Multichannel CT Imaging of Orthopedic Hardware and Implants · Multichannel CT Imaging of Orthopedic Hardware and ... metal hardware cause more attenuation and more arti- ... MULTICHANNEL

Multichannel CT Imaging of OrthopedicHardware and ImplantsKenneth A. Buckwalter, M.D.,1 J. Andrew Parr, M.D.,2 Robert H. Choplin, M.D.,1

and William N. Capello, M.D.2

ABSTRACT

The introduction of multichannel CT scanners provides both radiologists andsurgeons with a new tool to image patients with orthopedic hardware. The key parametersthat have made it possible to image the implants and the surrounding bone withmultichannel CT are the higher available technical factors (kVp and mAs) coupled withthe ability to acquire thin slices over a large scan region. These properties make it possibleto produce high-quality multiplanar reformations that facilitate visualization of theorthopedic device and the surrounding bone. An important consideration for multichannelCT imaging of hardware is the reduction of cone beam artifacts caused by the geometry ofmultichannel CT scanners. This artifact is reduced by using a narrower x-ray beamcollimation and a low pitch setting. This article discusses CT scan parameters and imagepostprocessing used at our institution and illustrates common clinical problems encoun-tered when imaging implanted orthopedic devices. These include fracture healing, loosen-ing of joint prostheses, evaluation of particle disease, and the use of CT for preoperativeplanning in revision arthroplasty.

KEYWORDS: X-ray computed tomography, CT, bone, joint prosthesis, prostheses and

implants

Advances in imaging technology have resulted inthe creation of multichannel computed tomography(CT) scanners. These powerful devices have revolution-ized body imaging and provide new opportunities inmusculoskeletal imaging. Techniques optimized forbody imaging are not optimal for orthopedic imaging.It is important to be aware of the differences in scantechniques to produce high-quality images of bonestructures. Metal hardware introduces additional arti-facts and accentuates the need to minimize the artifactsintroduced by the geometry of multichannel machines.

Imaging of patients with metal hardware is im-portant in assessing fracture fixation and fracture healing.The bone around most hardware can be imaged success-fully. The artifacts introduced by the hardware are related

to the composition of the metal, the shape and orienta-tion of the hardware, and the scanner setup.1 The mostimportant scan parameter affecting image quality ispitch: low pitch settings should be used routinely whenscanning patients with metal hardware. Other parame-ters such as kVp, mAs, and image reconstruction algo-rithm are important, too, and are discussed.

The principles used to image metal plates andintramedullary nails can be applied to imaging of jointimplants. Failure of joint implants may have a variety ofcauses including mechanical failure or fracture of themetal components, fracture of the surrounding bone,infection, component wear, osteolysis, and particledisease. Most of these complications can be detectedwith CT images, particularly if high-quality multiplanar

An Update on Imaging of Joint Reconstructions; Editors in Chief, David Karasick, M.D., Mark E. Schweitzer, M.D.; Guest Editor, Theodore T.Miller, M.D. Seminars in Musculoskeletal Radiology, Volume 10, Number 1, 2006. Address for correspondence and reprint requests: Kenneth A.Buckwalter, M.D., Indiana University, Department of Radiology, Room 0615E, 550N. University Blvd., Indianapolis, IN 46202. 1Department ofRadiology, 2Department of Orthopedics, Indiana University, Indianapolis, Indiana. Copyright # 2006 by Thieme Medical Publishers, Inc., 333Seventh Avenue, New York, NY 10001, USA. Tel: +1(212) 584-4662. 1089-7860,p;2006,10,01,086,097,ftx,en;smr00388x.

86

Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

reformations (MPR) are available for review.2 Becausemany of these patients can undergo revision arthroplasty,preoperative imaging is extremely helpful when radiog-raphy fails to demonstrate the relevant anatomy ad-equately. The techniques used to image implants withCT and illustrations of some pertinent complications aredetailed in this article.

CT TECHNIQUEAn understanding of the CT scan parameters that affectimage quality is crucial to the production of diagnosticimages of bone structures. This is particularly relevantwhen imaging patients with implants and other hard-ware. Techniques optimized for body imaging do notproduce optimal results when applied to imaging ofpatients with orthopedic devices. Body imaging typicallyis performed with intravenous contrast material duringsuspension of respiration requiring short (< 20 second)scan times. These rapid scan times are achieved using thehighest x-ray tube rotation times and the highest tablefeed velocities resulting in higher pitch settings. Thesescan parameters degrade visualization of detail by exag-gerating cone beam artifacts, which are more conspic-uous when metal hardware is present. Thus, standardbody imaging protocols should be avoided when imagingpatients with orthopedic hardware.

The factors affecting our ability to visualize thebone structures surrounding implanted hardware includethe following: hardware composition, hardware geome-try and location, pitch, kVp, mAs, and image recon-

struction algorithm. Each of these factors is discussed inthe following paragraphs.

Metal hardware attenuates the x-ray beam andalters the spectral characteristics of the radiation causingbeam hardening. There are three major types of metalalloys used in orthopedic devices: cobalt-chrome based,iron based, and titanium based. Tantalum is anothermaterial used occasionally for some prosthesis compo-nents. Cobalt-chrome alloys and stainless steel (ironbased) attenuate the x-ray beam significantly, resultingin major artifacts. Conversely, titanium and titaniumalloys produce minimal artifacts. Cobalt-chrome alloysare encountered commonly in hip and knee prostheses.Stainless steel is used often for fracture plates and screws.Titanium is used in the stems of newer hip prosthesisand for some intramedullary nails. Because it is imprac-tical to identify the specific metal alloy used in animplant prior to the CT examination, evaluation of theopacity of the metal on the CT scout image is a helpfulindicator of the density of the metal; dense cobalt-chrome implants are white on the scout image, whereasthe less dense titanium devices are gray to gray-white onthe scout image. These are guidelines that can be used toadjust technical factors at the time of the examination.

As might be expected, thicker portions of themetal hardware cause more attenuation and more arti-facts than thinner portions of the hardware. Manyintramedullary nails and prosthesis stems are symmetricin cross section and the attenuation of the x-ray beam isrelatively uniform, dispersing the artifacts across theentire image (Fig. 1). In contrast, fracture plates are

Figure 1 (A) Axial CT image through midtibia. Artifacts from intramedullary nail are dispersed relatively uniformly throughout image.(B) Coronal reformation through posterior tibia of the same patient shows persistent fracture line. Intramedullary nail used to stabilizethe fracture causes no visible artifacts. (C) Sagittal reformation, same patient. Healing is incomplete as there is no bone bridgingat the fracture site.

MULTICHANNEL CT IMAGING OF ORTHOPEDIC HARDWARE AND IMPLANTS/BUCKWALTER ET AL 87

Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

rectangular in cross section and artifacts usually are mostsevere in the direction of the greatest cross section(Fig. 2A). Because fracture plates are placed along theouter margin of the affected bone, one can predict thatthe most severe artifacts will propagate tangentially tothe surface of the bone and that the adjacent bone will bevisualized adequately (Fig. 2B). Thus, a successful ex-amination can be anticipated when lateral fixation platesare present.

Because artifacts are most severe immediatelyadjacent to the hardware, the farther the hardware islocated from the region of interest, the less it interfereswith visualization. For example, a patient with a unilat-eral hip prosthesis can undergo a detailed examination ofthe contralateral hip with little or no artifact. Anexception to the proximity rule is the patient withbilateral hip prostheses. In this circumstance, the attenu-ation from both prostheses coupled with the overlyinglateral soft tissues can produce severe artifacts. Except forthe smallest of patients, the density of the two prosthesesoverwhelms the ability of the x-ray beam to penetrate thetissues in the horizontal direction, resulting in left-to-right streaking.

Metal produces artifacts only when the x-raybeam traverses the metal object; consequently, it maybe possible to reorient the body part containing metal toreduce or avoid the x-ray beam entirely. For example,patients with bilateral knee prostheses may be positionedso that a cushion or pillow is placed behind the un-affected knee, displacing this knee so that the densest

portions of the prostheses are not simultaneously trav-ersed by the x-ray beam, improving visualization of theaffected knee, which remains straight (Fig. 3A).

Perhaps the most important parameter affectingimage quality is pitch. One definition of pitch is the tableincrement divided by the detector element width foreach complete revolution of the x-ray beam. Thus, highpitch settings represent rapid table motion relative to x-ray beam revolution, quickly covering a large portion ofthe patient’s anatomy. Rapid scanning is useful to followa contrast bolus or to facilitate motion-free imagingduring suspended respiration, but the higher pitch set-tings exaggerate image artifacts. These artifacts areworse for hard edges such as the bone–soft tissue inter-face and are even worse when metal is present. Despitethe deployment of cone beam correction software, theartifacts cannot be avoided. The cause of these artifacts isrelated partly to the geometry of multichannel scanners,which introduces cone beam effects.

The cone beam artifacts become worse as thedetector elements become more numerous. Thus, conebeam artifacts are minimal on a 4-channel machine butcan become severe on 16- and higher channel devices,which have wider detector arrays. The remedy is two-fold. First, the use of lower pitch settings can reducedramatically the cone beam artifacts and improve overallimage quality. For four-channel scanners, pitch settingsof 0.6 to 0.9 are recommended. If pitch cannot bechanged, alter the table feed speed; slower table feedscorrespond to lower pitch settings. For 16-channel and

Figure 2 (A) Axial CT through midtibia. Artifacts from lateral side plate project in the direction of densest metal (arrows). (B) Coronalreformation of axial series shown in A. Fractures of distal femur and proximal tibia were stabilized with lateral side plates. Both fracturesites have healed as there is significant bone bridging and remodeling. Presence of side plate does not interfere with bone visualization.Cortical screws used to secure the plate cause additional artifacts in medial to lateral direction; however, most screws are placedproximal and distal to the fracture zone and do not interfere with fracture visualization.

88 SEMINARS IN MUSCULOSKELETAL RADIOLOGY/VOLUME 10, NUMBER 1 2006

Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

higher channel scanners, a pitch setting of 0.3 works wellif your equipment permits; otherwise, use the lowestpossible setting. The second remedy is the use of thenarrower detector element collimation. For 16-channelmachines, the choices of array elements are 16� 0.75(12 mm collimation) or 16� 1.5 (24 mm collimation)for Philips and Siemens CT scanners and 16� 0.625(10 mm collimation) or 16� 1.25 (20 mm collimation)for GE scanners. Some investigation into protocol setupmay be required to select the thinner collimation setting.Both the thinner collimation setting and the lower pitchsetting are necessary for optimal results.

Technical factors such as the use of a higher kVpsetting and higher mAs may be helpful in addressingsome of the metal artifacts.3 The use of the filtered backprojection algorithm in CT image reconstruction isresponsible for the streak artifacts encountered in images

containing metal hardware. The metal attenuation of thex-ray beam may be severe, resulting in imperfect CTprojections. For less dense metal, the use of a higherkVp setting (e.g., 140 kVp) may result in more photonspassing through the hardware, improving the quality ofthe image reconstructions. Similarly, higher mAs factorscan be employed. Both of these strategies are recom-mended when dense hardware is present. The downsideof this approach is an increase in dose to the patient. Infact, if all other parameters remain unchanged (i.e.,pitch, slice width, mAs), the patient experiences 30 to40% higher dose when the kVp is raised from 120 to 140.

If the region of interest is remote from radiation-sensitive tissues, such as the knee or ankle joint, in-creased dose should not be a serious issue. Similarly, ifthe region of interest is the wrist and can be positionedoverhead, the unnecessary radiation of other body parts

Figure 3 (A) CT scout view through the knees of a 69-year-oldwoman with long-stemmed knee prosthesis and knee pain. Un-affected leg is bent to prevent overlap of both femoral compo-nents. (B) Sagittal reformation of long-stemmed hinged total kneeprosthesis, same patient. Small size of patella suggests priorpatellar resurfacing component, removed during revision arthro-plasty when the long-stemmed components were placed. Largeradiolucent zone surrounds stem of femoral component, indicatingloosening and failure of this component. Periosteal reaction sur-rounds femoral shaft. Proximal cortical break through is consistentwith bone destruction secondary to infection. Proximal location ofthis finding underscores the need to image the full extent ofhardware, not just symptomatic region. (C) Coronal reformationof femoral portion of knee prosthesis. Periosteal reaction isseen to better advantage than in B. Transverse fracture throughdistal femoral metaphysis is another finding of component failure.(D) Axial image through proximal femur shows cortical breakthrough and focal fluid collection (arrows) consistent with abscess.


Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

can be avoided. To obtain a diagnostic examination in anolder patient past childbearing years, the use of highertechnical factors should be considered an acceptable risk.If at all possible, minimal exposures are advised whendealing with pediatric and young adult patients.

Although it may seem paradoxical when imagingbone, the use of a smoother image reconstruction algo-rithm such as a standard or soft tissue filter rather than asharper reconstruction algorithm such as a bone filtermay assist in reducing some of the fine streak artifactsseen in the soft tissues surrounding metal hardware. Thisstrategy works well when dense metal, such as a hipprosthesis, is present. The trade-off is a reduction in thespatial resolution of the surrounding bone. For largedevices such as a hip prosthesis, a typical field of view(FOV) used to reconstruct the image would be 20 cm;with a 512 image matrix, pixel size is about 0.4 mm.Using liberal approximations, this corresponds veryroughly to a 13 lp/cm spatial resolution, which is closeto the resolution limits of a standard CT reconstructionfilter. Thus, spatial resolution is not compromised sig-nificantly when using the smoother reconstruction filter.In addition, the absence of significant streak artifact mayreveal details that would otherwise be obscured whenusing a sharper reconstruction filter.

When limited hardware such as screws and pins ispresent in small joints like the ankle and wrist, imagesare reconstructed using submillimeter slice width at 50%overlap (e.g., 0.8 mm thick at 0.4-mm increments). Asharp reconstruction filter is employed because metalartifacts are usually minimal. For more substantial hard-ware such as a fixation plate, hip, or knee prosthesis,images are reconstructed using 1- to 1.5-mm-thick slicesat 50% overlap. Depending on the density of the metaland the location or amount of surrounding soft tissues, astandard or soft tissue algorithm should be used.

IMAGE POSTPROCESSINGFor most orthopedic imaging, frontal and lateral projec-tion radiographs are the conventional way to reviewjoints and hardware implants. These projections corre-spond to sagittal and coronal multiplanar reformationsgenerated from the source axial CT images. Thus,interpretation of CT studies of patients with orthopedicimplants is dependent on the availability of high-qualitymultiplanar reformations and other forms of imagepostprocessing.

The development of highly integrated powerfulpostprocessing software and workstations makes itpossible to generate multiplanar reformations rapidlyand easily. Sagittal, coronal, and true axial (if neces-sary) planes are created routinely. For each plane, a setof 24 to 36 images is generated; more are necessarywhen evaluating a large region such as a postoperativeexamination of an extensive pelvic fracture. When

reviewing images with dense hardware, it is beneficialto increase the thickness of the multiplanar reforma-tions.2 The reformations are created using 1- to2-mm-thick slices. Although thicker reformations in-crease the likelihood of introducing partial volumeartifacts and image blurring, these undesirable effectsare offset by a significant reduction in image streakinginduced by the averaging effects of the thicker slice. Ingeneral, the default thickness for multiplanar reforma-tions is set by the workstation software to the in-planepixel size, so the slice thickness of the reformationsmust be increased manually.

Sometimes it is valuable to demonstrate the globalrelationships of hardware and bone. If radiographs arenot available, this relationship is best illustrated using avolume rendering technique. Volume rendering allowsthe simultaneous display of different density objects suchas metal (very dense) and bone (less dense). If thethreshold for bone is lower than that of metal, the metalhardware can be visualized through the less dense bone.The volume rendering technique is relatively powerfuland can be used to assign different colors to the bone andthe metal, amplifying the differences between the ob-jects. In addition, volume rendering techniques are morerobust than shaded surface display (three-dimensional)techniques, valuable when moderately severe streak ar-tifacts are present.

ORTHOPEDIC HARDWAREOne of the most common indications for hardwareplacement is fracture fixation. Long bone fractures maybe spanned by intramedullary nails or lateral side plates(Figs. 1, 2). When fractures occur in irregularly shapedbones such as the ilium or ischium, malleable plates maybe used to secure fixation (Fig. 4). It is important toconsider imaging the full extent of the implanted hard-ware because the hardware alters the mechanical proper-ties of the bone and complications such as loosening maybegin proximal or distal to the original fracture site(Fig. 3). Also, it is advisable to include the nearest jointwithin the scanned volume to provide anatomic cuesabout general orientation on subsequent multiplanarreformations and to demonstrate the relationship of aknown landmark to the fracture site. This is particularlyimportant if revision surgery is under consideration.

When imaging occurs soon after surgical repair,the goal of postoperative fracture imaging is the assess-ment of joint alignment or the assessment of the prox-imity of hardware to sensitive soft tissue structures suchas nerves or articular surfaces. When imaging occursweeks to months after surgical repair, the goal is usuallythe evaluation of fracture healing by assessing the pres-ence of bone bridging (Figs. 1, 2, 5). Because immatureosteoid is less dense than ordinary bone, very earlyhealing may not be evident at imaging. In addition,


Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

the determination of delayed union, malunion or non-union is a complex issue requiring a comprehensiveassessment of the patient’s clinical state. Furthermore,the terms malunion and nonunion have negative con-notations and should probably be avoided in the radio-graphic report because some of these patients pursuelitigation. It is much better to focus on the presence orabsence of bone bridging in the report and avoid thenegative terminology altogether.

HIP IMPLANTSImplantation of a hip prosthesis is an effective way totreat patients with debilitating joint disease. A properlyimplanted prosthesis can have a life span of decades.However, some prostheses fail and need to be removedor revised. There are a variety of reasons for failureincluding mechanical failure of the prosthesis compo-

nents, mechanical loosening, stress fracture, particledisease, and infection. The reason for failure is notalways clear. Most patients can undergo revision arthro-plasty, and these patients may benefit from preoperativeimaging to assist in surgical planning.

Some knowledge about the hip components isuseful when interpreting CT studies of patients with hipprostheses. Placement of a hip prosthesis can involve asimple femoral arthroplasty with implantation of a uni-polar or bipolar femoral component or a complete hipreplacement that includes an acetabular implant as well.The femoral stem can be cemented in place or may be aningrowth component. In the latter case, the stem is oftencoated with tiny metal spheres sintered to the metalstem. Because of their small size (50–200 mm), thesespheres are difficult to appreciate when they are in theirnormal location adjacent to the metal, but micromotionand wear may cause the spheres to dislodge, referred to

Figure 4 (A) Anteroposterior (AP) radiograph ofpelvis shows malleable plates and screws appliedto the posterior acetabular column for fracture repair.Patient presents for arthroplasty and placementof total hip prosthesis. Preoperative CT performedto evaluate available bone stock. (B) Coronal refor-mation of same patient shows bone remodeling ofthe lateral acetabulum. Inferior margin of malleableplate contacts superior femoral head resulting indestruction of the head. (C) Sagittal reformation,anterior to the left, same patient. Presence of malle-able plates causes minor artifact. Considerable pos-terior bone stock remains, ensuring an adequatesubstrate for reconstructive surgery.


Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

as ‘‘bead shedding.’’ Aggregate clumps of beads can bevisible in the pseudocapsule surrounding the implant ashigh-density irregularly shaped material. This is onehelpful finding when searching for signs of prosthesisfailure on CT examinations.

In the case of a unipolar hip prosthesis, the stemand femoral head are usually one monolithic unit. For avariety of reasons, this type of implant has fallen out offavor. The composition of a bipolar hip prosthesis ismore complicated, consisting of a femoral stem, a head, aplastic liner, and a hollow hemisphere that accommo-dates the liner and the head of the femoral component.The outer surface of the hemisphere articulates with thecartilage of the native acetabulum. These modular com-ponents are designed to enable future revision arthro-plasty to a total hip with the addition of an acetabularcup; the femoral stem is preserved, simplifying therevision.

A modern total hip prosthesis consists of a fem-oral stem, a head, a plastic liner, and a metal acetabularcup (Fig. 6). The plastic liner is made of high-densitypolyethylene and may be held in place by a circum-ferential clip placed near the rim of the metal cup. Mostof the acetabular cups we encounter have an ingrowthcoating and are secured to the native acetabulum withone or two screws. Another design, the gap cup pros-thesis, has a U-shaped flange that fits under the acetab-ular tear drop and one or two tabs that are screwed to thelateral margins of the iliac wing. Occasionally, a simpleplastic cup is cemented in place to the native acetabulum.

The polyethylene liner of the acetabular cup wearsover time, and the head of the femoral component canmigrate superiorly. Early evidence of this finding is best

appreciated on radiographs (Fig. 7A). More significantwear can be detected on sagittal and coronal reforma-tions of the CT examination, and wear is easy to detectbecause the head is normally centered symmetrically inthe acetabular cup (Fig. 6). In some patients, the poly-ethylene debris incites a granulomatous reaction in theadjacent bone. This ‘‘particle disease’’ can result inerosion of bone resulting in cyst-like changes in theperiprosthetic bone, weakening the bone and contribu-ting to failure of the prosthesis (Fig. 7). The erosions canoccur around the acetabular cup as well as the proximalfemur.4 In addition, patients can develop pseudomem-branes in the soft tissues surrounding the pseudocapsuleof the prosthesis.5,6 These pseudomembranes are cyst-like regions of fluid or soft tissue density material thatprotrude into the periacetabular region and communi-cate with the hip joint (Fig. 7F). The cysts can extendsuperiorly for many centimeters, and it is important toextend the scan to the superior pelvic brim routinely toidentify the full extent of the pseudomembrane forma-tion. There remains some controversy about the etiologyof the periprosthetic erosions and pseudomembraneformation: particle disease is not the sole cause.

In an effort to address the polyethylene wearproblem, some manufacturers of orthopedic implantsproduce ceramic femoral heads designed to workin conjunction with ceramic acetabular liners. On CTimages, the ceramic composite is homogeneous in ap-pearance and relatively high in density, although lessdense than metal (Fig. 8). The acetabular liners may havea metal backing and are secured to a conventional metalacetabular cup. The ceramic components are preciselymachined and the interface between components should

Figure 5 (A) AP radiograph of an 18-year-old male following fusion of the left hip and placement of a cobra plate. Implanted bonestimulator is overlapped partially by the proximal portion of the plate. One of two proximally broken screws is visible (arrow). (B) Coronalreformation through the fusion shows bone graft material in the arthrodesis site. There is no bone bridging.


Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

Figure 6 (A) Coronal reformation of anuncomplicated right total hip prosthesis;only right side was replaced in this pa-tient. The acetabular component has aningrowth coating and the femoral com-ponent is cemented in place. Full extentof femoral stem was scanned but is notillustrated. Plastic polyethylene linerseparates head of femoral stem fromacetabular cup. Femoral head shouldbe symmetrically located within cup.(B) Sagittal reformation, anterior to theleft, same patient. Two screws securecup to acetabulum. Trabecular bone canbe seen extending directly to peripheralmargin of cup, indicating integration ofcomponent into host bone.

Figure 7 (A) AP radiograph of pelvis of an 81-year-old woman with bilateral total hip prostheses. Liner wear has resulted in cephaladmigration of both femoral heads. Medial right acetabulum is absent and right acetabular cup has migrated medially. Large radiolucentzone noted lateral to right acetabular cup. Proximal right femur has resorbed. (B) Close-up of right hip prosthesis shows medial wallfracture (arrow). Innumerable faint radiopaque beads are present in right inferior acetabular region. New beads visualized after initialpostoperative radiographs often signal loosening of implant. (C) Coronal reformation of same patient shows liner wear with asymmetriclocation of femoral head in cup. Largemedial acetabular wall deficiency readily identified. (D) Sagittal reformation, same patient, anteriorto the left of image. Large posterior acetabular defect (*) indicates severe bone stock loss. Radiopaque beads are faintly visible (arrows).(E) Coronal reformation, more posterior than image in C, shows posterior cup is floating without support in a large osteolytic zone.(F) Coronal reformation, same location as E, narrower window to emphasize soft tissue details. Large cyst-like pseudomembrane(arrowheads) arises from medial acetabulum and extends into inner pelvis.


Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

be barely visible at CT imaging. Wear between thesecomponents should be minimal over the life of theprosthesis. If there is asymmetry, there may be a mis-match between the femoral head and the acetabular cupor a manufacturing defect.

Imaging of patients with hip prostheses using CTis relatively straightforward. All imaging is performedusing the highest 140 kVp setting. For most patientswith a single hip prosthesis, 350 to 450 mAs producesgood images; when bilateral hip prostheses are present,450 to 600 mAs produces good images. We use a smallfocal spot; however, depending on equipment vendor, itmay not be possible to achieve the recommended hightechnique settings using a small focal spot. Higher mAssettings can be used in larger patients, but results maynot be as consistent and there are diminishing returns toincreased technical factors. The metal of the implantscan confound x-ray beam dose modulation software, so itis best to turn this feature off. Most patients with hipimplants are past childbearing years and, although careshould be taken to avoid excessive radiation, it is less ofan issue in these patients. To minimize the cone beamartifacts, the smaller detector array configuration isselected (for 16 slice scanners: 16� 0.75 mm for Philipsand Siemens; 16� 0.625 for GE) and a very low pitchsetting (0.3, if possible).

Scanning begins just above the top of the iliaccrests and continues through the tips of the femoralstems and any associated cement. The entire scan volumeis reconstructed with both sides in the FOV (�350 mm)using a standard image reconstruction filter (C onPhilips) creating 3-mm contiguous images. It is impor-tant to survey the entire pelvis as the patient may haveanother reason for symptoms such as a sacral or pubicinsufficiency fracture. In addition, the full extent ofpseudomembrane formation can be imaged when it is

encountered. Despite the fact that one can bill legiti-mately only for the symptomatic side, it is important toconsider reconstruction of the asymptomatic hip forseveral reasons. If the hip has been replaced, it is notuncommon to encounter particle disease or signs ofloosening in the contralateral hip. If the hip has notbeen replaced, one can look for degenerative joint diseaseor findings of osteonecrosis, both of which may beimportant to the orthopedic surgeon.

Additional reconstructions can be made ofeach hip from the raw data. For the side containingan implant, images are reconstructed using a 160- to220-mm FOV with 1- or 1.5-mm-thick slices at50% overlap using a soft image reconstruction filter(B on Philips). These images are reconstructed fromthe midpelvis level to the end of the scanned volume.More cephalad slices are reconstructed if metal com-ponents or screws extend superiorly. Using these axialsource images, 24 to 36 MPRs are created in thesagittal and coronal planes using 1.5-mm-thick slices.Two sets of MPRs are created in each plane: one setcontains the entire series of images to illustrate the fulllength of the femoral stem; the second set is limited tothe acetabular cup and the proximal femoral stem.This additional set of reformations is important be-cause the majority of periprosthetic bone loss occursin this region. A native hip is reconstructed with a160-mm FOV with 0.8-mm-thick slices at 0.4-mmincrements using a standard (Philips C) or bone filter(Philips D), depending on the patient’s size. Thenative hip series is reconstructed from just above thelevel of the acetabulum to just below the lesser tro-chanter. A single set of 1-mm-thick sagittal andcoronal MPRs are created from these axial sourceimages.

Lucencies or erosions surrounding the prosthesismay be an indicator for loosening or failure of thecomponents.7 A normal metal cup closely approximatesthe reamed acetabulum. Trabecular bone should bevisible to the level of the cup and any associated screws(Fig. 6). If the cup has an ingrowth coating, there shouldbe no bead shedding. If the cup is cemented, thereshould be close approximation between the cementmantle and the underlying bone. A radiolucent zonemore than a few millimeters surrounding the cement issuggestive of linear osteolysis and may indicate looseningof the acetabular component. For noncemented cups,smaller cyst-like lucencies can be encountered normallyat unoccupied screw holes, possibly related to egress ofpressurized synovial fluid between the liner and themetal. Preexisting degenerative cysts account for someof the smaller lucencies encountered elsewhere. Largerlytic zones surrounding an acetabular component mayindicate particle disease. When present, a search forassociated pseudomembrane formation in the adjacentsoft tissues is recommended. For noncemented cups, the

Figure 8 Coronal reformation of a total hip prosthesis withceramic femoral head and ceramic acetabular liner.


Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

most important finding of stability is trabecular bone atthe rim. Large radiolucent zones in the nonperipheralzone of the cup may be tolerated if fixation is secure atthe periphery. The radiologist’s report should indicatethe location and approximate size of lytic zones. Osteol-ysis is not always caused by particle disease, so it is best toterm these zones of lysis ‘‘bone stock loss.’’

As the femoral neck is exposed to synovial fluid,erosion of the bone in this location is another findingencountered in particle disease. Large lytic zones andfractures can occur in the greater trochanter region aswell. Because most femoral stems transfer force to themore distal bone, it is possible to encounter proximalosteopenia normally caused by stress shielding, and thisshould not be confused with the osteolytic zones ofparticle disease. Other findings of failure include sub-sidence (inferior migration of the femoral stem relativeto the shaft) and, in the case of an ingrowth coated stem,bead shedding. In addition, lucency, erosion, and changein position over time, particularly at the tip, are worri-some signs.8

Many common problems can be evaluated withstandard radiography, and it is important to rememberthat radiographs have spatial resolution that is superiorto that of the most advanced clinical CT scanner. A hostof signs are used to assess stability of the prosthesis onradiographs.8 With the exception of a subtle fracture ormild bone stock loss in the greater trochanter, thefemoral portion of a total hip prosthesis can be evaluatedadequately with a simple frontal projection and Lauen-stein lateral set of radiographs. The entire length of thestem, including any cement, needs to be radiographed assome patients experience a fracture of the femur orfailure of the stem at or below the tip of the stifferfemoral stem. For this reason, inclusion of the entirestem is important for the CT technique.

KNEE IMPLANTSThere are various knee implants to accommodate thedifferent requirements of knee arthroplasty. These in-clude an isolated patellofemoral prosthesis, a unicondylarprosthesis, and a total knee prosthesis with or without apatellar resurfacing component. The components of atotal knee prosthesis can connect via a hinge mechanismto provide additional support when the cruciate liga-ments are absent. Long-stemmed prostheses are en-countered following a revision arthroplasty (Fig. 3).Regardless of the configuration, most total knee pros-theses consist of a metal femoral component that artic-ulates with a plastic liner affixed to the tibial component.The plastic liner can be replaced after implantation, ifnecessary.

Because of the dense metal, it is extremely chal-lenging to image the femoral portion of total kneeimplants and there is limited visibility of the condylar

region. The remainder of the prosthesis is usually vi-sualized very well. In general, the components arecemented in place, but ingrowth coating can be seen insome implants. When present, the patellar resurfacingcomponent consists of a plastic surface that is secured tothe patella with cement. As with any hardware or im-plant, it is advisable to examine the entire extent of bothprosthesis components.

The CT examination of a knee prosthesis is verysimilar to that of the hip. Because some patients withimplants have bilateral prostheses, it is helpful if thetechnologist questions the patient prior to positioning. Ifthere is a contralateral prosthesis, the asymptomatic sideshould be flexed and supported by a pillow or sponge tofacilitate penetration of the x-ray beam; flexion of thecontralateral side prevents the undesirable simultaneousoverlap of both condylar components by the beam(Fig. 3A). As with the hip examination, 140 kVp anda high mAs technique are suggested (400–600 mAs).Images are acquired using the thinner detector arrayconfiguration and a low pitch setting. Images are recon-structed using a soft tissue filter (Philips B) with 1- to2-mm-thick slices at 50% overlap. These axial sourceimages are subsequently reformatted into coronal andsagittal planes using 1- to 2-mm-thick slices.

Signs of loosening and failure are similar to thosefor the hip. These include fracture, radiolucent zonessurrounding the cement mantle of the components, andosteolysis. As with the hip, it is possible to encounterstress shielding around the tibial component of theknee prosthesis, which arises as bone atrophy at theproximal margins of the tibial component. For the kneeprosthesis, serial radiography remains the best means toassess stability and CT should be considered an adjunctexamination only.

SHOULDER IMPLANTSShoulders are replaced less frequently than hips andknees and the mechanics are different because theshoulder is a non–weight-bearing joint. Unlike thosewith hip and knee implants, patients undergoingshoulder arthroplasty may not experience an increase injoint mobility following surgery. Reduction of pain is themain purpose for shoulder arthroplasty. Analogous tothe hip, a humeral head prosthesis may be placed aloneor in conjunction with a plastic glenoid cup. Thesecomponents are usually cemented in place. Frictionalforces are minimal and wear of the glenoid cup isunusual.

The CT technique is similar to that for otherjoints. However, it is helpful to place the contralateralarm overhead to reduce the thickness of tissue at thelevel of the shoulder joint. The affected arm is placed bythe side of the body in mild external rotation. Scanningshould proceed from superior to inferior and technical


Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

parameters should be adjusted to allow a 20-secondscan time so that the respiration can be suspended.Images are acquired with technical factors similar tothose of the hip. Sagittal and coronal multiplanarreformations are created in oblique planes like theplanes used for routine shoulder magnetic resonanceimaging.

The humeral prosthesis is extremely dense, andvisualization of the proximal bone surrounding the headof the prosthesis is limited. It is usually possible to seethe full extent of the adjacent glenoid, with or without aglenoid component. It is not uncommon to perform theCT examination following a single contrast arthrogramto assess for loosening. This is challenging, particularly ifa noncontrast examination has not been performed. It isadvisable to consider digital subtraction arthrography inconjunction with the CT examination to increase thelikelihood of detecting contrast flow around the cement

of the components. Because the rotator cuff is inter-rupted when placing the humeral head prosthesis, asmall perforation of the rotator cuff may be encounteredas it is difficult to obtain a watertight seal followingrotator cuff repair.

IMPLANTS IN OTHER JOINTSImaging of implants in the elbow, wrist, and ankle(Fig. 9) is less common than imaging of implants inthe hip, knee, and shoulder. Serial radiography remainsthe main imaging method to assess for loosening andfailure, although CT may provide additional informa-tion when radiography is inconclusive. Our experience todate is limited, but the imaging principles and CTtechnique are similar to those previously outlined witha few exceptions. When imaging the elbow or wrist, it isadvisable to scan the patient with the affected arm

Figure 9 (A) AP radiograph of 56-year-old woman with implant placed for anklereconstruction following severe traumanow complaining of increasing pain. Dis-tal tibia and fibula are fused; inferior as-pect of fibular side plate abuts lateralcalcaneus. (B) Lateral radiograph, samepatient. Stability of ankle implant is diffi-cult to assess because of overlappingmetal and bone. (C) Coronal reformation,same patient. Tibial component of ankleprosthesis appears well seated. Limitedbone noted around talar portion of pros-thesis. (D) Sagittal reformation, same pa-tient. Large area of bone stock lossinferior to talar portion of ankle prosthe-sis. Degenerative cysts noted in talona-vicular joint. At surgery, talar componentwas loose; tibial component was stable.


Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

overhead to limit the tissue traversed by the x-ray beam.Imaging of the ankle can be performed in the ‘‘toes up’’position to facilitate positioning of the patient. Depend-ing on the density of the metal, multiplanar reformationsshould be created with 1- to 2-mm-thick slices. Loosen-ing or failure is predicated on the detection of radio-lucent zones surrounding the cement mantle of theprosthesis components or on the detection of fracturelines.

CONCLUSIONSThe development of multichannel CT scanners providesus with new imaging opportunities in musculoskeletalradiology. The radiologist and the orthopedic surgeonshould be aware that it is possible to image most patientswith hardware with good results. The additional capa-bilities of the equipment allow us to image most patientswith orthopedic hardware and joint implants. The keyfactors that enable us to image these patients are asfollows: higher technical factors, low pitch settings, andhigh-quality multiplanar reformations. The key param-eters that allow us to predict the likelihood of successfulimaging of implants and hardware are composition ofthe hardware, shape (geometry) of the device, andphysical location of the device relative to the region ofinterest.

The goal of most imaging following open reduc-tion and internal fixation is the assessment of healing.Because the determination of delayed union or nonun-ion is based on imaging as well as clinical findings, theradiologist’s role is to document the presence or absenceof bone bridging; when hardware is present, CT can bea useful tool to examine the underlying bone forbridging.

Multichannel CT is useful in examining patientswith joint implants. Attention to technical details isessential to obtain high-quality images. When dense

metal is present, such as a hip prosthesis, it is useful toreview the images using coronal and sagittal multiplanarreformations created with thicker slices (1–2 mm).Prosthesis failure can be related to fracture, infection,particle disease, and osteolysis. Most of these complica-tions can be detected with CT. Furthermore, these scansare useful for preoperative planning when revision ar-throplasty is under consideration.

If there is a question about fracture healing or animplant that cannot be answered with conventionalradiography, it is generally worthwhile to consider CTimaging.

REFERENCES

1. White LM, Buckwalter KA. Technical considerations: CTand MR imaging in the postoperative orthopedic patient.Semin Musculoskelet Radiol 2002;6:5–17

2. Fishman EK, Magid D, Robertson DD, et al. Metallic hipimplants: CT with multiplanar reconstruction. Radiology1986;160:675–681

3. Haramati N, Staron RB, Mazel-Sperling K, et al. CT scansthrough metal scanning technique versus hardware composi-tion. Comput Med Imaging Graph 1994;18:429–434

4. Chiang PP, Burke DW, Freiberg AA, et al. Osteolysis of thepelvis: evaluation and treatment. Clin Orthop Relat Res2003;(417):164–174

5. DeFrang RD, Guyer WD, Porter JM, et al. Synovial cystformation complicating total hip arthroplasty: a case report.Clin Orthop Relat Res 1996;(325):163–167

6. Adelman SC, Urquhart AG, Sondak V, et al. Polyethylenewear debris presenting as a retroperitoneal tumor. Surgery1998;123:111–112

7. Puri L, Wixson RL, Stern SH, et al. Use of helical computedtomography for the assessment of acetabular osteolysis aftertotal hip arthroplasty. J Bone Joint Surg Am 2002;84:609–614

8. Engh CA, Massin P, Suthers KE. Roentgenographic assess-ment of the biologic fixation of porous-surfaced femoralcomponents. Clin Orthop Relat Res 1990;(257):107–128


Dow

nloa

ded

by: D

artm

outh

Col

lege

. Cop

yrig

hted

mat

eria

l.

Documents

Multichannel CT Imaging of Orthopedic Hardware and Implants · Multichannel CT Imaging of Orthopedic Hardware and ... metal hardware cause more attenuation and more arti- ... MULTICHANNEL