Minireview - Home | Molecular Cancer Therapeutics · lesions within the brain influences the decision as to whether the most appropriate treatment is surgery, radiosurgery, or whole

Minireview

Multivoxel Magnetic Resonance Spectroscopy ofBrain Tumors1

Sarah J. Nelson2

Magnetic Resonance Science Center, University of California SanFrancisco, San Francisco, California 94143-1290

AbstractMagnetic resonance spectroscopy (MRS) is becomingmore widely available for clinical applications and isable to provide information about the metabolicproperties of regions of normal and abnormal tissuemorphology. A critical question for the clinicalmanagement of patients with brain tumors is whethermultivoxel MRS is able to add new information to thehigh-quality anatomical data provided by conventionalMR imaging techniques and whether this informationis relevant for the diagnosis and clinical managementof such patients. In this article, the state of the artfor acquiring and analyzing multivoxel MRS data isreviewed and placed in context relative to imagingfindings for metastatic and primary brain tumors. TheMRS data are seen to provide unique information thatwhen combined with high-quality anatomical MRimages has implications for defining tumor typeand grade, directing biopsy or surgical resection,planning focal radiation or biological therapies, andunderstanding the mechanisms of success and failureof new treatments.

IntroductionAlthough patients with brain tumors typically have a relativelypoor prognosis, the time to progression and median survivalcan be mediated for selected subpopulations of patients byapplying aggressive therapy (1–7). Selecting the treatmentthat is most appropriate for a specific patient and directingthat therapy to the region of active tumor is crucial for achiev-ing the best possible outcome. Critical factors in evaluatingprognosis are tumor type, grade, and volume (1, 5, 8, 9).Despite the excellent soft tissue contrast provided by MRI3,the sensitivity and specificity with which this modality definestumor type and grade is limited. This is partly attributable tothe existence of Gadolinium-enhanced necrosis that may bemistaken for tumor and partly to the difficulty in distinguish-

ing between tumor, edema, and nonspecific treatment ef-fects in the region of hypointensity on T2-weighted images(10, 11).

Overcoming these problems requires the development ofnew imaging modalities that highlight functional or metabolicproperties of the tumor. Although there have been studiesthat have investigated positron emission tomography andsingle photon emission tomography as candidates for suchanalysis (12–15), there would be a considerable savings incost and patient discomfort if similar information could bedefined using an MR methodology. Two new imaging meth-ods that have been considered for this application are per-fusion-weighted and diffusion-weighted MRI (16–23). Thesemake use of echo planar pulse sequences to examine thetissue architecture and microvasculature in the lesion andsurrounding brain parenchyma. Although they do provideinformation about physiological properties of the tumor thathave been linked to cellularity, structural integrity, and an-giogenesis, they are not yet used routinely for clinical man-agement of patients.

Single voxel proton MR spectroscopy has been applied forcharacterizing the metabolic signatures of brain tumors forsome time (24–32). There is strong evidence for a reductionin N-acetylaspartate and increase in choline containing com-pounds in tumor relative to normal brain parenchyma (28,29). Also, for some metastatic lesions and for high-gradegliomas, typically, there are resonances corresponding tolactate or lipids (26, 27). Initial reports concerning the poten-tial for single voxel MRS in tumor grading gave varied resultswith a large variability in data quality between institutions(30). With the introduction of automated packages for per-forming single voxel proton MRI on clinical scanners, thequality and reproducibility of the data have been improved.Multivariate statistical analysis procedures have also beenapplied in an attempt to identify patterns that characterizespecific tumor types and grades (31, 32). More recent studieshave also included the acquisition of spectra with both shortand long echo times so that levels of myo-inositol, glutamine,and glutamate can be included in the analysis and providethe potential for improved discrimination between differenttypes of lesions (33, 34).

Although single voxel proton MRS is a relatively rapidmethod for obtaining information about the metabolism in a4–8-cc region within the lesion, it does not address spatialheterogeneity and is unable to contribute to defining thespatial extent of the lesion. These factors are particularlyimportant for planning focal treatments such as radiation andsurgical resection and for following response to therapy. Toaddress these issues, it is necessary to consider multivoxelproton MRSI (1-H). In this article, we examine the limitationsof conventional MRI for metastatic and primary brain tumors,

Received 12/12/02; accepted 1/28/03.1 Parts of this work were funded by NIH Grants CA79719 and CA59880.2 To whom requests for reprints should be addressed, Magnetic Reso-nance Science Center, Box 1290, 1 Irving Street, University of California,San Francisco, CA 94143-1290.3 The abbreviations used are: MRI, magnetic resonance imaging; MRS,magnetic resonance spectroscopy; MRSI, magnetic resonance spectro-scopic imaging; CNI, choline to N-acetylaspartate index; PRESS, pointresolved acquisition mode.

497Vol. 2, 497–507, May 2003 Molecular Cancer Therapeutics

Research. on October 15, 2020. © 2003 American Association for Cancermct.aacrjournals.org Downloaded from

http://mct.aacrjournals.org/

discuss techniques for obtaining 1-H MRSI of brain tumors,and present examples of situations in which the informationthat is obtained differs from conventional MRI. The objectivewill be to highlight situations where the 1-H MRSI has thepotential for changing the clinical management of the patientand thereby make a significant contribution to improving theoutcome. We will focus on the application to adult braintumors as the types of lesions that occur in pediatric patientsand prognosis are quite different.

MRI of Metastatic LesionsFor these lesions, the volume of Gadolinium enhancement onT1-weighted spin echo or gradient echo images is thought toencompass all of the active tumor. In many cases, there arealso central regions of hypointensity within the enhancinglesion that correspond to necrosis. The region of hypointen-sity on the T1-weighted image and corresponding hyperin-tensity on T2-weighted images that typically surrounds theenhancing volume is thought to correspond to edema ornonspecific treatment effects rather than to infiltrative tumor.Detection of small lesions and improved visualization oflarger enhancing lesions is possible using double or tripledoses of Gadolinium (35). The number and size of metastatic

lesions within the brain influences the decision as to whetherthe most appropriate treatment is surgery, radiosurgery, orwhole brain radiation therapy (5–7). Other factors taken intoaccount are the status of the primary lesion and metastasesin other organs.

Fig. 1 gives three examples of Gadolinium-enhancedT1-weighted and T2-weighted images from patients withbrain metastases. The first patient had a metastasis frommelanoma. Even at this time, the lesion had a small centralregion of hyperintensity corresponding to necrosis. It grewrapidly and survival was only for an additional 2 months.The second patient had a large region of central necrosisat the time of presentation for this scan. The primary lesionwas from lung and was stable at this time. Despite treat-ment with gamma knife radiosurgery, the lesion did notrespond, and the patient survived an additional 3 months.The third patient also had a primary lung lesion and hadreceived prior radiation therapy for the metastasis on theleft side of the brain. The enhancing lesion in the vermiswas seen as a new lesion on a follow-up imaging study.Treatment at this time was by gamma knife radiosurgery toboth the original and the new lesion. After radiosurgery,there was a significant reduction in the size of the newer

Fig. 1. Post-Gadolinium T1-weighted and T2-weighted MR images from 3 patients with brain metastases. The patient on the left has a metastasis froma melanoma, the patient in the center has a lung metastasis, and the patient on the right has two lung metastases.

498 Multivoxel MR Spectroscopy of Brain Tumors



lesion and the patient survived an additional 33 monthswith no additional treatment.

MRI of Primary Brain TumorsGadolinium-enhanced T1-weighted MR images from 12 pa-tients with gliomas are seen in Fig. 2. These are the mostcommon types of primary brain tumor in adults and areextremely heterogeneous in terms of imaging appearanceand in their response to therapy. They are infiltrative lesionswith poorly defined margins on both T1- and T2-weightedimages (36–39). The prognosis varies significantly with tumorgrade, ranging from a median survival of 7–10 years for grade2 lesions, 2–4 years for grade 3 lesions, and � 1 year forgrade 4 lesions (1, 2). Definition of grade is based uponhistological analysis of tissue samples obtained by biopsy orduring surgical resection (8, 9). Because it is common forthere to be regions of different tumor grade within the samelesion, directing the surgeon to the region that is likely to beof highest grade is critical to obtain representative samplesfor histological analysis. Although a tumor is known to beoutside the lesion as well, the enhancing lesion is widelyused as the target for surgery resection or for planning ra-diation therapy (11).

As can be seen in Fig. 2, grade 2 gliomas are typicallynonenhancing after injection of Gadolinium. They are iden-tified as a region of hypointensity on T1-weighted imagesand hyperintensity on T2-weighted lesions. Grade 3 glio-mas have large regions of hypointensity on T1-weightedimages but may also have regions in which the blood brainbarrier is compromised and which therefore appear brighton postcontrast T1-weighted images. Although grade 4gliomas may also be nonenhancing, it is more common forthem to have substantial regions of Gadolinium enhance-ment with a central area of hypointensity that correspondsto necrosis. Higher doses of Gadolinium do improve thevisibility of the enhancing lesion and heavily T2-weightedfluid attenuated inversion recovery images contribute todistinguishing regions of edema and nonenhancing tumorfrom cerebrospinal fluid and cysts (36 –39). Despite thewidespread use of MRI for assessment of gliomas, thereare many circumstances where the data are ambiguous interms of defining tumor margins and in distinguishingtreatment-induced necrosis from recurrent tumor. It is withthese difficulties in mind that 1-H MRSI has been pro-posed as a technique that can add to the evaluation andcharacterization of these lesions (40 –51).

Fig. 2. Post-Gadolinium T1-weighted images from patients with newly diagnosed gliomas. The top row are grade 2 lesions, the middle row are grade 3lesions, and the bottom row are grade 4.

499Molecular Cancer Therapeutics



Methods for Obtaining Multivoxel 1-H MRSIThe most straightforward method for generalizing singlevoxel MR spectroscopy is to select a larger volume of inter-est and then apply phase encoding to obtain localization toa one-, two-, or three-dimensional array of voxels (52).PRESS or stimulated echo acquisition mode (STEAM) are thetwo most common methods used for volume selection, withPRESS being preferred when the echo time allows becauseof its intrinsically higher signal to noise ratio. An advantage ofthis approach is that the volume can be selected to eliminateas much of the s.c. lipids as possible and to avoid regionslikely to cause large variations in susceptibility such as thesinuses. This permits improved shimming and providesspectra with narrower peaks and higher signal to noise (46–51). The advantage of obtaining a two-dimensional or three-dimensional array of spectra is that it is possible to observenot only heterogeneity within the lesion but to examine sur-rounding tissue that may appear normal on MRI. This pro-vides a reference for comparing metabolite levels in thetumor and makes it possible to identify regions of abnormalmetabolism outside the morphological lesion (47–50). Fortreatment planning and long-term follow-up, it is necessaryto evaluate tumor progression, and it is critical to obtainthree-dimensional coverage of a large volume of interest.

Most of the studies that have been performed to date haveused echo times of 144 or 270 ms, which provide spectradominated by five different metabolite peaks: choline; crea-tine; N-acetylasparatate; lactate; and lipid (46, 47). The cho-line peak includes a number of different choline-containingcompounds and reflects membrane synthesis and turnover.Creatine is important in cellular energetics, whereas N-acety-laspartate is a neuronal marker. Lactate reflects anaerobicmetabolism, and lipids are observed in regions of cellularbreakdown caused by necrosis. Fig. 3 shows examples ofspectra from normal brain tissue, necrosis, and regions fromdifferent types of brain tumors. The normal brain has N-acetylaspartate that is approximately twice the intensity ofcholine and creatine. Tumor generally has decreased N-

acetylaspartate and increased choline and variable levels ofcreatine. Peaks corresponding to lactate and lipid may bepresent in regions of necrosis for both metastatic and pri-mary brain tumors (47). Because the nominal voxel size for avolume head coil at 1.5 T is 1 cc, individual voxels maycontain a mixture of tumor, necrosis, and normal brain tissue.

Although it possible to obtain three-dimensional 1-H MRSIdata with chemical shift selective water suppression andconventional volume selection radiofrequency pulses, thereare many circumstances where the water and lipid suppres-sion are inadequate and compromise the quality of the dataobtained. This is especially true for patients who have hadsurgical resection and can be a severe problem for patientsthat are treated by brachytherapy using permanent radioac-tive seeds. To improve the quality of water suppression, it ispossible to implement alternative radiofrequency pulses thatare able to provide improved spatial and frequency selection(53, 54). Another critical tool for obtaining full coverage of thelesion and for sharpening the edges of the selected volumehas been the implementation of very spatially selective sat-uration bands (55). These have a very sharp transition bandand can be applied parallel to the edges of the selectivevolume to make it more cubic in shape or at an obliqueorientation to conform the volume to the anatomy.

Other approaches for obtaining volumetric coverage of thelesion are to use multislice and multiple echo time techniquesfor providing spatial localization in a time effective fashion(56–58). In this case, lipid suppression has typically beenprovided by spatial and frequency selective pulses, inversionrecovery, and spatial saturation pulses. These techniqueshave the advantage of obtaining complete in plane coveragebut may be limited close to the sinuses or surgery cavitiesbecause of susceptibility artifacts. For multislice acquisi-tions, it is necessary to have a slice gap to avoid cross-talk,and two acquisitions are required to provide complete cov-erage of the lesion. Another strategy for obtaining volumetriccoverage of the brain with a reasonable acquisition time is touse echo planar spectroscopic imaging with either oscillating

Fig. 3. Spectra from normalbrain tissue, brain metastases,necrosis, and gliomas of differ-ent grades. These spectra areselected from arrays of three-dimensional 1-H MRSI datasetswith a nominal spatial resolutionof 1 cc, echo time TE � 144 ms,repetition time TR � 1 s, andtotal acquisition time for the spa-tial array of spectra was 17 min.




gradients in one spatial dimension or spiral sampling within agiven plane (59–61). This has been shown to provide gooddata quality for normal volunteers and patients with neuro-degenerative diseases but may be limited for patients withbrain tumors once they have undergone surgical resection.Future studies are expected to use a hybrid PRESS-echoplanar spectroscopic imaging technique with spatially selec-tive saturation bands that allow greater k-space coveragebut eliminate signals from regions that are likely to causesusceptibility artifacts.

Reconstruction and Postprocessing of 1-HMRSI DataThe reconstruction of 1-H MRSI data and analysis of theresulting arrays of spectra combines fourier transforms andapodization with automated methods of spectral processingto provide data that can be interpreted by visual inspection orquantified to generate maps of the spatial distribution ofdifferent metabolites (62). The first step is to apply an apo-dization function to the k-space-free induction decays andperform a fourier transform to produce k-space spectra. Thenext step is to reconstruct the spatial dependence of thedata. For spiral or irregular k-space sampling, the approachis to first re-grid the k-space data onto a rectangular array(59–61). For conventional phase encoding, this step is un-necessary. To center the data at the most appropriate spatiallocation, it is possible to phase-weight the k-space array withthe appropriate voxel shift. This is followed by applying anyrequired spatial apodization and then performing the spatialfourier transformations. The resulting array of spectra willtypically have spatially dependent frequency and phase er-rors that need to be corrected, as well as baseline variationsbecause of residual water.

Numerous methods for estimating frequency, phase, andbaseline corrections for spectral data have been reported inthe literature (63–69). Characteristics of the 1-H MRSI datathat guide the choice of methodology are the larger numberof spectra that need to be considered, and the need forwhatever method is chosen to be robust to differences insignal to noise and peak configurations corresponding todifferent tissue types. One approach is to acquire a separatedataset with no water suppression (64). This is time consum-ing, but the high signal to noise of the water resonanceallows for an accurate estimate of frequency and phaseparameters. Provided that the data acquisition window istimed correctly, there should be no need for frequency-dependent phase correction, and the phase of the water ineach voxel should be the same as for the other metabolites.

An alternative to acquiring a separate water referencedataset is to deliberately limit the water suppression to leavebehind a relatively large water peak in the spectrum. Theaccuracy of this approach depends upon the quality of thevolume selection and out of voxel suppression because in-complete suppression of water outside the excited volumemay cause spurious peaks with different frequency andphase to be folded into the selected volume. Another strat-egy is to make use of prior knowledge of possible peaklocations (69), obtain estimates of corrections from metab-olite peaks in voxels that have sufficient signal to noise, and

then apply spatial interpolation to fill in corrections for voxelswith low signal to noise (62). This provides arrays of phasedand frequency referenced spectra for visual correlation withthe anatomy as shown in Figs. 4–7.

More quantitative analysis of the data requires the estima-tion of peak locations, heights, and areas. Prior knowledge ofrelative peak locations is valuable for performing this analysisand provides the basis for a robust method that identifiesstatistically significant peaks, determines peak heights, andcalculates peak areas by integration within a defined range offrequencies for each metabolite. Additionally, more sophis-ticated fitting algorithms can be applied to spectra that havesufficient signal to noise for the optimization routines to bereliable (e.g., 69). The output of the analysis is a number ofspatial maps of metabolite parameters that can be applied toidentify regions of normal and abnormal metabolism. Asindicated in a previous publication (62), additional correc-tions for spatial variations in intensity caused by the dataacquisition procedures may also be required if comparingrelative intensities of metabolites such as choline, creatine,N-acetylaspartate, lactate, and lipid.

Our studies have indicated that the relative increase incholine and decrease in N-aceylasparate is critical for defin-ing the spatial extent of the metabolic abnormality corre-sponding to active tumor. We have developed an index forclinical applications, termed the CNI, that describes thisquantity and can be derived automatically from a given 1-HMRSI dataset without reference to the anatomical images(70). The CNI is a statistical quantity that is normalized by therange of metabolite levels in normal tissue and can be usedto infer the probability of each voxel that is abnormal. Notethat this analysis cannot guarantee that the abnormality isattributable to tumor in a randomly selected population be-cause other pathologies may also result in increased cholineand decreased N-acetylasparate. However, it does providean objective method for highlighting regions that have spec-tral characteristics consistent with tumor. The differences inCNI with tumor grade and correlation with histology aresummarized in recent publications (70–73).4 Similar indicescan be defined to represent the regions with abnormal cho-line relative to creatine and abnormal creatine relative toN-acetylaspartate.

Differences in MRI and Metabolic LesionsA critical question for using 1-H MRSI for evaluation of braintumors is whether the spatial extent of the metabolic lesionis different from the Gadolinium-enhancing region and hy-perintensity on T2-weighted images (47). If there is no dis-tinction between these lesions, there may be no added valuefor the 1-H MRSI data over and above conventional MRimages. For metastases, the focus is on whether it is possibleto distinguish between regions of tumor and enhancing ne-crosis. In practice, it is difficult to get definitive evidencebecause it is rare for such lesions to be biopsied. Evaluation

4 I. Catalaa, R. G. Henry, W. P. Dillon, E. Graves, T. McKnight, Y. Lu, D. B.Vigneron, and S. J. Nelson. Perfusion, diffusion and spectroscopy valuesin newly diagnosed cerebral gliomas, submitted for publication.




of whether a lesion corresponds to active tumor is typicallybased on whether it subsequently gets larger on Gadolinium-enhanced MRI. Another complication is that as the lesiongets larger, it forms central necrosis and, with a voxel size of1–2 cc, it is difficult to obtain spectra free from partial vol-uming of tumor and necrosis. In a study of 18 patients withbrain metastases that were treated with gamma knife radio-surgery, we found that all but two of the lesions had de-creased N-acetylasparate, as well as a peak correspondingto lactate or lipid. Of the lesions that were followed aftertreatment, all lesions that showed decreases in the volume ofthe enhancing lesion also showed reduction in lactate, lipid,and choline peaks. There were also three lesions thatshowed reduced metabolism but had stable or slightly in-creasing volume. Lesions that showed increased enhance-ment on long-term follow-up also had a corresponding in-crease in choline and lactate or lipid peaks (47). From thisstudy, it appears that the MRSI can provide additional con-fidence in as to whether lesions are responding to therapyand in some cases may give information that is not presentin anatomical images.

The situation is more complex for patients with gliomas asthere is the need to separate tumor from necrosis and todistinguish nonenhancing tumor from edema and treatmenteffects. To evaluate the feasibility of using 1-H MRSI in this

manner, we have looked at the differences in anatomical andmetabolic lesions in patients with newly diagnosed gliomaswho had been scanned before surgical resection (50, 51,71–73)4. The goal was to determine how many of theselesions were enhancing and had the voxel with most abnor-mal metabolic signature (maximum CNI) outside the enhanc-ing volume. Of 46 grade 2 lesions, 13% were enhancing and95% had their maximum CNI in the nonenhancing region ofthe T2 lesion. The mean and the median of the maximum CNIin each lesion were both 6.9. It was much more common forthe grade 3 lesions to be enhancing than for the grade 2lesions, with 23% considered as weakly enhancing and 38%fully enhancing. The location of the maximum CNI was innonenhancing tissue for 83% of the lesions, the medianmaximum CNI was 7.1 and the mean maximum CNI was 8.0.For the 31 grade 4 patients that were studied, 75% hadvisible regions of macroscopic necrosis and all of them hadenhancing regions. There was a large variation in the maxi-mum CNI within each lesion and, although the median valuewas 6.9, there were some lesions with very high CNIs so thatthe mean value was 9.4. Fig. 5 shows examples of grades 2,3 and 4 gliomas with regions having abnormal CNI highlighted.It is clear from these types of studies that the Gadoliniumenhancement does not usually correspond to the region withhighest choline. Another interesting finding was that the region

Fig. 4. Correlation of 3-D 1-H MRSI data with the anatomy as depicted by a post-Gadolinium T1-weighted image with superimposed grid correspondingto the spectral array. These data are from a patient with a melanoma metastasis and show two axial slices from the dataset. The nominal spatial resolutionof the data were 1cc, echo time TE � 144 ms, repetition time TR � 1 s and total acquisition time was 17 min.




with abnormal CNI was typically only a subset of the T2 lesion.In some cases this was because of low metabolite levels andpresumably represented necrosis, in others it was because ofnormal appearing spectra being within the T2 lesion (73). Thesefindings support the hypothesis that 1-H MRSI is able to definethe spatial extent of active tumor for patients with gliomas andimplies that 1-H MRSI is likely to be critical for directing biopsiesor surgical resection, planning focal therapy, and evaluatingtumor burden.

Potential for Tumor GradingGiven the differences in spatial extent of metabolic and an-atomical lesions, the question arises as to whether the met-abolic data are able to contribute to defining tumor type andgrade. In a study that used two-dimensional 1-H MRSI, Preulet al. (34) found an excellent classification of patients using amultivariate pattern recognition analysis of peaks corre-sponding to choline, creatine, N-acetylasparate, lactate,lipid, and alanine. From looking at the metabolite levels ineach class, it was clear that meningiomas were distinguishedas they were the only lesions that had alanine. Grade 2gliomas tended to have low lactate and lipid, some N-acety-laspartate, and some creatine. Grade 3 gliomas tended tohave low lactate and lipid, less N-acetylaspartate and crea-

tine, with higher choline. Grade 4 gliomas tended towardhigh lactate and lipid, with very low N-acetylaspartate. Al-though these results were very promising, there has not yetbeen a prospective study using the statistical classificationthat these authors derived.

One of the complications in analyzing data obtained with amultivoxel data acquisition technique is in determining whichspectrum to consider for each lesion. Suggestions that havebeen made include using the most abnormal voxel and theaverage of all voxels within the lesion. Both of these ap-proaches involve a subjective decision that takes the ana-tomical appearance of the lesion into account. For example,does the lesion include the entire T2 abnormality or is itrestricted to the enhancing volume. As seen in Fig. 5, thespectral characteristics of these regions may be quite differ-ent (72). The same issue is present with single voxel analysis,but in that case, the decision is made implicitly at the time ofdata acquisition by the choice of the selected volume. Ourstudies have suggested that although it may be possible todetect mean differences between populations of gliomaswith different grades based upon metabolite levels, there isconsiderable overlap, both for mean metabolite levels or forthe most abnormal voxels within the T2 lesion (72).4 Becausethere are such large variations in anatomical appearance, it

Fig. 5. Arrays of spectra from grade 2 (left), grade 3 (middle), and grade 4 (right) gliomas. The acquisition parameters were as in Fig. 4. Note theheterogeneity of the spectral patterns in these lesions. Spectra that have metabolic abnormalities are shaded, and those with peaks corresponding to lactateor lipid are marked with a “�”.




seems likely that both anatomical and metabolic patterns arerelevant for classification and that which ever procedure isconsidered should ensure that the influence of both types ofdata are explicitly considered in the analysis. Informationsuch as relative cerebral blood volume or apparent diffusioncoefficient may also help in grading tumors and in distin-guishing between tumors and other types of mass lesions.4

Potential for Evaluating Response to TherapyAnother potentially important clinical role for 1-H MRSI is theability to make an early evaluation of whether a lesion hasresponded to therapy. If this were possible, it would allowtailoring therapy to each individual patient and modifying anineffective treatment strategy before the lesion shows a largeincrease in volume. It would also be possible to avoid givingunnecessary treatment in the case that an increase in en-hancing volume is attributable to formation of treatment-induced necrosis as opposed to recurrent or residual tumor.For 1-H MRSI to be included in the clinical management ofthe patient in this manner, it is important to map out both thetemporal and spatial distribution of metabolite changes inresponse to the therapy of interest. This requires the use ofthree-dimensional 1-H MRSI and is most easily achieved forthe case of focal therapies such as surgery or radiation (74).Registration of the MR images and 1-H MRSI data are criticalfor correlating data from such sequential examinations.

We have already presented some of our results for gammaknife radiosurgery of metastatic lesions. To study this appli-cation for gliomas requires a consideration of whether thetreatment volume covered the metabolic lesion, as well ashow the lesion responded to therapy. It is clear from ouranalysis of data from 36 patients with recurrent gliomas thatthe Gadolinium-enhancing lesion is not an adequate targetfor gamma knife radiosurgery in at least 50% of the recurrentgrade 3 and grade 4 lesions that were considered (48, 49).Patients for whom the metabolic lesion extended outside thegamma knife target had a larger increase in enhancing vol-ume 6 months after therapy, shorter time to additional treat-ment and worse survival than patients with either no meta-bolic lesion or a with a metabolic lesion contained within thetarget (48). Fig. 6 shows an example of the time course ofresponse to gamma knife radiosurgery for a patient who hada relatively small initial lesion that was inside the target. Theanatomical and metabolic lesion shrunk by the first follow-upscan at 2 months after therapy and shrunk additionally by 8months after treatment. The changes in the metabolic lesionwere a reduction in choline and increase in N-acetylaspar-tate. The latter was interpreted as being because of thereturn of normal brain tissue into the voxel as the tumorshrunk.

Fig. 7 shows an example of a patient who had a muchlarger metabolic than anatomical lesion before radiosurgery.

Fig. 6. Sequential changes in spectra for a patient with a recurrent grade 3 glioma, which was treated with gamma knife radiosurgery to the enhancingvolume. In this case, the metabolic lesion was restricted to the treated volume. The examinations were obtained before treatment, 2 and 8 months aftertreatment. Note the reduction in volume of the anatomical lesion and of choline in the treated voxels.




The target addressed only a few abnormal voxels in thecenter of the metabolic lesion. Two months after therapy, thecholine had not changed significantly inside the target, butthere was an increase in a resonance corresponding to lac-tate or lipid. The voxels outside the irradiated region did notchange. Three months after therapy, the choline had de-creased inside the target and the lactate/lipid had increasedadditionally. The Gadolinium-enhancing region increased involume at both time points, and hence, the treatment wasconsidered to have failed. If we had used the metabolitelevels as indicators of response, the treatment would havebeen classified as having some effect within the target re-gion. Similar findings were observed in patients being treatedwith brachytherapy (46). For fractionated radiation therapy,where the treated volume is much larger and the dose dis-tribution more variable, it is necessary to take into accountchanges in the normal appearing brain tissue caused byintermediate doses of radiation, as well as the changeswithin the tumor (74).

ConclusionsFrom the studies that have been performed thus far, it is clearthat 1-H MRSI is an important adjunct to anatomical imagingfor evaluation of tumor type and grade, as well as for target-

ing and evaluating response to therapy. Although the prog-nostic value of this technique is still under investigation, thereis every reason to expect that it will provide information thatwill be relevant for choosing the most appropriate therapy forindividual patients and for understanding the mechanisms ofsuccess and failure of new treatments. This is particularlycritical for screening therapies based upon the biologicalproperties of the tumor, where it is important to knowwhether the lack of response was because of the agent beingunable to access the tumor or to the lesion being insensitiveto that particular approach. Possibilities for improving thesensitivity and specificity of the 1-H MRSI data include theuse of shorter echo times, the application of radiofrequencycoils with improved signal to noise and of magnets withhigher field strength.

AcknowledgmentsI thank Tracy McKnight, Edward Graves, Andrea Prizkall, Xiaojuan Li,Isabelle Catalaa, and Daniel Vigneron for their contributions to the workpresented here.

References1. Prados, M. D. Treatment strategies for patients with recurrent braintumors. Semin. Radiat. Oncol., 1: 62–68, 1991.

Fig. 7. Sequential changes in spectra for a patient with a recurrent grade 4 glioma, which was treated with gamma knife radiosurgery to the enhancingvolume. The examinations were obtained before treatment, 2 and 3 months after treatment. In this case, there were numerous voxels outside the target thathad abnormal metabolism. Note the reduction of choline in the treated voxels but stable or slightly increasing choline outside the target volume. Althoughthe enhancing volume dose increased after therapy, it is still smaller than the metabolic lesion.




2. Leibel, S. A., Scott, C. B., and Loeffler, J. S. Contemporary approachesto the treatment of malignant gliomas with radiation therapy. Semin.Oncol., 21: 198–219, 1994.

3. Fitzek, M. M., Thornton, A. F., Rabinov, J. D., Lev, M. H., Pardo, F. S.,Munzenrider, J. E., Okunieff, P., Bussiere, M., Braun, I., Hochberg, F. H.,et al. Accelerated fractionated proton/photon irradiation to 90 cobalt grayequivalent for glioblastoma multiforme: results of a Phase II prospectivetrial. J. Neurosurg., 91: 251–260, 1999.

4. Nakagawa, K., Aoki, Y., Fujimaki, T., Tago, M., Terahara, A., Karasawa,K., Sakata, K., Sasaki, Y., Matsutani, M., and Akanuma, A. High-doseconformal radiotherapy influenced the pattern of failure but did not im-prove survival in glioblastoma multiforme. Int. J. Radiat. Oncol. Biol. Phys.,40: 1141–1149, 1998.

5. Zimm, S., Wampler, G. L., Stablein, D., Hazra, T., and Young, H. F.Intracerebral metastases in solid tumor patients: natural history and re-sults of treatment. Cancer (Phila.), 48: 384–394, 1981.

6. Flickinger, J. C., Kondziolka, D., Lunsford, L. D., Coffey, R. J., Good-man, M. L., Shaw, E. G., Hudgins, W. R., Weiner, R., Harsh, G. R., 4th,Sneed, P. K., et al. A multi-institutional experience with stereotactic ra-diosurgery for solitary brain metastases. Int. J. Radiat. Oncol. Biol. Phys.,22: 797–802, 1994.

7. Moriarty, T. M., Loeffler, J. S., Black, P. M., Shrieve, D. C., Wen, P. Y.,Fine, H. A., Kooy, H. M., and Alexander, E. Long term follow-up of patientstreated with stereotactic radiosurgery for single or multiple brain metas-tases. In: D. Kondziolka (ed.), Radiosurgery, Vol. 1, pp. 83–91, Basel,Switzerland: Karger, 1995.

8. Russell, D.S., and Rubenstein, L. J. Pathology of Tumors of the Nerv-ous System, Ed. 5, London: Williams and Wilkins, 1989.

9. Kleihues, P., Burger, P. C., and Scheithauer, B. W. Histological typingof tumors of the central nervous system, Ed. 2, Berlin: Springer-Verlag,1993.

10. Dean, B. L., Drayer, B. P., Bird, C. R., Flom, R. A., Hodak, J. A., Coons,S. W., and Carey, R. G. Gliomas: classification with MR imaging. Radiol-ogy, 174: 411–415, 1990.

11. Earnest, F. I. V., Kelly, P. J., Scheithauer, B. W., Kall, B. A., Cascino,T. L., Ehman, R. L., Forbes, G. S., and Axley, P. L. Cerebral astrocytomas:histopathologic correlation of MR and CT contrast enhancement withstereotactic biopsy. Radiology, 166: 823–827, 1988.

12. Janus, T. J., Kim, E. E., Tilbury, R., Bruner, J. M., and Yung, W. K. A.Use of {18F} fluorodeoxyglucose positron emission tomography in pa-tients with primary malignant brain tumors. Ann. Neurol., 33: 540–548,1993.

13. Glantz, M., Hoffman, J. M., Coleman, R. E., Friedman, A. H., Hanson,M. W., Burger, P. C., Herndon, J. E., 2nd, Meisler, W. J., and Schold, S. C.,Jr. Identification of early recurrence of primary central nervous systemtumors by {18F} fluorodeoxyglucose positron emission tomography. Ann.Neurol., 29: 347–355, 1991.

14. Di Chiro, G. Positron emission tomography using F-18-fluorodeoxy-glucose in brain tumors. A powerful diagnostic and prognostic tool. In-vestig. Radiol., 22: 360–371, 1986.

15. 6. Kim, E. E., Chung, S. K., Haynie, T. P., Kim, C. G., Cho, B. J.,Podoloff, D. A., Tilbury, R. S. Yang, D. J., Yung, W. K., Moser, R. P., Jr.,et al. Differentiation of residual or recurrent tumors from post-treatmentchanges with F-18 FDG PET. Radiographics, 12: 269–279, 1992.

16. Aronen, H. J., Gazit, I. E., Louis, D. N., Buchbinder, B. R., Pardo, F. S.,Weisskoff, R. M., Harsh, G. R., Cosgrove, G. R., Halpern, E. F., Hochberg,F. H., et al. Cerebral blood volume maps of gliomas: comparison withtumor grade and histologic findings. Radiology, 191: 41–51, 1994.

17. Pardo, F. S., Aronen, H. J., Kennedy, D., Moulton, G., Paiva, K.,Okunieff, P., Schmidt, E. V., Hochberg, F. H., Harsh, G. R., Fischman,A. J., Linggood, R. M., and Rosen, B. R. Functional cerebral imaging in theevaluation and radiotherapeutic treatment planning of patients with ma-lignant gliomas. Int. J. Rad. Oncol. Biol. Phys., 30: 663–669, 1994.

18. Wenz, F., Rempp, K., Brix, G., Hess, T., Weisser, G., Debus, J.,Knopp, M. V., Engenhart, R., and van Kaick, G. Radiation induced rCBVchanges of low grade astrocytomas and normal brain tissue. InternationalSociety of Magnetic Resonance in Medicine 2nd annual meeting, p. 667,1994.

19. Henry R. G., Vigneron D. B., Fischbein N., Grant P. E., Day, M. R.,Noworolski, S. M., Star-Lack, J. M., Dillon, W. P., Chang, S., and Nelson,S. J. Comparison of proton MRSI and cerebral blood volume imaging ofgliomas, International Society of Magnetic Resonance in Medicine 5thannual meeting. p. 1123, 1997.

20. Brunberg, J. A., Chenevert, T. L., McKeever, P. E., Ross, D. A., Junck,L. R., Muraszko, K. M., Dauser, R., Pipe, J. G., and Betley, A. G. In vivo MRdetermination of water diffusion coefficients and diffusion anisotropy:correlation with structural alteration in gliomas of the cerebral hemi-spheres. AJNR, 16: 361–371, 1995.

21. Castillo, M., Smith, J. K., Kwock, L., and Wilber, K. Apparent diffusioncoefficients in the evaluation of high-grade cerebral gliomas. AJNR, 22:60–64, 2001.

22. Sugahara, T., Korogi, Y., Kochi, M., Ikushima, I., Shigematu, Y., Hirai,T., Okuda, T., Liang, L., Ge, Y., Komohara, Y., et al. Usefulness of diffu-sion-weighted MRI with echo-planar technique in the evaluation of cellu-larity in gliomas. J. Magn. Reson. Imaging, 9: 53–60, 1999.

23. Tien, R. D., Felsberg, G. J., Friedman, H., Brown, M., and MacFall, J.MR imaging of high-grade cerebral gliomas: value of diffusion-weightedecho planar pulse sequences. Am. J. Roentgenol., 162: 671–677, 1994.

24. Damaerel, P., Johannik, K., Van Hecke, P., Van Ongeval, C., Verellen,S., Marchal, G., Wilms, G., Plets, C., Goffin, J., Van Calenbergh, F.,Lammens, M., and Baert, A. L. Localized 1H NMR spectroscopy in fiftynew cases of newly diagnosed intracranial tumors. J. Comput. Assist.Tomogr., 15: 67–76, 1991.

25. Heesters, M. A. A. M., Kamman, R. L., Mooyaart, E. L., and Go, K. G.Localized proton spectroscopy of in operable brain tumors. Response toradiation therapy. J. Neuro-Oncol., 17: 27–35, 1993.

26. Usenius, J. P., Kauppinen, R. A., Vaino, P., Hernesniemi, J. A., Va-palahti, M. P., Paljarvi, L. A., and Soimakallio, S. Quantitative metabolitepatterns of human brain tumors: detection by 1H NMR spectroscopy invivo and in vitro. J. Comput. Assist. Tomogr., 18: 705–713, 1994.

27. Usenius, J-P., Vaino, P., Hernesniemi, J., and Kauppinen, R. A.Choline-containing compounds in human astrocytomas studied by 1HNMR spectroscopy in vivo and in vitro. J. Neurochem., 63: 1538–1543,1994.

28. Chang, L., Mc Bride, D., Miller, B. L., Cornford, M., Booth, R. A.,Buchthal, S. D., Ernst, T. M., and Jenden, D. Localized in vivo 1H magneticresonance spectroscopy and in vitro analyses of heterogeneous braintumors. J. Neuroimaging, 5: 157–163, 1995.

29. McBride, D. Q., Miller, B. L., Nikas, D. L., Buchthal, S., Chang, L.,Chiang, F., and Booth, R. A. Analysis of brain tumors using 1H magneticresonance spectroscopy. Surg. Neurol., 44: 137–144, 1995.

30. Negendank, W. G., Sauter, R., Brown, T. R., Evelhoch, J. L., Falini, A.,Gotsis, E. D., Heerschap, A., Kamada, K., Lee, B. C., and Mengeot, M. M.Proton magnetic resonance spectroscopy in patients with glial tumors: amulticenter trial. J. Neurosurg., 84: 449–458, 1996.

31. Sijens, P. E., Kopp, M. V., Brunetti, A., Wicklow, K., Alfano, B.,Bachert, P., Sanders, J. A., Stillman, A. E., Kett, H., Sauter, R., et al. 1HMR spectroscopy in patients with metastatic brain tumors: a multicenterstudy. Magn. Reson. Med., 33: 818–826, 1995.

32. Shimizu, H., Kumabe, T., Tominaga, T., Kayama, T., Hara, K., Ono, Y.,Sato, K., Arai, N., Fujiwara, S., and Yoshimoto, T. Noninvasive evaluationof malignancy of brain tumors with proton MR spectroscopy. AJNR Am. J.Neuroradiol., 17: 737–747, 1996.

33. Somorjai, R. L., Dolenko, B., Nikulin, A. K., Pizzi, N., Scarth, G.,Zhilkin, P., Halliday, W., Fewer, D., Hill, N., Ross, I., West, M., Smith,I. C. P., Donnelly, S. M., Kuesel, A. C., and Briere, K. M. Classification of1H MR spectra of human brain neoplasms: the influence of preprocessingand computerized consensus diagnosis on classification accuracy. J.Magn. Reson. Imaging, 6: 437–444, 1996.

34. Preul, M. C., Caramanos, Z., Collins, D. L., Villemure, J. G., Leblanc,R., Olivier, A., Pokrupa, R., and Arnold, D. L. Accurate non-invasivediagnosis of human brain tumors by using proton magnetic resonancespectroscopy. Nat. Med., 2: 323–325, 1996.

35. Vogl, T. J., Friebe, C. E., Balzer, T. Mack, M. G., Steiner, S. Schedel,H. Pegios, W. Lanksch, W. Banzer, D., and Felix, R. Diagnosis of cerebralmetastasis with standard dose gadobutrol versus a high dose protocol.




Intraindividual evaluation of a Phase II high dose study. Radiologe, 35:508–516, 1995.

36. Yuh, W. T., Nguyen, H. D. Tali, E. T., Mayr, N. A., Fisher, D. J., Atlas,S. W., Carvlin, M. C., Drayer, B. P., Pollei, S. R., Runge, V. M., et al.Delineation of gliomas with various doses of MR contrast material, Am. J.Neuroradiol., 15: 983–989, 1994.

37. Kurki, T., Lundbom, N., Kalimo, H., and Valtonen, S. MR classificationof brain gliomas: value of magnetization transfer and conventional imag-ing. Magn. Reson. Imaging, 13: 501–511, 1995.

38. Rydberg, J. N., Hammond, C. A., Grimm, R. C., Erickson, B. J., Jack,C. R., Jr., Huston, J., III, and Riederer, S. J. Initial clinical experience in MRimaging of the brain with a fast fluid-attenuated inversion-recovery pulsesequence. Radiology, 193: 173–180, 1994.

39. De Coene, B., Hajnal, J. V., Gatehouse, P. Longmore, D. B., White,S. J., Oatridge, A., Pennock, J. M., Young, I. R., and Bydder, G. M. MR ofthe brain using fluid-attenuated inversion recovery (FLAIR) pulse se-quences. Am. J. Neuroradiol., 13: 1555–1564, 1992.

40. Alger, J. R., Frank, J. A., Bizzi, A., Fulham, M. J., DeSouza, B. X.,Duhaney, M. O., Inscoe, S. W., Black, J. L., van Zijl, P. C. M., Moonen,C. T. W., and Di Chiro, G. Metabolism of human gliomas: assessment withH-1 MR spectroscopy and F-18 flourodeoxyglucose PET. Radiology, 177:633–641, 1990.

41. Gill, S. S., Thomas, D. G., Van Bruggen, N., Gadian, D. G., Peden,C. J., Bell, J. D., Cox, I. J., Menon, D. K., Iles, R. A., and Bryant, D. J.Proton MR spectroscopy of intracranial tumors: in vivo and in vitro studies.J. Comput. Assist. Tomogr., 14: 497–504, 1990.

42. Barker, P. B., Glickson, J. D., and Bryan, R. N. In vivo magnetic reso-nance of human brain tumors. Top. Magn. Reson. Imaging, 5: 32–45, 1993.

43. Segebarth, C. M., Baleriaux, D. F., Luyten, P. R., and den Hollander, J. A.Detection of metabolic heterogeneity of human intracranial tumors in vivo byH-1 NMR spectroscopic imaging. Magn. Reson. Med., 13: 62–76, 1990.

44. Luyten, P. R., Marien, A. J., Heindel, W., van Gerwen, P. H., Herholz,K., den Hollander, J., Friedmann, G., and Heiss, W. D. Metabolic imagingof patients with intracranial tumors: H-1 MR spectroscopic imaging andPET. Radiology, 176: 791–799, 1990.

45. Fulham, M. J., Bizzi, A., Dietz, M. J., Shih, H. H., Raman, R., Sobering,G. S., Frank, J. A., Dwyer, A. J., Alger, J. R., and Di Chiro G. Mapping ofbrain tumor metabolites with proton MR spectroscopic imaging: clinicalrelevance. Radiology, 185: 675–686, 1992.

46. Wald, A. A., Day, M. R., Nelson, S. J., Moyher, S. E., Henry, R. G.,Sneed, P. K., Huhn, S. L., Chang, S., Prados, M. D., Dillon, W. P., Gutin,P. H., McDermott, M., and Vigneron, D. B. Response of glioblastomamultiforme to brachytherapy detected by 3D proton magnetic resonancespectroscopic imaging. J. Neurosurg., 87: 525–534, 1997.

47. Nelson, S. J., Vigneron, D. B., and Dillon, W. P. Serial evaluation ofpatients with brain tumors using volume MRI and 3D 1H MRSI. NMRBiomed., 12: 123–138, 1999.

48. Graves, E. E., Nelson, S. J., Vigneron, D. B., Chin, C., Verhey, L.,McDermott, M., Larson, D., Sneed, P. K., Chang, S., Prados, M., Lamborn,K., and Dillon, W. P. A preliminary study of the prognostic value of1H-spectroscopy in gamma knife radiosurgery of recurrent malignantgliomas. Neurosurgery (Baltimore), 46: 319–328, 2000.

49. Graves, E. E., Nelson, S. J., Vigneron, D. B., Verhey, L. McDermott,M., Larson, D., Chang, S., Prados, M. D., and Dillon, W. P. Serial protonMR spectroscopic imaging of recurrent malignant gliomas after gammaknife radiosurgery. AJNR, 2: 613–624, 2001.

50. Dowling, C., Bollen, A. W., Noworolski, S. M., McDermott, M. W.,Barbaro, N. M., Day, M. R., Henry, R. G., Chang, S. M., Dillon, W. P.,Nelson, S. J., and Vigneron, D. B. Preoperative proton MR spectroscopyin brain tumor patients with a mass lesion: correlation with resectionspecimen histology. AJNR, 22: 604–612, 2001.

51. Vigneron, D., Bollen, A., McDermott, M., Wald, L., Day, M., Moyher-Noworolski, S., Henry, R., Chang, S., Berger, M., Dillon, W., and Nelson,S. Three-dimensional magnetic resonance spectroscopic imaging of his-tologically confirmed brain tumors. MRI, 19: 89–101, 2001.

52. Nelson, S. J., Vigneron, D. B., Star-Lack, J., and Kurhanewicz, J. Highspatial resolution and speed in MRSI. NMR Biomed., 10: 411–422, 1997.

53. Star-Lack, J., Nelson, S. J., Kurhanewicz, J., Huang, L. R., andVigneron, D. B. Improved water and lipid suppression for 3-D press CSI

using RF band selective inversion with gradient dephasing (BASING).Magn. Reson. Med., 38: 311–321, 1997.

54. Star-Lack, J., Vigneron, D. B., Pauly, J., Kurhanewicz, J., and Nelson,S. J. Improved solvent suppression and increased spatial excitation band-widths for 3-D PRESS CSI using phase-compensating spectral/spatialspin-echo pulses. J. Magn. Reson. Imaging, 7: 745–757, 1997.

55. Tran, T-K. C., Vigneron, D. B., Sailasuta, N., Tropp, J., Le Roux, P.,Kurhanewicz, J., Nelson, S. J., and Hurd, R. Very selective suppressionpulses for clinical MRSI studies of brain and prostate cancer. Magn.Reson. Med., 43: 23–33, 2000.

56. Spielman, D. M., Pauly, J. M., Macovski, A., Glover, G. H., andEnzmann, D. R. Lipid-suppressed single- and multisection proton spec-troscopic imaging of the human brain. J. Magn. Reson. Imaging, 2: 253–262, 1992.

57. Duyn, J. H., Gillen, J., Sobering, G., van Zijl, P. C., and Moonen, C. T.Multisection proton MR spectroscopic imaging of the brain. Radiology,188: 277–282, 1993.

58. Duyn, J., and Moonen C. T. Fast proton spectroscopic imaging of humanbrain using multiple spin-echoes. Magn. Res. Med., 30: 409–414, 1993.

59. Adelsteinsson, E., Irarrazabal, P., Spielman, D. M., and Macovski A.,Three-dimensional Spectroscopic Imaging with Time-Varying Gradients.Mag. Res. Med., 33: 461–466, 1995.

60. Posse, S., Tedeschi, G., Risinger, R., Ogg, R., and Le Bihan, D. HighSpeed 1H spectroscopic imaging in human brain by echo planar spatial-spectral encoding, Magn. Res. Med., 33: 34–40, 1995.

61. Posse, S., DeCharli, C., and Bihan, D. L. Three-dimensional echo-planar MR spectroscopic imaging at short echo times in the human brain.Radiology 192: 733–738, 1994.

62. Nelson, S. J. The analysis of volume MRI and MR spectroscopicimaging data for the evaluation of patients with brain tumors. Magn.Reson. Med., 46: 228–239, 2001.

63. Spielman, D., Webb, P., and Macovski A., A Statistical Framework forin vivo spectroscopic imaging. J. Magn. Reson., 79: 66–77, 1988.

64. Spielman, D., Webb, P., and Macovski, A. Water referencing forspectroscopic imaging. Magn. Reson. Med., 12: 38–49, 1989.

65. Van der Veen, J. W. C., de Beer, R., Luyten, P. R., and van Ormondt, D.,Accurate quantification of in vivo 31P nmr signals using the variable projectionmethod and prior knowledge. Magn. Res. Med., 6: 92–98, 1988.

66. Derby, K., Hawryszko, H., and Tropp J. Baseline deconvolution,phase correction and signal quantification in Fourier localized spectro-scopic imaging. Magn. Res. Med., 12: 235–240, 1989.

67. Nelson, S. J., and Brown, T. R. A new method for automatic quanti-fication of 1-D Spectra with low signal to noise ratio. J. Magn. Reson., 75:229–243, 1987.

68. Nelson, S. J., and Brown, T. R. A study of the accuracy of quantifi-cation which can be obtained from 1-D nmr spectra using the PIQABLEalgorithm. J. Mag. Reson., 84: 95–109, 1989.

69. Provencher, S. W. Estimation of metabolite concentrations from local-ized in vivo proton NMR spectra. Magn. Reson. Med., 30: 672–679, 1993.

70. McKnight, T. R., Noworolski, S. M., Vigneron, D. B., and Nelson, S. J.An automated technique for the quantitative assessment of 3D-MRSI datafrom patients with glioma. J. Mag. Reson. Imaging, 13: 167–177, 2001.

71. Pirzkall, A., McKnight, T. R., Graves, E. E., Carol, M. P., Sneed, P. K.,Wara, W. W., Nelson, S. J., Verhey, L. J., and Larson, D. A. MR-spec-troscopy guided target delineation for high-grade gliomas. Int. J. Radiat.Oncol. Biol. Phys., 50: 915–928, 2001.

72. Li, X., Lu, Y., Pirzkall, A., and Nelson, S. J. Analysis of spatial extentof the metabolic abnormality for newly diagnosed glioma patients. J.Magn. Reson. Imaging, 16: 229–237, 2002.

73. McKnight, T. R., von dem Bussche, M. H., Vigneron, D. B., Lu, Y.,Berger, M. S., McDermott, M. W., Dillon, W. P., Graves, E. E., Pirzkall, A.,and Nelson, S. J. Histopathological validation of a three-dimensionalmagnetic resonance spectroscopy index as a predictor of tumor pres-ence. J. Neurosurg., 97: 794–802, 2002.

74. Graves, E. E., Pirzkall, A., Nelson, S. J., Verhey, L., and Larson, D.Registration of magnetic resonance spectroscopic imaging to computedtomography for radiotherapy treatment planning. Med. Phys., 28: 2489–2496, 2001.




2003;2:497-507. Mol Cancer Ther Sarah J. Nelson

1TumorsMultivoxel Magnetic Resonance Spectroscopy of Brain

Updated version

http://mct.aacrjournals.org/content/2/5/497

Access the most recent version of this article at:

Cited articles

http://mct.aacrjournals.org/content/2/5/497.full#ref-list-1

This article cites 68 articles, 6 of which you can access for free at:

Citing articles

http://mct.aacrjournals.org/content/2/5/497.full#related-urls

This article has been cited by 14 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

SubscriptionsReprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://mct.aacrjournals.org/content/2/5/497To request permission to re-use all or part of this article, use this link



http://mct.aacrjournals.org/content/2/5/497.full#ref-list-1

http://mct.aacrjournals.org/content/2/5/497.full#related-urls

http://mct.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

Minireview - Home | Molecular Cancer Therapeutics · lesions within the brain influences the decision as to whether the most appropriate treatment is surgery, radiosurgery, or whole