Imaging the adult brain

rournal of Neurology, Neurosurgery, and Psychiatry 1995;58:7-21

NEUROLOGICAL INVESTIGATIONS

Imaging the adult brain

Ivan Moseley

The techniques which form the neuroradiolo-gist's armamentarium depend in one form oranother on transformation of energy waves;with the exception of ultrasonography, thoseof the electromagnetic spectrum. None ofthese energy forms is free from potentiallynoxious biological effects, and this is particu-larly true of the socalled ionising radiations, x,,B, and (5 waves. It is probably true that, as inthe case of new drugs vis-a-vis the formulary,attempts to introduce x ray based techniquesas innovative procedures today might well fail.It is therefore realistic to consider diagnosticimaging procedures primarily in relation tothe energy source employed: x rays and otherionising radiations, radiofrequency, andsound waves.

Plain radiographsThe neuroradiologist's use of x ray techniqueshas changed considerably over the past 20years. "Radiographic" (x ray) procedures reg-ister the differential absorption of a beam of xrays by the various tissues of the body. Theydiffer greatly in the manner in which that reg-istration is effected so as to preserve anatomicalrelations and maximise clinically useful data,while attempting to limit radiation exposure.In the classical radiograph or "plain film" thepatient is simply placed between a source of xrays and a film sensitive to radiation. Becausemost of the unabsorbed radiation is emitted ina straight line, anatomical relations are main-tained and spatial distortion is minimal unlessthe film is not orthogonal to the beam; mostmodem radiographic tables ensure that obliq-uity is avoided.

Lyshohm DepartmentofRadiology, NationalHospital for Neurologyand Neurosurgery,Queen Square, LondonWC1N, UK andDepartment ofRadiology, MoorfieldsEye Hospital, CityRoad, London EC1,UKI Moseley

Contrast media in plain radiographyContrast media can be employed to increaseradiographic contrast between anatomicalstructures, demonstrating some not normallyvisible. They are usually introduced into pre-existing body spaces, such as the alimentarytract or subarachnoid space (for myelographyor cistemography) or into the vascular system,where their flow within (in arteriography or

phlebography) or excretion from it (in urogra-

phy) can be followed with a suitably timedseries of x ray exposures, otherwise sinilar tothose when no contrast medium is used.

Computed tomographyIn x ray tomography of all kinds, some motionoccurs between the three elements, x ray

tube-patient-film. In x ray computed tomog-raphy (CT) the beam emerging from thepatient is registered not on a film but by anarray of radiation detectors, as the x ray tubepasses around the part being examined (inthis case, the head). A computer is used totransform the intensity of the emergent beamrelative to the incident beam into measures ofx ray attenuation, to identify in two dimen-sions the site of the tissue causing that attenu-ation, and to construct a two dimensionalmap of attenuation analogous to an anatomi-cal section. The planes of section of the headgiven by CT are effectively limited by the con-struction of the gantry bearing the x ray tubeand detectors to axial and coronal planes.Reformatted images in other planes can beconstructed by analysing the data from aseries of contiguous sections. One of the latestdevelopments in CT is "spiral", or more cor-rectly "helical", scanning, in which the tuberotates continuously around the head whilethe patient is moved through the gantry. Therapidly acquired, continuous sequence ofattenuation measurements can be "cut up"into segments, corresponding to two dimen-sional sections the thickness of which is deter-mined by the operator. The merits of thismethod are speed, which makes it more usefulfor examination of children and uncooper-ative adults, and greater spatial resolution inimages reconstructed in planes other than thatof acquisition.The relative "attenuation coefficients" of

certain structures can also be modified bycontrast media, usually containing iodine,introduced into the vascular system or sub-arachnoid space. Other types of contrastmedium, such as inhaled stable xenon gas, arerarely used in clinical practice.

Digital radiographyThe third main method of registering x rayattenuation data is by using a high speed cam-era to scan a screen coated with a phosphorthat emits light when radiation falls upon it. Ifthe camera transforms the light energy into anelectronic signal, this can be digitised and acomputer employed to compare the results ofone scan of the phosphor screen with that ofthose which precede or follow it. A digital toanalogue converter can then store and displayonly the difference between two sets of attenu-ation data-before and after arrival of contrastmedium, for example-in a format similar tothat of the original phosphor screen picture,but with unwanted detail removed. This is the

7

group.bmj.com on April 8, 2018 - Published by http://jnnp.bmj.com/Downloaded from

http://jnnp.bmj.com/

http://group.bmj.com

Moseley

basis of digital subtraction techniques, usedparticularly in angiography to produce imagesin which only the blood vessels are displayed.The major advantages of this technique arethat the detection system is much more sensi-tive than plain film angiography to minorchanges in intensity of the emergent beam,and that, because the data are stored on acomputer, they are amenable to interrogationand postprocessing.

Radionuclide studiesThe other techniques using ionising radiationsgenerally consist of the introduction into thebody of a radionuclide-namely, a source ofradiation, usually,B or 5 rays-and registrationof the radiation it emits with external detec-tors. Originally, these detectors gave no spa-tial information, but with the development ofscanning methods subsequently applied toCT, and the gamma camera, a scintillationdetector similar to the phosphor screen devicereferred to earlier, anatomical localisationbecame possible, although spatial resolution isinferior to that of most of the techniquesdescribed. Modern techniques such as singlephoton emission computed tomography(SPECT) use a combination of scanning andscintillation detection. Positron emissiontomography (PET), to be discussed in a laterarticle in this series, uses the same fundamen-tal principles, but with some technical modifi-cations.The great potential merit of the radio-

nuclides used in these nuclear medicine studiesis that they can be attached to chemical com-pounds which have a greater or lesser degreeof tissue specificity, and provide metabolic ordynamic information.

Nuclear magnetic resonance methodsThese exploit the behaviour of nuclei withunpaired electrons (paramagnetic nuclei)within the body in a strong magnetic field.The basic principle is very simple: some ofthese nuclei align themselves with that exter-nal field and can be displaced by a pulse ofradiofrequency energy. The extra energyimparted to the system is dissipated, also asradiofrequency waves, once the radiofre-quency pulse ceases and the nuclei return totheir previous position. The speed at whichthe energy is given up, and thereby thestrength of the radiofrequency signal at anygiven moment, depend on the biophysicalcharacteristics of the tissue in which the nuclei(typically hydrogen protons in clinical imag-ing) lie; the decay of energy release is expo-nential, the curve being determined mainly bytwo "relaxation time constants", Ti and T2.For many biological tissues Ti is of the orderof 1000 ms and T2 75ims. The intensity ofthe energy released from a given tissuedepends on the number of susceptible nuclei:water has numerous mobile protons, whereasdense bone has very few. As the frequency ofthe radio signal emitted also depends on thestrength of the magnetic field, a "gradient"-

that is, a variation in strength of the fieldacross the region being examined-enableslocalisation of the nuclei in one plane ofspace. Thus by examining the intensity andthe rate of decay of a radiofrequency signal ofknown frequency, a number of the characteris-tics of the proton emitting it can be assessed.Simple though the principle may be, theelectronic switching and computation re-quired for three dimensional localisation andtissue characterisation are exceedingly complex.The characterisation can be refined by

modifying the way in which the nuclei areexcited, applying a "sequence" of radiofre-quency pulses instead of simply one, and bysampling the emitted signal after varyingdelays. Consideration of how this is achievedis beyond the scope of this article; suffice it tosay that most clinical MR images are maps ofwater distribution, but that they can be modi-fied to show mainly differences in T1 (T1weighted images) or T2 (T2 weighted images)of the tissues being studied. As a simpleguide, the TI weighted images give moremorphological data, whereas the T2 weightedimages provide more information about theconstitution of the tissues. A full MRI studyusually combines at least these types of image.Some of the unsatisfactory diagnostic applica-tions noted in the early days of MRI resultedfrom the use of a limited number ofsequences, or sequences which did not pro-vide sufficient contrast between normal andpathological tissues. Intravascular contrastmedia, mainly containing the paramagneticrare earth gadolinium, can be used, as with xray CT.

Flow imagingBecause electromagnetic waves travel at thespeed of light, techniques which employ themare essentially instantaneous. Because in MRIthere is a finite delay between the applicationof the excitatory radiofrequency pulse andreception of the emitted magnetic resonancesignal, however, protons which are movingthrough the tissue being imaged have differentproperties from those which are macroscopi-cally stationary. This enables the relative grossmovement in blood vessels and the CSF to becharacterised and imaged with speciallydesigned sequences.

Magnetic resonance angiography (MRA)was first presaged in the imaging literaturealmost a decade ago,' but has been availablein clinical practice and of a quality whichmakes it a useful investigation for about threeyears; image quality is still improving. A moredetailed account of the technique and itsapplications will be found in a later article inthis series. In combination with MRI, clinicalapplication of MRA will bring about a majorchange in the investigation of at least thosepatients in whom information about the cervi-cal or intracranial vessels is useful, but notcentral to their management-for example,patients with parasellar tumours. In the longterm, it may completely replace diagnosticangiography.

8





The present and future role of CSF imag-ing is less unambiguous. A number ofresearch papers have been published concern-ing motion and pulsatility of the intracranialfluid in such puzzling conditions as normalpressure hydrocephalus2 and benign intracra-nial hypertension,3 but without as yet castingany clear light on pathophysiology. Onereviewer4 identified other applications forphase contrast MRI of CSF movement:assessment of the functional significance ofdrainage via extra-arachnoid structures or thecentral canal of the spinal cord; investigatingwhy some arachnoid cysts enlarge; and char-acterisation of obstructive hydrocephalus"without having to introduce contrast agentsinto the subarachnoid space"-a practicewhich virtually disappeared with the introduc-tion of CT! In another study, a fast field echotechnique, applied to determination of thepatency of ventricular shunts, was shown tobe fallible, despite which the authors claimedthat the technique "may be useful".5 Thefuture clinical utility and, indeed, adoption ofthese applications are at present questionable.

Diffusion weighted MRIMagnetic resonance imaging techniques canbe used to show other physical characteristicsof tissues. One method which has been devel-oped over the past five years is diffusionweighted imaging.6 Depending on the gradi-ents used, the contrast in the images candepend on magnetic susceptibility, perfusion,and diffusion. To date the genuine clinicalapplications of this technique have also notbeen clarified, although research has been car-ried out on brain myelination, stroke, andsome tumours.7

Magnetic resonance spectroscopyThe phenomena of nuclear magnetic reso-nance were originally employed for in vitrochemical analysis. In vivo analysis of body tis-sues can be carried out with magnetic reso-nance spectroscopy (MRS), which relies onthe same variation in the resonant frequencyof given nuclei due to their physicochemicalenvironment to create spectra which reflectthe concentration of chemical compoundscontaining them. In the brain, spectra fromboth hydrogen and phosphorus containingcompounds have been investigated; muchclinical research on proton spectroscopy iscurrently oriented towards relative concentra-tions of choline, creatine, n-acetyl aspartate,and lactate in normal and pathological braintissues. Magnetic resonance spectroscopy isnot strictly an imaging technique, although itemploys essentially the same imager; it iscombined with MRI to localise the volume ofbrain from which the spectrum is obtained.Until recently, a major problem with clinicalapplications of MRS has been the low signalto noise ratio, which has entailed large vol-umes of interest and long acquisition times.Smaller volume spectra have, however,shown, for example, that choline tends to be

increased in solid brain tumours, whereas ala-nine concentration is high in meningiomas;however, even though such information mightbe useful in cases in which the imaging fea-tures were otherwise inconclusive, spectro-scopic results are by no means diagnostic.8Wider application to cerebral metabolic dis-ease is awaited.9

Unlike MRA, for which the major futurecontribution to clinical neurology and neuro-surgery seems unequivocal, these other tech-niques are perhaps still in search of clinicalapplications.

SonographyAn ultrasound probe contains a "transducer"which emits 1-20 MHz sound waves and,when they bounce back off the tissues, con-verts the reflected energy into an electrical sig-nal, which is analysed, usually to form animage. Bone, however, is not a good mediumfor transmission, so that current applicationsof sonography to the brain are confined tosmall children, in whom the fontanelles act as"acoustic windows", patients with skulldefects (including intraoperatively'0), andexamination of the eye and the major arteriesin the neck. The major arteries will be consid-ered in another article in this series.

The use and utility ofimaging proceduresAn article in this journal in May 199411 con-sidered the possibility and utility of "periodi-cally updated reviews of all randomisedcontrolled trials relevant to neurology andneurosurgery". Virtually none of the trialsreferred to was primarily concerned withimaging the brain, and the unfortunate truthis that many clinicians appropriately guidedby controlled trials of treatment approachimaging in what seems to their radiologicalcolleagues to be a decidedly aleatoric way,guided by personal prejudices and what theirchiefs used to do rather than by the latestaudit on best practice. That radiologists arealso prone to promote procedures that theyenjoy ("cath time"' 2) and that far too manyradiological publications are uncritical in theirpraise for recently developed procedures"3 isunhelpful.My intention is to review, very broadly, cat-

egories of disease, and to attempt to indicatewhich imaging tests are the most likely to pro-vide the information most appropriate formanagement of the patient; in general, I shallnot dwell on how imaging research has clari-fied disease processes. I shall not look atpatients likely to have cerebrovascular disease,as this is the subject of another review in thisseries. Neither shall I attempt to review all theimaging literature of recent years. Regrettably,in the countries from which the bulk of thisliterature arises there are strong motives forboth promoting the use of medical facilitiesand for publishing one's work, which mayexplain the remarkable, and sometimeslamentable absence of outcome studies. Tomy knowledge, there is, for example, very

9




10

little evidence that the outcome for patientswith even the "surgical" neurologicaldiseases (intracranial tumours, abscesses,haematomas, etc) has been appreciablyimproved by the most recent advances inimaging, except where these are indissolublywedded to a surgical technique, as in imageguided stereotaxic surgery. Indeed, such stud-ies as have been performed have producedvery disheartening results, showing no longlasting improvement in the quality of ourpatients' lives.'4Outcome in this sense is, of course, not the

only measure. The great advantage of manymodem imaging methods is that the patientendures far less unpleasant experiences, orruns less risk, to achieve the outcome, albeitidentical. In this sense, CT represented a con-siderable advance over its predecessors, notnecessarily improved on by progression toMRI, with the notable exception of MRA.Even short term outcome may not be affectedmuch. Thus in MRI studies of the brain inpatients with AIDS, there is no clear benefitfor the large majority if contrast medium isgiven intravenously,'5 but some radiologistswould defend its routine use. "Some of us arewilling to pursue diagnostic accuracy withimaging studies to greater lengths than others,and each of us is likely to have some diseases towhich we are willing to devote more time thanothers. This phenomenon [the "compulsive-ness factor"], rooted in human nature, willsurvive even the current rage for outcomeanalysis".'6 One may counter this by observ-ing that radiologists, like too many other doc-tors, become so accustomed to sticking sharpobjects into their patients that they cease toregard it as an assault. Were the doctor inject-ing a patient to explain that his main motivewas diagnostic compulsiveness rather thanany expectation of altering management, hemight well expose himself to a charge of bat-tery! 17With the advent of CT about 20 years ago,

followed less than a decade later by MRI, therange of methods by which neuroradiologistsmake diagnoses and carry out follow up studieshas paradoxically decreased. Both these crosssectional imaging techniques have led to avery significant fall in the number of otherinvestigations performed, both plain radi-ographic and invasive, something which didnot occur with the introduction of radionu-clide examinations into neuroradiology in theyears preceding the introduction of CT.'8 Weare still going through the phase, familiar toradiologists, in which patients are submittedto multiple tests which yield essentially thesame information, as occurred with CT andsome of the older invasive investigations.-'Thus many patients currently undergo bothCT and MRI when either (or even neither)would suffice, but one may hope that this is atemporary phenomenon.The choices of imaging procedure(s) the

clinician has to make when confronted with agiven clinical problem were summarised bythe Royal College of Radiologists in a bookletMaking the best use of a department of clinical

radiology. Guidelines for doctors,'0 the secondedition of which appeared at the end of 1993.Its recommendation was that every timesomeone thinks of requesting an examinationthey ask themselves "Do I need it?"; "Do Ineed it now?"; "Has it been done already?";"Have I explained the problem?"-that is, dothe radiographer and radiologist understandwhat I need to know and why?; and "Is thisthe best study?". The answers to several ofthese questions are different in Britain in 1995from what they might be in an ideal world.Most neurologists and neurosurgeonsacknowledge that they are perforce temperedby considerations of cost and availability, aswell as inconvenience and risk to the patient(from radiation and other hazards), but theyshould also be informed by the answers to theimplied questions "Is this test really likely tohave any impact on my management of thispatient? If not, will it nevertheless give datawhich will genuinely help my understandingof this disease and management of otherpatients with similar problems?", and, onemight add "If so, does the patient who willsuffer the inconvenience and risk of this test-a threat to life of 1 in 40 000 in contrastenhanced CT studies, for example21-knowthat?"

Head injuriesThe controversy over the optimal imagingapproach to head injuries continues to bumand, regrettably, the Royal College ofRadiologists' booklet only serves to fan theflames. A major problem with rational man-agement in Britain is that, more than in anyother group of potentially neurologicalpatients, the primary care of those with acutehead injuries is largely entrusted to inexpertjunior doctors who lack both the necessarymaturity of judgement and a clear knowledgeof the facts. The confusion over the relativeroles of plain radiography and other tests,which in this context means essentially CT,pervades the Royal College's recommenda-tions. Complex non-invasive techniquesshould not be employed to identify fracturesof the cranial vault in patients with significantintracranial injuries. The presence of a linearfracture itself is irrelevant to the patient'smanagement. It is often said that lawyersplace great emphasis on injuries having beensevere enough to cause a fracture; if this is so,it should be the clinician's task to re-educatethem. If plain radiographs have any part toplay it is in the assessment of complex frac-tures and, more debatably,22 in raising suspi-cion of an underlying injury which mayrequire treatment. Thus the advice that apatient "disorientated or worse . . . drunk/difficult to assess [or having] stable focal neu-rological signs" should have a skull radiographand be admitted for observation seems likerecourse to both belt and braces; that consul-tation with the neurosurgeons be immediate ifthe skull film shows a fracture but only after12 hours if the patient fails to improve is ques-tionable at best. Even more debatable is the

Moseley





suggestion that for a patient developing focalsigns or whose conscious level is deteriorating,even for one whose pupil is dilating, pulseslowing, and blood pressure rising CT or askull film is appropriate-as though the twowere equivalent! A nationwide survey in theUnited States, reported in 1991, indicatedthat skull radiographs were obtained fortrauma rarely or never in almost half of thehospitals surveyed, and performed less oftenin larger hospitals, with free access to CT, andat the behest of neurologists or neuro-surgeons.23As indicated earlier, CT would at present

seem to be the primary investigation fordetecting the intracranial complications ofhead injuries for several reasons: cheapnessand general availability, lack of the logisticproblems posed by carrying out MRI in emer-gency situations with a patient on a ventilator,etc, the ready detection of acute haemorrhage,and the fact that the management and prog-nostic relevance of the changes shown isunderstood. Magnetic resonance imaging ismore sensitive to contusion or shearinginjuries, but currently the therapeutic conse-quences of their demonstration are less welldefined. This point applies not only in headinjury. Because more intracranial lesions canbe shown by MRI than by other techniques inpatients with suspected disseminated malig-nancy, etc, staging of these patients, based onthe sensitivities of older techniques, may haveto be revised.A late consequence of head injury which

can lead to longstanding problems is the CSFfistula, usually giving rise to rhinorrhoea. Anumber of investigations, some ofwhich (con-ventional tomography of the skull base,radionuclide cisternography) deliver a heftyradiation dose for little or no return, havebeen proposed in the past. For patients whoare surgical candidates because of persistentproblems, however, CT with intrathecal watersoluble contrast medium, performed in theactive phase of leakage,24 is often the onlyradiological investigation required.

Brain deathInnumerable techniques have been proposedfor use in the diagnosis or recognition of braindeath, often with little or no regard to the rel-evant national or international recommenda-tions. Indeed, many of the suggestions seemto be predicated on the presumed inability ofneurologists to make the diagnosis reliably.The Guidelines for the determination of deathdrawn up in the United States by thePresident's Commission almost 15 years agoindicated that the diagnosis is indeed clinical,and that paraclinical tests, including imaging,might be used firstly to determine the cause ofirreversible coma and secondly to shorten theperiod of observation required for definitivediagnosis.25 There is little or no evidence thatthese imaging procedures are in practice usedto supplement clinical examination in earlierdetermination of brain death, their only puta-tive merit. One should not forget that, as

Griner and Glaser26 observed, diagnostic testsmay be useful in four ways; two of these-screening and diagnosis-may be of directutility to the patient. The third-assessmentof treatment-is just as likely to be beneficial,if at all, only to the doctor, whereas thefourth-income generation-is uniquely so.The presumably self interested claim in theRoyal College of Radiologists' Guidelinesthat in brain death radionuclide investigationshave a "first line" role for "confirmation ofneuronal loss" (!) seems as inappropriate asfactually incorrect.

Acute lesions of cranial nervesThe plain film has no part to play in the initialassessment of patients with visual loss, exceptwhen this is thought to be due to an intraocu-lar foreign body27; neither are there any indi-cations for optic foramen views.28 Shouldexamination suggest an ocular cause otherthan intrinsic disease of the lens or retina,sonography of the globe (a specific techniquebest reserved for experts) may be employed.Magnetic resonance imaging may produceimages which are more readily interpreted bythe non-specialist, but it is doubtful that it hasmuch to contribute when expert sonographyis available.When the signs suggest involvement of the

intraorbital optic nerve, sonography is con-traindicated; CT is probably as useful in mostcases as MRI, and may show some significantabnormalities such as calcified, buried drusenof the optic disc or small, or even large,29plaques of calcification in a meningioma ofthe sheath of the optic nerve, to which MRI isinsensitive; MRI may, however, be moreinformative when CT shows a mass lesionwith a questionable anatomical relation to theoptic nerve (and therefore its amenability toresection), or when it is important to deter-mine whether a tumour arising from the nerveitself or its sheath extends intracranially.

Lesions within the substance of the anterioroptic pathways that are not space occupying,essentially inflammatory, or degenerative, canbe shown only by MRI. Thus pronouncedswelling and contrast enhancement of thenerve-sheath complex may be shown insarcoidosis. With high resolution coronal sec-tions, it may be clear that the inflammatoryprocess involves the meninges rather than thenerve itself, and the same contrast enhancedMRI study may show multifocal intracranialmeningeal involvement as a bonus. The gen-uine utility of demonstrating plaques ofdemyelination, signal change in Leber's opticneuropathy, etc, is debatable, unless the pres-ence and extent of a lesion within the opticnerve could influence the decision whether totreat the patient with high dose steroids. InBritain, about two thirds of patients present-ing with classical retrobulbar neuritis will goon to develop multiple sclerosis, and there isevidence that the large majority of those atrisk can be identified by the presence oflesions in the brain at the time of initial pre-sentation with visual disturbance30; if this

1 1




Moseley

knowledge is considered important, a strongerargument can be made for requesting MRI ofthe brain rather than the optic nerves. Thereare nevertheless some conditions, such asradiation neuropathy, in which imaging maybe used not only to exclude a compressivelesion (a recurrent skull base tumour, forexample) but also to establish a positive diag-nosis, by showing pathological contrastenhancement within the nerve(s). Moreover,as damage to the blood-brain barrier isevanescent, follow up studies showing itsrestitution can resolve any diagnosticdoubts."Mass lesions confined to the optic canal or

affecting solely the adjacent bone are veryrare. Magnetic resonance imaging would bethe best technique for showing such masses,although CT often shows the bone to bettereffect. Evidence that surgical or other treat-ment is effective for visual loss due to traumato the optic canal region is largely unconvinc-ing, and a recent authoritative review suggeststhat it should be the subject of a serious con-trolled trial.'2 Recommendations about radio-logical attempts to show fractures, etc, musttherefore be viewed critically. It may, how-ever, be important for the faciomaxillary sur-geon to be forewarned of the presence of afracture through the canal when a visuallyintact patient is to undergo treatment forother facial injuries; this would be best docu-mented by CT. As indicated, MRI is theexamination of choice for demonstratingintracranial extension of a primary tumour ofthe optic pathways, manifest as signal changeor expansion of the optic chiasm. With theaddition of intravenous contrast medium, it isalso the best way of demonstrating spread tothe intracranial meninges of an optic nervesheath meningioma."

If the pattern of field loss localises thevisual pathway disturbance to the intracranialoptic nerve (by virtue of a contralateral junc-tional scotoma) or to the optic chiasm, MRI isin general superior to CT not only for detec-tion, but even more so for effective preopera-tive characterisation. Images can be obtainedin multiple planes, and MRA, which addsonly a few minutes to the duration of theexamination, can adequately show theparasellar arteries. An important studyshowed that aneurysms of sufficient size tocompress the proximal intracranial optic path-ways were unlikely not to be visible on MRI orCT.'4 The investigation of lesions affectingthe optic chiasm is essentially similar to thatof pituitary tumours, and is considered below.Patients with retrochiasmal visual loss can bedivided clinically into two groups: those inwhom the visual deficit is isolated and thosewith other evidence of a cerebral lesion. For acerebral lesion the role of imaging is to showthe presence or absence of a mass lesion,plaques of demyelination, etc, and these areconsidered elsewhere. An isolated homony-mous hemianopia is usually due to an infarctinvolving the optic radiations or visual cortex,and CT or MRI are adequate for its demon-stration. Angiography is not required unless

some form of intervention would be contem-plated.

PapilloedemaWhen patients presenting with longstandingheadaches or visual disturbances are found tohave bilateral disc swelling, the assumption isthat they have papilloedema due to raisedintracranial pressure until proved otherwise.From a strictly pragmatic viewpoint, a simpleCT study without intravenous contrastmedium only is required to establish firstlythat there is no intracranial space occupyinglesion, and secondly that lumbar puncture,indispensable for the diagnosis of benignintracranial hypertension, may be performedsafely. It should be unnecessary to add thatthe cerebral ventricles are of normal size in thiscondition.'5 Demonstration of an empty sellaturcica and dilated optic nerve sheaths is abonus. MRI will show the sheaths in moredetail, which may conceivably be useful ifvisual impairment is such that operativedecompression of the sheaths is considered. Itcan also give additional information on thepatency or otherwise of the dural venoussinuses, but the true value of that informationin someone with chronic disease is highlyquestionable; the employment of invasiveangiography can scarcely be justified.

Acute ocular motor nerve palsiesThe investigation of acute cranial nervepalsies presumed to be due to structural vas-cular lesions, specifically aneurysms, will bediscussed in a later article in this series. Manyof the larger non-vascular lesions, particularlythose amenable to treatment, will be visibleon CT but, in as much as isolated acute cra-nial nerve lesions may be a manifestation ofdemyelination, MRI is clearly preferable.'6Even then, no causative lesion will be shownin most patients presenting in this way; mod-ern imaging has largely failed so far to eluci-date such lesions as the "microvascular" thirdnerve palsy.

Extending the CT or MRI study to theorbit can be of value in identifying lesions ofthe superior oblique or lateral rectus musclesthat mimic fourth or sixth nerve palsies.

Acute facial palsyThere have been reports of positive findingson MRI, particularly with gadolinium con-trast enhancement, in patients with isolatedacute facial weakness, and their relation toprognosis.'7 In typical cases of Bell's palsy,however, the genuine therapeutic utility ofthese expensive investigations may be on a parwith that of showing uncomplicated rib frac-tures with a chest radiograph, and it seemsentirely reasonable to reserve imaging foratypical cases. One must be aware that, aswith some inflammatory lesions-cerebralabscess being a prime example-contrastenhancement may persist once clinical resolu-tion has occurred. One must remember that

12





imaging can reveal only a relatively smallrange of abnormalities (swelling, shrinkage,change in density or intensity, etc); it cannotinterpret images without the clinical picturefirmly in mind, as a report of findings on con-trast enhanced MRI in the Ramsay Hunt syn-drome-in a patient with symptoms for twodays-mimicking an intracanalicular vestibu-lar schwannoma38 emphasises.

Acute lower cranial nerve palsies are com-monly either vascular or demyelinating andare therefore treated elsewhere.

Chronic lesions ofthe lower nervesThe number of patients presenting with dis-turbances of hearing or balance is very large,and were all to be intensively investigated, theimaging services would be in danger of beingswamped. The number of patients referred forimaging can, however, be controlled so thatthe services are used much more appropriatelyif neuro-otological consultations precedeimaging. This is particularly true in patientswho have vertigo without hearing loss, inwhom a positive diagnosis can often be madewithout recourse to imaging, and a structurallesion is rarely found. When the history,examination, and otological tests indicate thepossibility of a vestibular schwannoma, thereis now little justification for carrying out anytest other than MRI. Plain films and conven-tional tomography have no part to play in the1990s (and it was very difficult to rationalisetheir use in the preceding decade).39 It is nowevident that, when MRI is not available, CT isa poor second best, although it can demon-strate relatively large tumours. The promotionof gadolinium enhanced images for MRIdemonstration of smaller tumours40 wasalways questionable, as most could be shownwith adequate clarity by a combination of Tland T2 weighted images, except perhaps onthe least satisfactory machines. The recom-mendations for the use of contrast medium bythe 1991 NIH Consensus DevelopmentPanel, not published until 1994,4' have beenovertaken by technical developments: it isnow widely accepted that high resolution T2weighted images of the region of the internalauditory meatus in a single plane are adequatefor exclusion of even small tumours42; whenthe results are equivocal, addition of a secondplane is quicker, kinder, and much cheaperthan subjecting the patient to an intravenousinjection. Optimal techniques may, of course,vary depending on local factors such asmachine strength, availability of specific coils,high resolution programmes, etc; discussionwith the radiologist may be necessary toachieve the optimal results in any given unit.

"Vascular compression syndromes" in theposterior fossa continue to stir up debate. Ifone believes that compression of the fifth orseventh cranial nerves by relatively small ves-sels can produce facial pain or twitchingrespectively, and that surgery is the answer,high resolution MRI similar to that used forinvestigation of the eighth nerve is the optimalimaging technique. It is the only one which

shows clearly the cranial nerves and the adja-cent vessels: it may be supplemented by MRAshowing the whole course of the "offending"artery,43 but if the surgical approach does notdepend on the detailed vascular anatomy, thismay be redundant.The diagnostic yield of CT in patients with

palsies of the lower cranial nerves (IX-XII) issufficiently low compared with that of MRIthat the older technique should be foregonewhenever feasible. Intrinsic lesions-neoplas-tic, vascular, or inflammatory-of the lowerbrain stem, often quite small, can often beidentified only with MRI.44 Most extra-axiallesions are also clearly visible, with the excep-tion of malignant or inflammatory basalmeningitis, in which even quite extensivelesions may be evident only on images afterintravenous gadolinium.When extra-axial masses are identified in

the lower part of the posterior cranial fossa,MRI will confirm the nature of aneurysms orprovide preoperative vascular mapping in thecase of soft tissue lesions. When MRI revealsa lesion in the region of the foramen magnum,however, particularly one showing inhomoge-neous contrast enhancement that suggests thepresence of large vessels within it, an intra-arterial study with the aim of proceeding totherapeutic preoperative embolisation isstrongly indicated.

Magnetic resonance imaging is also theprime investigation for patients in whom brainstem or cerebellar disorders are thought to bedue to anomalies at the foramen magnum.Numerous studies have suggested that thecerebellar tonsils may, in normal subjects,extend a few millimetres below the foramenmagnum, whereas older studies, usually myel-ographic, suggested that this was not the case.More recent work suggests that the olderdata, although obtained with a techniqueliable to alter the craniospinal hydrodynamics,were indeed correct, and that it is the normalbiventral lobules extending through the cen-tral part of the foramen, which lie higher thanits lateral borders, which give the impressionof cerebellar ectopia.45 In practice this meansthat it is not abnormal for a midline sagittalimage to show the inferior part of the cerebel-lum extending slightly below the plane joiningthe anterior and posterior lips of the foramenmagnum. More important pathological obser-vations are that the neuraxis seems to pack theforamen and that the anterior surface of themedulla seems compressed against the odon-toid peg. Needless to say, the appearances of"cerebellar ectopia" in a patient withoutappropriate symptoms or signs are usuallyirrelevant. Imaging for suspected foramenmagnum lesions should include the uppercervical spine, as an asymptomatic associatedsyrinx may be detected. Plain films or CTmay be required before surgery for foramenmagnum abnormalities, because of theirsuperior demonstration of the bone.

Acute cerebral and cerebellar lesionsIt is only a few years since editorials published

13




14Moseley

in a number of neuroscience journals empha-sised that, despite the inroads of MRI, thediagnosis of multiple sclerosis was essentiallyclinical, backed up by neurophysiological andCSF testing.46 The same can be said of manydiseases, especially those in which initial clini-cal presentation is protean and subsequentcourse equally variable. In pragmatic terms,however, one may question this judgement. Inthe appropriate clinical context, particularly inyoung people, the demonstration of multiple,characteristic lesions, usually in the brain, butoccasionally solely in the spinal cord, gives asufficiently strong clue to the diagnosis toobviate further investigation. For example,the predictive value of a positive MRI study ofthe brain in patients presenting with retrobul-bar neuritis or myelitis as an initial neurologicaldisturbance is so strong that extensive addi-tional investigations would seem meddlesomeunless there are discordant features. Whetherthe patient with optic neuritis should undergoMRI of the brain is an ethical question; inpurely pragmatic terms, the efforts of the pop-ular press in drawing attention to this condi-tion as the harbinger of multiple sclerosis aresuch that many young women consider them-selves inadequately investigated without MRI.

It is certainly no longer effectively the casethat multiple sclerosis is a diagnosis of exclu-sion. For both clinical and technical reasons,however, routine MRI is more likely to showlesions in the head than in the spinal cord;indeed, the detection of lesions in the spinalcord is one of the most stringent tests of thequality of the imager. Estimates of the propor-tion of patients in whom MRI will show spinallesions when images of the brain appear nor-mal vary, but in the most reliable series thepercentage is small. This being the case, astrong argument can be made for imaging thehead first and foregoing images of the spine inyoung patients with relapsing symptoms,when cranial MRI clearly shows typical, dis-seminated disease. Even when a spinal cordepisode is the first neurological illness,demonstration of brain lesions increases therelative risk ofprogression to multiple sclerosisin less than two years by 36 times, or by about15 times if patients with complete transversemyelitis are excluded.47 The proportion of thisgroup, when attacks involve different spinallevels, in which an additional structural lesionwill be found is minuscule. As an analogy,when a patient with multiple enlarged lymphnodes is shown by biopsy to have classic sar-coidosis, one does not proceed to sample allthe other nodes in case one might showHodgkin's disease. In older patients with pro-gressive symptoms, which might be due tosuperimposed spondylosis, for example, thesituation is obviously different.

It is axiomatic that on MRI the lesions ofmultiple sclerosis are not pathognomonic.Some of the diseases that may simulate multi-ple sclerosis, including sarcoidosis and tropi-cal spastic paraparesis,48 may produce verysimilar radiological appearances. Clinical fea-tures, together with CSF analysis, are there-fore paramount in a proportion of cases.

Fortunately, these conditions are rare, as arethe often suggested but rarely documentedcerebral vasculitides, although these tend toproduce more peripheral and basal ganglionlesions,49 as does Behcet's disease, which alsoshows a predilection for the brain stem.i0Small vessel ischaemia, however, is ubiqui-tous, and in the individual patient, where sta-tistical analyses are unhelpful, it may bedifficult to make a firm diagnosis on imaginggrounds. Lesions which are predominantlyperiventricular, involving the corpus callo-sum, and with transverse diameter greaterthan their anteroposterior extent, favour mul-tiple sclerosis, as does finding multiple, appar-ently asymptomatic supratentorial andinfratentorial abnormalities in a youngpatient. None of these characteristics isabsolute; I am not convinced that statisticalanalysis of lesion numbers, etc51 helps anymore than an experienced radiologist'simpression in any given case. Once the diag-nosis of multiple sclerosis is established or

strongly suggested, repeated MRI studies todemonstrate the appearance of new lesionsare rarely indicated, particularly in establish-ing a differential diagnosis between, for exam-ple, small vessel ischaemic disease andmultiple sclerosis in a middle aged person.Occasionally, the finding of new lesions mayfacilitate the differentiation of acute dissemi-nated encephalomyelitis, a monophasic ill-ness, and multiple sclerosis; whether this isworthwhile clinically is, however, debatable.In established multiple sclerosis, there is littlepoint in repeat MRI studies charting thecourse of the individual patient's disease,unless there are strong reasons for thinkingthat a new neurological episode is due to someother condition.

Before the advent of MRI, some workersproposed double dose, delayed contrastenhanced CT for detection of cerebralplaques,52 but this was always of doubtfulvalue as most of the patients with positiveresults had diagnostic changes in the CSF.Accepting the caveat that the diagnosis ofmultiple sclerosis is essentially clinical, andthat the role of radiology before the advent ofMRI was largely exclusory, the use of thistechnique in patients who cannot for somereason undergo MRI is highly debatable. Ingeneral it is also true that when MRI studiesare positive in patients with suspected multi-ple sclerosis, recourse to gadolinium enhance-ment is unnecessary. Occasionally, however,when alternative diagnoses seem likely, theabsence of contrast enhancement of anylesion may strengthen the suspicion of multi-ple sclerosis, particularly when the patient isnot in an acute relapse, whereas enhancementof all or most of the lesions visible on T2weighted images may point to acute dissemi-nated encephalomyelitis-in which some ofthe lesions are often large,53-or an inflamma-tory condition such as sarcoidosis.

Intracranial infectionsThere is little doubt that CT revolutionised





the neurologists' and surgeons' approach tomany intracranial infections, including cere-bral abscesses, empyemas, granulomatous dis-ease, etc. It is also clear that for certain types ofinfection, notably encephalitis, MRI is signifi-cantly more sensitive, particularly in the earlystages of the disease.54 This is also true ofother conditions, such as meningitis andAIDS, but the added sensitivity in some ofthese may simply allow observation of diseaseprogression rather than conferring any benefiton the individual patient, much though theymay advance medical knowledge.Many subacute or chronic infections pro-

duce changes that can be identified on CT, orwith greater ease on MRI. Unfortunately, asin so many situations, the radiological findingsare non-specific; thus the diagnosis ofintracranial Whipple's disease, Lyme dis-ease,55 etc, is invariably based as much onclinical and epidemiological factors as onimaging. This also applies to opportunisticinfections in AIDS: although in many parts ofthe world the major organism is toxoplasma,cryptococcus may be relatively common insome areas,56 whereas in Spain, for example,tuberculosis is an important consideration.57Differentiation of toxoplasmosis from lym-phoma by means ofMRI is sufficiently unreli-able for antimicrobial treatment or biopsy tobe the recommended course58 rather thanreliance on the radiologist's diagnostic "com-pulsiveness".

Chronic and progressive syndromesCEREBRAL LESIONSHistorically, one of the fields to which mosteffort has been directed in imaging the head isthat of progressive focal or generalised neuro-logical deficit-that is, detection and charac-terisation of intracranial tumours. This is not,however, a single task, and for practical pur-poses, it can be subdivided as follows59:* detecting a mass lesion within the head* indicating whether it lies within or without

the brain* whether it requires biopsy for tissue char-

acterisation* indicating the best means of intervention* directing that intervention* monitoring the natural history of untreated

lesions* secondary effects of the lesion* the effects of treatment.

It is a truism that skull radiographs andradionuclide investigations are not indicated,even as non-invasive screening tests, for detec-tion of intracranial masses; both have unac-ceptably high false negative rates.6061Magnetic resonance imaging yields a signifi-cantly higher detection rate than CT, espe-cially in the anatomical sites to which CT isknown to be relatively blind-namely, theanteroinferior part of the middle cranial fossaand the posterior cranial fossa, particularly thebrain stem. Studies carried out when MRIwas first introduced, however, gave conflictingresults as regards the detectability of supra-tentorial neoplasms other than in the tempo-

ral lobes.62 Some of these were methodologi-cally questionable, and more recent scientifi-cally conducted comparisons are few, but aclinical impression is that, with modemequipment, the detection rate for tumours inor adjacent to the cerebral hemispheres withthe two techniques is not greatly different.Even clinicians' awareness of this does not,however, prevent many patients having bothtests.

Contrary to a widespread belief, numerousstudies have confirmed the radiologists'impression that there is only a very small (lessthan 1%) decrease in detection rate of pri-mary cerebral tumours if intravenous contrastmedium is not given; furthermore, this wastrue even when image quality was much infe-rior to that available today.6' It is recognisedthat "false negative" CT studies occur withfast growing malignant gliomas, particularly inpatients presenting with seizures; a secondexamination months later, by which time neu-rological signs have often developed, mayshow a large tumour genuinely invisible onthe first64; the same may also occur with MRI.Not surprisingly, contrast enhanced MRI hasbeen found to be more effective than CT fordetection of intracranial metastases. Thisapplies particularly to extra-axial deposits, towhich CT is notoriously insensitive,65 but isalso true when only parenchymal lesions arepresent66; there seems to be little or no benefitin delaying the postinjection imaging, amanoeuvre claimed to increase the detectionrate in CT.Once the radiologist has decided that an

intracranial mass is neoplastic, which is notdifficult in most cases, his long term aimwould clearly be to be able to provide a tissuediagnosis sufficiently reliable to obviatebiopsy. Until we reach that happy state, it canbe argued that the neuroradiologist shouldrather "ask himself whether or not a precisediagnosis is really useful",67 and throw in thetowel. Situations which engender negativeanswers to that question are almost certainlymore common than otherwise. Exceptionsinclude the fundamentally important, butoften rather low level decision, as to whether amass lies within or outside the brain, as in theformer case it is potentially resectable;whether multiple or less commonly singlemasses are likely to be part of disseminatedmetastatic disease, in which case surgery onthe head, even biopsy, may not be the pre-ferred option; or whether it might be a lym-phoma, in which case one should abstain fromsteroid treatment before biopsy.68 It seemed atone time that immunolocalisation of braintumours, using tissue specific antibodieslabelled with radionuclides69 or some com-pound detectable by magnetic resonance,would obviate radiological diagnostic guess-work, but that promise does not seem as yetto have been fulfilled.

Similar remarks apply to attempts to gradedegrees of malignancy by imaging. Con-siderable effort has been devoted to establish-ing correlations between malignancy andradiological criteria derived from angiography,

15




16

CT, MRI, and isotope studies. When theseinvolve complex formulae or multifactorialanalyses which barely achieve statistical signif-icance,70 their clinical utility is effectively nil.Simple Gestalt analyses of some of these tests,however, can contribute to prognosis whenbiopsy is either unhelpful or patently at vari-ance with the radiological data, indicating thatatypical tissue has been obtained. Theseshould be the minority of cases, as histologicalexamination is in general more reliable.Unfortunately, many recent enthusiasticreports on radiological differential diagnosisof intracerebral or intraventricular7l tumourshave the same fatal flaw; the differential diag-nostic game may be entertaining for the radi-ologist to play, but clinical features may beequally important and the images rarely sup-plant biopsy. It is perhaps puzzling that areport showing a close correlation betweengrades of malignity of intracranial tumoursand the methyl and methylene peaks ofperipheral blood on MRS72 seems to haveengendered little excitement among neurosur-geons-or perhaps not!Where extracerebral tumours are con-

cerned, the value of current imaging tech-niques in general is in similarly providingmorphological detail rather than refining diag-nosis. Magnetic resonance signal characteris-tics, for example, do not seem particularlyuseful for assessing preoperatively the likelyconsistency of meningiomas.73

Surgical use of imaging technology forlocalisation of lesions to be removed or biop-sied has progressed in less than 20 years fromthe complex but rather crude, using veryrough data given by air studies and angiogra-phy, to the highly sophisticated, using a vari-ety of techniques and combining theinformation into computer generated imageswhich often contain more information thanthe surgeon can use. The ways in which CTor MRI guided stereotaxy is currently per-formed are almost as various as the institu-tions involved. Suffice it to say that by thetime these words are published routineintracranial surgery carried out, perhaps byremote control, within a dedicated MRimager, with effectively real time data presen-tation, may no longer be science fiction.

Angiography remains part of the preopera-tive investigation of patients with intracranialtumours in some institutions, although thebenefits it confers are debatable. There wouldseem to be no justification for using an inva-sive technique with a known morbidity forlocalisation of masses, particularly when, as inthe large majority of cases, non-invasive pro-cedures are more reliable. Grading of intrinsictumours on the basis of pathological vessels,etc, has not been shown to be more sensitivethan grading by CT or MRI. When informa-tion about patency of dural sinuses, etc, isthought necessary before surgery, this can beacquired by intravenous digital subtractionangiography or MRA, and the same may besaid for the demonstration of large vesselsbefore stereotaxic biopsy. Virtually the onlyfirm indication for preoperative intra-arterial

angiography is when intravascular interven-tion is foreseen. Presurgical embolisation ofsome vascular neoplasms, especially thosearound the skull base such as glomus jugularetumours or juvenile angiofibromas, is todaygenerally considered the best practice, butsurgeons' universal enthusiasm for intra-arter-ial treatment of, for example, convexitymeningiomas, seems to have waned as doubthas been cast on the real benefits.74Unfortunately, meningiomas arising from theskull base, in which surgical access to thefeeding vessels is less straightforward andhaemostasis would be a real gain, are oftennot strikingly vascular on angiography.75

If cerebral tumours are treated by biopsyand radiotherapy (with or without other typesof treatment) rather than by excision, it isimportant to know the extent of brain involve-ment, as it is undesirable to expose normalbrain to therapeutic doses of radiation. Thereis evidence from surgical biopsies, post-mortem examinations, and correlation withPET studies of cerebral metabolism that CTunderestimates tumour spread,76 and thesame is almost certainly true of MRI. Therehave been a number of accounts of treatmentof cerebral77 and brain stem neoplasms byintra-arterial infusion of cytotoxic agents.Theoretically at least, delivering the noxiousagent close to the vascular bed of the tumourhas advantages, but it remains to be provedthat these outweigh the morbidity of the tech-nique.Computed tomography or MRI are excel-

lent methods of following up patients withcerebral tumours, treated or untreated. It is,however, reasonable to restrict the use ofthese or any other investigations to clinical sit-uations in which there is some realistic likeli-hood that they will modify management.When surgery seems inappropriate for apatient with a low grade intrinsic tumour,imaging follow up is required only when a sig-nificant change in clinical status occurs, whichmight tip the balance in favour of surgery, orin the unlikely event that some new operativetechnique renders intervention a betterchoice. Much the same applies in many situa-tions; imaging as occupational therapy for theoutpatient whose recent clinical status isunchanged some months or years after totalremoval of a meningioma, for example, isfruitless (Hodgson, Kingsley, and Moseley,unpublished data).

Differentiating recurrent tumour fromradionecrosis remains difficult with imaging;neither CT nor MRI is sufficiently reliable.There have been several reports of the use ofPET for this purpose, but its expense and lim-ited availability tend to exclude it from rou-tine use; SPECT techniques, which areconsiderably cheaper, may be informative,however, in at least some cases.78

HEADACHEA number of surveys have confirmed thevery low diagnostic yield of imaging inpatients with either non-specific headacheunaccompanied by neurological disturbances

Moseley





or recognisable forms of migraine, migrainousneuralgia, etc. This is one area in which sensi-ble clinical assessment is invaluable, given thefrequency of complaints of headache and thedesire of many patients to be investigated.The other side of the coin is that CT is now acheap and almost completely anodyne test,provided that contrast medium is avoided.When CT was considerably less widely avail-able and relatively more expensive than today,one audit of its utility suggested that an EEGmight be a good pre-CT screening test inpatients with chronic headache79; one couldprobably reverse this argument on cost-bene-fit grounds today. Plain films of the skull arean even cheaper way of confirming the diag-nosis in unusual conditions in which the skullitself is the source of pain, such as Paget's dis-ease. In a more common situation, films ofthe paranasal sinuses will usually show orexclude inflammatory or other changes severeenough to cause chronic headache, althoughthey may not suffice for further management.Conversely, plain films of the craniocervicaljunction or cervical spine hardly ever elucidateisolated occipital pain; moreover, they are sooften abnormal in elderly people80 that theycannot genuinely support a diagnosis of cer-vicogenic pain.

Investigation of types of headache known tobe associated with intracranial or craniocervicallesions-as in subarachnoid haemorrhage orcough headache-will depend on the pre-sumed nature of the underlying lesion: anintracranial aneurysm or cerebellar tonsillarectopia, for example. Because, save for suchcases, there is little to indicate that MRI issuperior to CT in the investigation ofheadache, it is probably unjustifiable to usethe more expensive method as a primaryimaging test.

EPILEPSYThe contribution of imaging to investigationof patients whose illness is dominated by orconsists primarily of epileptic attacks has tra-ditionally been the detection of potentiallytreatable causes, typified by intracranialtumours or vascular malformations. With theadvent of CT, causative lesions not amenableto such treatment-small cortical infarcts, orthe effects of previous trauma-could be iden-tified. This may have satisfied the neurolo-gist's curiosity, but positive effects onmanagement were minor.8' When theincreased sensitivity of MRI to parenchymalpathology first became evident, it was sug-gested that the detection rate for lesionsresponsible for epilepsy was very much higherthan when using CT. This has not beenentirely borne out by subsequent experienceand, with some exceptions, the abnormalitiesto which the newer technique is indeed moresensitive tend to be those in which their pres-ence explains why the patient should have fitsrather than those to which present surgicalapproaches are suited.Thus when a patient presents with seizures,

the aim of imaging should be. firstly to deter-mine whether he or she has a lesion that

should be treated itself, and of which the fitsare simply the clinical manifestation. A con-vexity meningioma is the paradigm; in a rela-tively young patient the likely morbidity ofinvestigation and treatment is outweighed notmerely by the presentation, but also by thenatural history. When preliminary imagingshows a complex arteriovenous malformationin an eloquent area in someone of middle age,the equation is more finely balanced, but willdepend to some extent on imaging data.Needless to say, plain films and radionuclideinvestigations are sufficiently insensitive as tobe effectively useless, but there is little evi-dence that for detection of such "surgical"lesions MRI is superior to CT, although fur-ther investigations may be required beforesurgical decision making. When CT is normalin patients whose epilepsy is not so severe asto require more radical surgery, there is littleor no justification for further investigation.There are some adults and children in whomMRI will show anomalies of cerebral cortica-tion, or heterotopic grey matter, etc, but if,given today's surgical possibilities, a mildlyaffected patient would not be a candidate forcerebral resection surgery, this type of investi-gation cannot be justified in clinical practice,other than as research.When the fits are so intractable, however,

or the side effects of conservative treatment soonerous that the patient would be offeredsurgery as primary treatment, MRI is theexamination of choice. This is particularlytrue when the presumed epileptogenic focuslies in the temporal lobe, to which CT isrecognised to be relatively blind. Thin coronalTI and T2 weighted images, which permitvolumetric analysis and three dimensionalreconstructions, can show not only hamar-tomas and indolent tumours, but minordegrees of hippocampal atrophy and signalchange, indicating mesial temporal sclerosis.As always when images are used to generatemeasurements there are important method-ological sources of error, from which a num-ber of the most often cited publications maynot be free.82 Even when these are taken intoconsideration, for successful surgery theresults must be considered in combinationwith clinical and electrophysiological data,including preoperative tests of cerebral domi-nance using intra-arterial sodium amytal.Possibly because of the expense of both tech-niques, the combination of MRI with magne-toencephalography,83 to produce what at leastone radiologist, possibly with political fore-sight, has termed "magnetic source imaging",has not become widespread as yet. It has beensuggested that interictal or particularly ictaland periictal functional radionuclide studiesgive more demonstrative lateralising data84but, even so, MRI demonstration of detailedanatomy greatly assists the neurosurgeon. Afew years ago, some centres were reportingsuccessful outcomes of surgery for mesialtemporal sclerosis in as few as 60% ofpatients.82 This should improve with currentMRI techniques, and in the near future intra-operative MRI will almost certainly prove

17




18

extremely valuable in guiding the extent ofresection.

MOVEMENT DISORDERSRegrettably, static imaging methods, con-cerned with structure rather than function,have contributed little to clinical investigationof involuntary movements, except where theseare symptomatic of structural disease, as inthe rare meningioma causing tremor.Demonstration by high field MRI of putativechanges in brain iron distribution in parkin-sonism variants was very enthusiastically pro-moted some years ago,85 but has sincespectacularly failed to change clinical practice.Some disorders, such as Hallervorden-

Spatz disease, have highly suggestive MRIfindings (the "eye of the tiger" globus pal-lidus).86 The findings in Wilson's disease,however, are variable, inconstant (althoughmuch less so than with CT), and do not nec-essarily correlate with clinical status orresponse to treatment87; they may serve tostrengthen a clinical diagnosis, but clearly donot replace metabolic investigations. Theatrophic changes seen in "multiple systematrophy" are similarly non-specific and bolsterclinical impressions rather than throwing upunsuspected diagnoses.88

DEMENTIAInvestigation of patients presenting with orfound to have dementia is highly variable inpractice, although the underlying controversyis rarely articulated. Published evidence is cer-tainly contradictory. On the one hand, thereare studies, often initiated by radiologists,which show the very poor cost-benefit gainsfrom imaging in such patients, especiallythose without neurological signs and that,contrary to widespread belief, the proportionof patients with genuinely treatable lesionsdoes not increase greatly with earlier onset ofdementia.89 On the other hand, hardly anissue of a self respecting neurological journalappears without an article on changes thatmay be shown by detailed analysis of expen-sive investigations (CT, MRI, SPECT, orPET) in demented patients. Few, if any, ofthese have shown any useful, reliable correla-tion between structural or even functionalimaging and specific causes of dementia(rather than dementia in general); indeed, inmost published studies the accepted diagnosisis that made on clinical and psychometricgrounds. This is especially regrettable, as,despite some claims to greater accuracy, atleast one series, in which clinical, radiological,and neurophysiological assessments werecompared with postmortem findings, revealedonly 47% sensitivity and 65% specificity indiagnosing Alzheimer's disease.90 It may betrue that other neuropsychologists are betterat clinical differential diagnosis, but those whohave not put themselves to the test should notassume this to be the case. Nevertheless, thefrequency and "scientific" tone of such stud-ies seem to have convinced some neurologiststhat they should be carrying out these tests asdiagnostic aids on their patients. Were some

specific treatment for Alzheimer's disease, forexample, to become available, it would becomprehensible that an imaging test of highspecificity and sensitivity might be used toselect which demented patients were treated.But unless that test were quick, cheap, freelyavailable, and virtually infallible, it wouldseem unethical not to extend treatment toother demented patients, unless it provedunacceptably toxic or expensive; and neitherthat test nor the treatment are currently tohand. It has been suggested that simple linearmeasurement of the distance between themedial border of the uiicus on the two sidesmight be useful for identifying patients withAlzheimer's disease, but some groups havebeen unable to distinguish patients with mildto moderate dementia from normal subjectsof the same age with this criterion.9' A recentupdate on the OPTIMA group's suggestionthat measurement of the minimum width ofmedial temporal lobe tissue at the level of thebrain stem92 indicates that this simple testdoes have high discriminant value; however, itis by definition not useful until the disease hasalready reached a stage at which structuralchanges have occurred.

It cannot be overemphasised that loss ofbrain volume is a normal phenomenon inelderly people, and its demonstration cannotbe assumed to have advanced the investiga-tion of dementia. For this reason many radiol-ogists prefer the term "cerebral involution",because of the traditionally "pathological"overtones of "atrophy". It should also beunderstood that correlations between thelesions due to small vessel ischaemia whichcan be shown by CT or, with greater sensitiv-ity by MRI, and cognitive decline is suffi-ciently vague to be of no clinical value in theindividual case, particularly in the elderly,mildly demented patient.93 In a number ofbrain diseases known to be associated in somecases with dementia (such as parkinsonism),although some workers have suggested focalatrophy of the medial frontal lobes andaround the third ventricle,94 others have foundno good correlation with specific imagingchanges.95 Not surprisingly, some correlationbetween cognitive decline and the extent ofbrain disease has been identified in multiplesclerosis.96

It was to be hoped that modern imagingmight prove more helpful than invasive tech-niques in diagnosis of early Huntington's dis-ease, but clinical features and, of course,family history and DNA analysis seem to bebetter guides. It has been suggested, however,that the degree of atrophy of the caudatenucleus may be a guide to prognosis.97A persistent controversy concerns the

frequency (perhaps even the existence) ofnorrnal pressure hydrocephalus. In Britainthis is a rather infrequent diagnosis in patientswithout a clear antecedent history of intra-cranial bleeding or infection, and very smallnumbers are subjected to shunt procedures onthe basis of a diagnosis of idiopathic normalpressure hydrocephalus, but one report des-cribes 22 patients in whom normal pressure

Moseley





hydrocephalus had been diagnosed under-going a non-standard MRI procedure beforeoperation in a relatively short time at a singleGerman university hospital.98 When the clinicaland laboratory criteria for the diagnosis seemso geographically variable, assessment ofclaims about the therapeutic significance ofputative radiological signs is virtually impossi-ble. With recent reports about the relevanceof MRI findings of concomitant white matterdisease and arterial hypertension to the patho-physiology,99 water on the brain is probably atits muddiest ever, at least in the imaging liter-ature.'00

OTHER NEUROPSYCHIATRIC DISORDERSSeveral imaging studies in recent years haveapplied CT, MRI, and radionuclide methodsto patients with schizophrenia, autism, psy-chosis, etc.1'0 The range of imaging featuresexamined has been extensive and various, andfew groups have used widely accepted criteria.In addition, many workers have published thestatistical analyses of their findings rather thanthe measurements themselves, which areessential for replication and comparison. Themain thrust seems to have been to show thatpeople with psychiatric disorders do indeedhave structural abnormalities, a fact whichsome psychiatrists, presumably battling withtheir psychoanalytical colleagues, find "excit-ing'02 As yet, the data are too soft and con-tradictory for any clue to pathophysiology tohave emerged, and their clinical relevance isdoubtful.

Neuroendocrine diseaseThe history of the application of imaging tothe investigation of neuroendocrine disease,specifically to small but hormonally activepituitary tumours, makes entertaining read-ing. As each of a series of techniques hasbecome available, it has made clear the highlevel of self deception by both radiologists andendocrinologists necessary for the use of itspredecessor. Thus plain radiography, conven-tional tomography, and even CT103 have allbeen shown to be extremely unreliable in thelight of later technology; the particularly highinaccuracy of CT was recognised even beforecomparisons were made with MRI.'04Although MRI is also not infallible, it wouldseem to be by far the best investigation cur-rently available. As I have pointed out previ-ously, however, "the desirability of locatingthe tumour within the pituitary gland of apatient with hyperprolactinaemia is predi-cated solely and absolutely on surgery beingthe treatment of choice"; it is otiose whenchemotherapy will be employed.When larger pituitary region tumours are

present, the multiplanar capacity of MRI,linked to MRA for demonstration of the vas-cular anatomy, renders this the technique ofchoice. Combining image quality with econ-omy, in ideal circumstances, it is probablynecessary to use only a fat suppressed contrastenhanced TI weighted volume acquisitionbut once again this will depend to some extent

on the technical possibilities of the apparatusavailable. This simple technique can also beused for follow up after surgery, although, as Ihave indicated elsewhere, clinicians' fondnessfor follow up studies may have to yield toaudits showing their lack of cost or benefiteffectiveness. This would seem to be particu-larly true when women being treated with, forexample, bromocriptine are asymptomatic asregards both their hormonal status and vision.The greater sensitivity to both the internal

structure of pituitary tumours and their rela-tions with the anterior optic pathways makeMRI the ideal method for investigating sus-pected pituitary apoplexy, and for decidingwhich patients actually need treatment, as theapoplectic episode may itself result in involu-tion of tumour usually found to underlie it. O5

Magnetic resonance imaging has openeddoors to other endocrine disturbances towhich CT was effectively blind. As an exam-ple, the normal posterior lobe of the pituitarygives high signal on Ti weighted MRI, forreasons not entirely clear. This is absent inmost patients with central or nephrogenic dia-betes insipidus, but not in those with primarypolydipsia.106

ConclusionsIt is difficult to draw overall conclusions fromthe foregoing, as I have tried to indicate ineach case the sort of selection process that theneurologist or neurosurgeon should makewhen thinking of referring a patient for imag-ing. To best use his services to the patient'sadvantage, the clinician should: (a) refrainfrom any imaging for which there is no clearclinical indication and (b) prefer anodyne toinvasive, cheap to expensive, but (c) not per-form a harmless, low cost examination if itwill not obviate a more aggressive or expen-sive one.

In Britain the legal power to determinewhether any imaging investigation is carriedout is unequivocally vested in the radiolo-gist.107 This gives force to the Royal College ofRadiologists' uncontentious recommendationthat "if you are in doubt as to whether aninvestigation is required or which investiga-tion is best, it makes sense to ask the radiolo-gists who, like other consultants, will knowmuch more about their speciality than thosewhose primary interests are in other fields".20

1 Dumoulin CL, Hart HR. MR angiography. Radiology1986;161:717-20.

2 Nitz WR, Bradley WG, Watanabe AS, Lee RR, BurgoyneB, O'Sullivan RM, et al. Flow dynamics of cerebrospinalfluid: assessment with phase-contrast velocity MRimaging performed with retrospective cardiac gating.Radiology 1992;183:395-405.

3 Gideon P, Sorensen PS, Thomsen C, Stahlberg F, GjerrisF, Henriksen 0. Assessment of CSF dynamics andvenous flow in the superior sagittal sinus by MRI inidiopathic intracranial hypertension: a preliminary study.Neuroradiology 1994;36:350-4.

4 McComb JG. The usefulness of phase-contrast MRmeasurement of cerebrospinal fluid flow. AJNR Am JNeuroradiol 1993;14:1309-10.

5 Castillo M, Hudgins PA, Malko JA, Burrow BK, HoffmanJC Jr. Flow-sensitive MR imaging of ventriculo-periteonal shunts: in vitro findings, clinical applications,and pitfalls. AJNRAmJ Neuroradiol 199 1;12:667-71.

6 Le Bihan D, Breton E, Iallemand D, Grenier P, Cabanis E,Laval-Jeantet M. MR imaging of intravoxel incoherent

19




Moseley

motions: applications to diffusion and perfusion inneurological disorders. Radiology 1988;161:401-7.

7 Henkelman RM. Diffusion-weighted MR imaging: auseful adjunct to clinical diagnosis or a scientific curiosity?AJNR Am J Neuroradiol 1990;12:932-4.

8 Dorothee 0, Henning J, Ernst T. Human brain tumors:assessment with in vivo proton spectroscopy. Radiology1993;186:745-52.

9 Vion-Dury J, Meyerhoff DJ, Cozzone PI, Weiner MW.What might be the impact on neurology of the analysisof brain metabolism by in vivo magnetic resonancespectroscopy? J Neurol 1994;241:354-71.

10 Rubin JM, Mirfakhree M, Duda EE, Dohrmann GJ,Brown F. Intraoperative ultrasound examination of thebrain. Radiology 1980;137:831-2.

11 Counsell CE, Fraser H, Sandercock PAG. ArchieCochrane's challenge: can periodically updated reviewsof all randomised controlled trials relevant to neurologyand neurosurgery be produced? J Neurol NeurosurgPsychiatry 1994;57:529-33.

12 Daves ML. Radiologic overkill. JAMA 1967;200:999-1000.

13 Kent DL, Larson EB. Magnetic resonance imaging of thebrain and spine: is clinical efficacy established after thefirst decade? Ann Intern Med 1988;108:402-24.

14 Dixon AK, Southern JP, Teale A, Freer CEL, Hall LD,Williams A, et al. Magnetic resonance imaging of thehead and spine: effective for the clinician or the patient?BMJ 1991;302:79-82.

15 Friedman D, Rapaport R. Routine use of contrast-enhanced MR scans in AIDS. AJNR Am J Neuroradiol1993;14: 1324.

16 Zimmerman RD. Routine use of contrast-enhanced MRscans in AIDS. AJNR Am J Neuroradiol 1993;14:1326-84.

17 McLean SAM. A patient's right to know. Information dis-closure, the doctor and the law. Aldershot, Dartmouth,1989:14.

18 Moseley IF. Long term effects of the introduction of non-invasive investigations in neuroradiology. Part 1: overalltrends. Neuroradiology 1988;30: 187-92.

19 Tatler GLV, Moseley IF. Use of invasive neuroradiologicalinvestigations in patients with normal computerisedtomography. BMJ 1982;ii: 1026-8.

20 Roberts GM, ed. Making the best use of a Department ofClinical Radiology. Guidelines for doctors 2nd ed. London:Royal College of Radiologists, 1993.

21 Ansell G. Clinical considerations in the administration ofcontrast media to high-risk patients. In: Carr DH, ed.Contrast media. Edinburgh: Churchill Livingstone, 1988:78-88.

22 Feuerman T, Wackym PA, Gade GF, Becker DP. Value ofskull radiography, head computed tomographic scan-ning, and admission for observation in cases of minorhead injury. Neurosurgery 1988;22:449-53.

23 Hackney DB. Skull radiography in the evaluation oftrauma: a survey of current practice. Radiology 1991;181:711-4.

24 Colquhoun IR. CT cisternography in the investigation ofcerebrospinal fluid rhinorrhoea. Clin Radiol 1993;47:403-8.

25 Anon. Guidelines for the determination of death: reportof the medical consultants on the diagnosis of death tothe President's Commission for the study of ethicalproblems in medicine and biomedical and behavioralresearch. JAMA 1981;246:2184-6.

26 Griner PF, Glaser RJ. Misuse of laboratory tests and diag-nostic procedures. N Engl J Med 1982;207: 1336-9.

27 Moseley IF. The plain radiograph in ophthalmology: awasteful and potentially dangerous anachronism. JR SocMed 1991;84:76-80.

28 Moseley IF. The diagnostic value of "optic foramenviews": experience from an eye hospital. Bry Ophthalmol1990;74:235-7.

29 Lewis TT, Kingsley DPE, Moseley IF. Do bilateral opticnerve meningiomas exist? BrJ Neurosurg 199 1;5:13-8.

30 Miller DH, Ormerod IEC, McDonald WI, ManManusDG, Kendall BE, Kingsley DPE, et al. The early risk ofmultiple sclerosis after optic neuritis. J Neurol NeurosurgPsychiatry 1988;51:1569-71.

31 Guy J, Mancuso A, Beck R, Moster ML, Sedwick LA,Quisling RG, et al. Radiation-induced optic neuropathy:a magnetic resonance imaging study. Y Neurosurg 1991;74:426-32.

32 Steinsapir KD, Goldberg RA. Traumatic optic neuro-pathy. Surv Ophthalmol 1994;38:487-518.

33 Zimmerman C, Schatz NJ, Glaser JS. Magnetic resonanceimaging of optic nerve meningiomas. Enhancement withgadolinium-DPTA. Ophthalmology 1990;97:585-91.

34 Day AL. Aneurysms of the ophthalmic segment. A clinicaland anatomical analysis. J Neurosurg 1990;72:677-9 1.

35 Jacobson DM, Karanja PN, Olson KA, Warner JJ.Computed tomography ventricular size has no value indiagnosis pseudotumor cerebri. Neurology 1990;40:1454-5.

36 Bronstein AM, Rudge P, Gresty MA, du Boulay G, MorrisJ. Abnormalities of horizontal gaze and magneticresonance findings. II. Gaze palsy and internuclearophthalmoplegia. J Neurol Neurosurg Psychiatry 1990;53:200-7.

37 Schwaber MK, Larson TC, Zealear DL, Creasy J.Gadolinium-enhanced magnetic resonance imaging inBell's palsy. Laryngoscope 1990;100:1264-9.

38 Anderson RE, Laskoff JM. Ramsay Hunt syndromemimicking intracanalicular acoustic neuroma on

contrast-enhanced MR. AJNR Am J Neuroradiol 1990;11:409.

39 Anon. Imaging patients with suspected acoustic neuroma.Lancet 1988;ii: 1294.

40 Welling DB, Glasscock ME, Woods CI, Jackson CG.Acoustic neuroma: a cost-effective approach. OtolaryngolHead Neck Surg 1990;103:364-70.

41 Consensus Development Panel. National Institutes ofHealth Consensus Development Conference statementon acoustic neuroma, 11-13 December 1991. ArchNeurol 1991;51:201-7.

42 Renowden SA, Anslow P. The effective use of magneticresonance imaging in the diagnosis of acoustic neuro-mas. Clin Radiol 1993;48:25-8.

43 Bernardi B, Zimmerman RA, Savino PJ, Adler C.Magnetic resonance tomographic angiography in theinvestigation of hemifacial spasm. Neuroradiology 1993;35:606-11.

44 Bradley WG. MR of the brain stem: a practical approach.Radiology 1991;179:319-32.

45 Savy LE, Stevens JM, Taylor DJ, Kendall BE. Apparentcerebellar ectopia: a reappraisal using volumetric MRI.Neuroradiology 1994;36:360-3.

46 Asbury AK, Herndon RM, McFarland HF, McDonaldWI, McIlroy WJ, Paty DW, et al. National MultipleSclerosis Society working group on neuroimaging for theMedical Advisory Board. Neuroradiology 1987;29: 119.

47 Miller DH, Ormerod IEC, Rudge P, Kendall BE, MoseleyIF, McDonald WI. The early risk of multiple sclerosisfollowing isolated syndromes of the brainstem and spinalcord. Ann Neurol 1989;26:635-9.

48 Rudge P, Ali A, Cruickshank JC. Multiple sclerosis, tropicalspastic paraparesis, and HTLV-1 infection in Afro-Caribbean patients in the United Kingdom. Y NeurolNeurosurg Psychiatry 1991;54:689-94.

49 Miller DH, Ormerod IEC, Gibson A, du Boulay EPGH,Rudge P, McDonald WI. MR brain scanning in patientswith vasculitis: differentiation from multiple sclerosis.Neuroradiology 1987;29:226-31.

50 Banna M, El-Ramahi K. Neurologic involvement inBehcet disease: imaging findings in 16 patients. AJNRAmY Neuroradiol 1991;12:791-6.

51 Fazekas F, Offenbacher H, Fuchs S, Schmidt R,Niederkorn K, Horner S, et al. Criteria for an increasedspecificity of MRI interpretation in elderly subjects withsuspected multiple sclerosis. Neurology 1988;38: 1822-5.

52 Viniuela FV, Fox AJ, Debrun GM, Feasby TE, Ebers GC.New perspectives in computed tomography of multiplesclerosis. Am J Radiol 1982;139: 123-7.

53 Kesselring J, Miller DH, Robb SA, Kendall BE, MoseleyIF, Kingsley D, et al. Acute disseminated encephalo-myelitis. MRI findings and the distinction from multiplesclerosis. Brain 1990;113:291-302.

54 Lunsford LD, Martinez AJ, Latchaw RE, Pazin GJ. Rapidand accurate diagnosis of herpes simplex encephalitiswith computed tomographic stereotaxic biopsy. SurgNeurol 1984;21:249-57.

55 Fernandez RE, Rothberg M, Ferencz G, Wujack D. Lymedisease of the CNS: MR imaging findings in 14 cases.AJNR Am J Neuroradiol 1990;11:479-81.

56 Tien RD, Chu PK, Hesselink JR, Duberg A, Clayton W.Intracranial cryptococcosis in immunocompromisedpatients: CT and MR findings in 29 cases. AJNR Am JNeuroradiol 1991;12:283-9.

57 Mercader JM, Perich J, Berenguer J, Pujol T, Cardenal C.Intracranial tuberculoma in AIDS: neuroradiologicalfindings. Neuroradiology 1991;33:569-70.

58 Dina TS. Primary central nervous system lymphomaversus toxoplasmosis in AIDS. Radiology 1991;179:823-8.

59 Moseley IF. Radiological investigation of cerebral masses.Current Imaging 1991;3:31-6.

60 Hillemacher A. Die Wert amnesticher and klinische Datensowie apparativer Intersuchundsbefunde bei derDiagnose von Hirntumoren. Fortschr Neurol Psychiatr1982;50:93-112.

61 Baker HL, Houser OW, Campbell JK. National CancerInstitute study: evaluation of computed tomography inthe diagnosis of intracranial neoplasms. I. Overallresults. Radiology 1980;136:91-6.

62 Haughton HV, Rimm AA, Sobocinski KA, Papke RA,Daniels DL, Williams AL, et al. Blinded clinical com-

parison of MR imaging and CT in neuroradiology.Radiology 1986;160:751-5.

63 Kazner E, Wende S, Grumme T, Lanksch W, Stochdorph0. Computed tomography in intracranial tumours. Berlin:Springer-Verlag, 1982.

64 Bolender NF, Cromwell LD, Graves V, Margolis MT,Kerber CW, Wendling L. Interval appearance ofglioblastoma not evident in previous CT examination.Y Comput Assist Tomogr 1983;7:599-605.

65 Enzmann DR, Krikorian J, Yorke C, Hayward R.Computed tomography in leptomeningeal spread oftumours. J Comput Assist Tomogr 1978;2:448-55.

66 Sze G, Milano E, Johnson C, Heier L. Detection of brainmetastases: comparison of contrast-enhanced MR withunenhanced MR and enhanced CT. AJNR Am JfNeuroradiol 1990;11:785-91.

67 Ruggiero G, Picard L. The true value of computed tomog-raphy, angiography and pneumoencephalography for theprecise diagnosis of the brain tumours. In: WackenheimA, du Boulay GH (eds) Choices and characteristics in com-

puterized tomography. Amsterdam: Kugler, 1980:13-21.68 De Angelis LM. Primary nervous system lymphoma: a new

clinical challenge. Neurology 1991;41:619-21.

20





69 Davies AG, Richardson RB, Bourne SP, Kemshead JT,Coakham HB. Immunolocalisation of human braintumours. In: Bleehen NM, ed. Tumours of the brain.Berlin: Springer-Verlag, 1986, 83-99.

70 Asari S, Makabe T, Katayama S, Itoh T, Tsuchida S,Ohmoto T. Assessment of the pathological grade ofastrocytic gliomas using an MRI score. Neuroradiology1994;36:308-10.

71 Jelinek J, Smirniotopoulos JG, Parisi JE, Kanzer M.Lateral ventricular neoplasms of the brain: differentialdiagnosis based on clinical, CT and MR findings. AJNRAm J Neuroradiol 1990;11:567-74.

72 Peeling J, Sutherland G, Marat K, Tomchuk E, Bock E.1H and 13C nuclear magnetic resonance studies ofplasma from patients with primary intracranial neo-plasms. Jf Neurosurg 1988;68:931-7.

73 Carpeggiani P, Crisi G, Trevisan C. MRI of intracranialmeningiomas: correlations with histology and physicalconsistency. Neuroradiology 1993;35:532-6.

74 Brismar J, Cronqvist S. Therapeutic embolisation in theexternal carotid artery region. Acta Radiologies 1978;19:715-31.

75 Halbach W, Hieshima GB, Higashida RT, David CF.Endovascular therapy of head and neck tumors. In:Vifiuela F, Halbach W, Dion JE, eds. Inteventionalneuroradiology. Endovascular therapy of the central nermoussystem. New York: Raven Press, 1992:17-28.

76 Bergstrom M, Collins P, Ehrin E, Ericson K, Ericsson L,Greitz T, et al. Discrepancies in brain tumour extent asshown by computed tomography and positron emissiontomography using [68Ga]EDTA, ['Clglucose, and["C]methionine. J Comput Assist Tomogr 1983;7:1062-6.

77 Piani C, Pasquini U, De Nicola M, Bevilacqua F,Menichelli F, Salvolini U, et al. Intraarterial ACNUchemotherapy in malignant cerebral gliomas: prelimin-ary results. Neuroradiology 1991;33:319-21.

78 Schwartz RB, Carvalho PA, Alexander E III, Loeffler JS,Folkerth R, Holman BL. Radiation necrosis vs high-grade recurrent glioma: differentiation by using dual-isotope SPECT with 20T1 and 99-Tc-HMPAO. AJNRAmJNeuroradiol 1991;12:1187-92.

79 Larson EB, Omenn GS, Lewis H. Diagnostic evaluation ofheadache. Impact of computerised tomography andcost-effectiveness. jAMA 1980;243:359-62.

80 Heller CA, Stanley P, Lewis-Jones B, Heller RF. Value ofxray examinations of the cervical spine. BMJ 1983;287:1276-8.

81 Daras M, Tuchman AJ, Strobos RJ. Computed tomog-raphy in adult onset epilepsy. Can 7 Neurol SCi 1987;14:286-9.

82 Morrell F. In vivo imaging of human anatomy in tem-poral lobe epilepsy. A3'NR Am I Neuroradiol 1991;12:948-9.

83 Orrison WW, Davis LE, Sullivan GW, Mettler FA Jr,Flynn ER. Anatomic localization of cerebral corticalfunction by magnetoencephalography combined withMR imaging and CT. A3tNR Am J Neuroradiol 1990;11:713-6.

84 Borbely K, Balogh A, Halasz P, Vajda J, Czirjak S, Nyary I.Reliability of ictal and peri-ictal SPECT investigations inlocalizing epileptogenic foci in patients undergoingepilepsy surgery. Neuroradiology 1994;36(suppl):S76.

85 Drayer BP, Olanow W, Burger P, Johnson GA, HerfkensR, Riederer S. Parkinsonism plus syndrome: diagnosisusing high field MR imaging of brain iron. Radiology1986;159:493-8.

86 Savoiardo M, Halliday WC, Nardocci N, Strada L,D'Incerti L, Angelini L, et al. Hallervorden-Spatz dis-ease: MR and pathological findings. AJNR Am JNeuroradiol 1993;14: 155-62.

87 Nazer H, Brismar J, Al-Kawi MZ, Gunasekaran TS, JorulfKH. Magnetic resonance imaging of the brain inWilson's disease. Neuroradiology 1993;35: 130-3.

88 Savoiardo M, Strada L, Girotti F, Zimmerman RA, Grisoli

M, Testa D, et al. Olivopontocerebellar atrophy: MRdiagnosis and relationship to multisystem atrophy.Radiology 1990;174:693-6.

89 Bradshaw JR, Thomson JLG, Campbell MJ. Computedtomography in the investigation of dementia. BMJ1983;286:277-80.

90 Ettun TM, Staehelin HB, Kischka U, Ulrich J, Scollo-Lavizzari G, Wiggli U, et al. Computed tomography,electroencephalography and clinical features in thedifferential diagnosis of senile dementia. Arch Neurol1989;46:1217-20.

91 Early B, Rodrigo Escalona P, Boyko OB, DoraiswamyPM, Axelson DA, Patterson L, et al. Interuncal dis-tance in healthy volunteers and in patients withAlzheimer disease. AJNR Am J Neuroradiol 1993;14:907-10.

92 Jobst KA, Smith AD, Szatmari M, Molyneux A, EsiriME, King E, et al. Detection in life of confirmedAlzheimer's disease using a simple measurement oftemporal lobe atrophy by computed tomography.Lancet 1992;340:1179-83.

93 Erkinjuntti T, Gao F, Lee DH, Eliasziw M, Merskey H,Hachinski VC. Lack of difference in brain hyper-intensities between patients with early Alzheimer's dis-ease and control subjects. Arch Neurol 1994;51:260-8.

94 Steiner I, Gomori JM, Melamed E. Features of brainatrophy in Parkinson's disease. Neuroradiology 1985;27:158-60.

95 Huber SJ, Shuttleworth EC, Christy JA, Chakeres DW,Curtin A, Paulson GW. Magnetic resonance imaging indementia of Parldnson's disease. J Neurol NeurosurgPsychiatry 1989;52:1221-7.

96 Callanan MM, Logsdail SJ, Ron MA, Warrington ER.Cognitive impairment in patients with clinically isolatedlesions of the type seen in multiple sclerosis. Brain1989;112:361-74.

97 Wardlaw JM, Sellar RJ. Early caudate nucleus atrophy inHuntington's disease: does it correlate with presentingsymptoms? Neuroradiology 1991;33:238-40.

98 Wakhloo AK, Jungling F, Krauss JK, Schumacher M,Hennig J. Measurements of pulsatile brain motionwith MR-interferography in patients with NPH.Neuroradiology 1992;34(suppl):S47.

99 BradleyWG Jr, Whittemore AR, Watanabe AS, Davis SJ,Teresi LM, Homyak M. Association of deep whitematter infarction with chronic communicatinghydrocephalus: implications regarding the possibleorigin of normal-pressure hydrocephalus. AJNR Am JNeuroradiol 1991;12:31-9.

100 George AE. Chronic communicating hydrocephalus andperiventricular white matter disease: a debate withregard to cause and effect. AJNR Am J Neuroradiol1991;12:42-4.

101 Moseley I. Neuropsychiatric disease. Curr Opin NeurolNeurosurg 1990;3:884-9.

102 Andreasen NC. Neuroradiology and neuropsychiatry: anew alliance. AJNRAmJYNeuroradiol 1992;13:841-3.

103 Johnson MR, Hoare RD, Cox T, Dawson JM, Maccabe JJ,Llewellyn DEH, et al. The evaluation of patients with asuspected pituitary microadenoma: computed tomo-graphy compared to magnetic resonance imaging. ClinEndocrinol 1992;36:335-8.

104 Marcovitz S, Wee R, Chan J, Hardy J. The diagnosticaccuracy of preoperative CT scanning in the evaluationof pituitary ACTH-secreting adenomas. AJNR Am YNeuroradiol 1987;8:641-4.

105 L'Huillier F, Combes C, Martin N, Leclerc X, Pruco JP,Gaston A. MRI in the diagnosis of so-called pituitaryapoplexy. 7 Neuroradiol 1989;16:221-37.

106 Moses AM, Clayton B, Hochhauser L. Use of Ti-weighted MR imaging to differentiate between primarypolydipsia and central diabetes insipidus. AJ7NR Am JNeuroradiol 1992;13:1273-7.

107 Moseley IF. Diagnostc imaging in neurological disease.Edinburgh: Churchill Livingstone, 1986:1-2.

21




Imaging the adult brain.

I Moseley

doi: 10.1136/jnnp.58.1.71995 58: 7-21 J Neurol Neurosurg Psychiatry

http://jnnp.bmj.com/content/58/1/7.citationUpdated information and services can be found at:

These include:

serviceEmail alerting

box at the top right corner of the online article. Receive free email alerts when new articles cite this article. Sign up in the

Notes

http://group.bmj.com/group/rights-licensing/permissionsTo request permissions go to:

http://journals.bmj.com/cgi/reprintformTo order reprints go to:

http://group.bmj.com/subscribe/To subscribe to BMJ go to:


http://jnnp.bmj.com/content/58/1/7.citation

http://group.bmj.com/group/rights-licensing/permissions

http://journals.bmj.com/cgi/reprintform

http://group.bmj.com/subscribe/



Documents

Imaging the adult brain