CLINICAL ORTHOPAEDICS AND RELATED RESEARCHNumber 412, pp. 117–124© 2003 Lippincott Williams & Wilkins, Inc.

117

Computed tomography can assist in the detec-tion of periprosthetic osteolysis, but it has notbeen used to measure the actual volume of bonedefects adjacent to hip replacement componentsbecause of the scanning artifact caused bymetal. The aim of the current study was to de-velop a spiral computed tomography techniquethat provides precise and reliable volumetricmeasurement of bone defects adjacent to unce-mented metal-backed acetabular components.Computed tomography scans were taken ofsmall and large defects of known volume cre-ated in the ilium in a bovine hemipelvis and apelvis from a cadaver. Scans were analyzed by

two independent observers. The computed to-mography operating conditions were determinedthat enabled volumetric measurements and thatwere accurate to within 96% for small and largedefects and precise to greater than 98% forsmall and large defects. This computed tomog-raphy technique has the capability to measureaccurately and precisely the volume of bone de-fects in the ilium adjacent to metal-backed ac-etabular components. This technique has clearadvantages over plain radiographs. It will allowinvestigation of the natural history of osteolyticlesions, enhance preoperative planning, and im-prove monitoring of the outcomes of treatmentsof osteolysis.

Although the middle to long-term clinical re-sults of total hip replacement using unce-mented metal-backed acetabular componentsare promising, loss of bone from around theseprostheses, termed periacetabular osteolysis,remains a significant concern.8,9,13,15,22,23 Theseacetabular components may remain well-fixedin the presence of progressive osteolysis evenwhile the patient is asymptomatic.22 This mayresult in prosthesis loosening, for which thepatient requires complicated and expensive re-vision surgery, or major bone grafting before

Measurement of Bone Defects Adjacent to Acetabular

Components of Hip Replacement

Roumen Stamenkov, MD*; Donald Howie, MBBS, PhD*; James Taylor, MBBS**; David Findlay, MSc, PhD*;

Margaret McGee, BSc, MPH*; George Kourlis, BSc**; Angelo Carbone, BSc*; and Matthew Burwell, MBBS*

From the *Departments of Orthopaedics and Trauma,Royal Adelaide Hospital and the University of Adelaide;and the **Department of Radiology, Royal AdelaideHospital, Adelaide, South Australia.Supported by the Royal Adelaide Hospital, the AdelaideBone and Joint Research Foundation, Novartis Pharma-ceuticals and Bristol-Myers Squibb/Zimmer.Reprint requests to Roumen Stamenkov, MD, Dept Or-thopaedics and Trauma, L4 Bice Building, Royal Ade-laide Hospital, North Terrace, Adelaide, South Australia5000.Received: April 10, 2002.Revised: September 18, 2002.Accepted: October 29, 2002.DOI: 10.1097/01.blo.0000069001.16315.f4

loosening. Although it is recommended thatpatients with these hip replacements be moni-tored closely to identify bone loss, there arecurrently no reliable techniques available tomeasure the amount of bone loss.

Early identification and quantification ofperiprosthetic osteolysis using plain radio-graphs is often difficult or inaccurate.10 Astudy by Robertson et al19 reported that plainradiographs can underestimate bone loss by atleast 20%. Computed tomography (CT) scanscan reveal osteolysis that is not detectable onplain radiographs and also gives much moreinformation about the extent and anatomic lo-calization of the osteolytic lesions.19 Computedtomography scanning is used for preoperativeplanning3,4,6 and is used as a diagnostic tool tomeasure volumetric changes of musculoskele-tal and other tumors.25 However, attempts toextend this application of CT scanning to themeasurement of osteolytic lesions around totalhip replacement prostheses have been limitedbecause of the artifact produced by the metalliccomponents of total hip replacement or thescrews used to fix the components.14,20

Metal artifact reduction software is availablewith spiral CT scanning, but, although the ad-vantages of spiral CT scanning in identifyingosteolysis have been described,19 the accuracyof the technique in measuring the volume of os-teolytic lesions has not been reported. There-fore, the aim of the current study was to developa technique using spiral CT scanning that al-lows precise and reliable measurements of thevolume of osteolytic lesions in the bone adja-cent to uncemented metal-backed acetabulartotal hip replacement components.

MATERIALS AND METHODS

Because of the limited availability of pelves fromcadavers, the initial development and validation ofthe CT scanning technique was done using a bovinehemipelvis containing a unilateral total hip re-placement prosthesis. Additional validation of thetechnique was done in a pelvis from a cadaver, intowhich unilateral and bilateral acetabular compo-nents were implanted consecutively.

It has been reported that most osteolytic lesionsadjacent to metal-backed acetabular components oftotal hip replacement are periacetabular, and le-sions commonly occur in the ilium in associationwith fixation screws.22 Therefore, the current au-thors examined CT measurements of simulated ac-etabular bone defects in the ilium in the presence ofthe acetabular component and the fixation screws.

SpecimensA fresh-frozen hemipelvis from a bovine, compa-rable in size to an acetabulum in a human, was de-frosted from �20� C to room temperature. Thebovine acetabulum was orientated in the sameanatomic orientation as in the human, and the spec-imen was mounted in a table vice and the acetabu-lum was prepared. The acetabulum was reamedinitially using a 44-mm outer diameter reamer, fol-lowed by 54-, 55-, 56-, and 57-mm outer diameterreamers (Zimmer Inc, Warsaw, IN) to accept a 58-mm outer diameter TiA1V Trilogy fiber metal-coated shell with three screw holes (Zimmer Inc).The acetabular component was implanted into theacetabulum of the specimen in 40� operative incli-nation and 25� operative anteversion17 and rotatedso that one screw hole was superior, one screw holewas posterosuperior, and one screw hole was an-terosuperior. A Kirschner (K) wire was placedthrough the superior screw hole into the ilium andoriented with its axis parallel to the longitudinalaxis of the specimen. After marking the position ofthe cup on the reamed surface of the acetabulum,the cup was removed. A large, 40-mm deep cylin-drical defect was created by drilling over the K wirewith a 12-mm cannulated drill. The axis of the largesimulated defect was parallel to the longitudinalaxis. The procedure was repeated in the posterosu-perior screw hole to create a small 25-mm deepcylindrical defect, orientated posterosuperiorly, us-ing a 9-mm cannulated drill (Fig 1).

The volumes of the simulated defects were de-termined by filling with water and measuring threetimes using a 0.01-mm graduated 1-mL syringe.The acetabular component was reimplanted usingan impactor-positioner (Zimmer Inc). A 30-mmlong, 6.5-mm diameter TiA1V self-tapping bonescrew (Zimmer Inc) was inserted into the supero-lateral screw hole so that it engaged in 5 mm ofbone distal to the created defect. A 25-mm longsimilar screw was inserted through the superome-dial screw hole with no defect. A Trilogy polyeth-ylene liner (Zimmer Inc) was inserted into the com-

Clinical Orthopaedics118 Stamenkov et al and Related Research

ponent. The specimen was placed into a plasticcontainer and orientated so that its longitudinal axiswas parallel to the midline of the container and itssagittal plane was perpendicular to the base of thecontainer. The proximal aspect of the specimenwas fixed to the vertical wall of the container withthree screws. The ischium was rested on the base ofthe container and was fixed with one screw. A ce-mentless PCA CoCr alloy femoral stem (Howmed-ica International, Ltd, Rutherford, NJ) and 28-mmdiameter CoCr alloy head (Howmedica Interna-tional) were positioned into the liner.

An intact pelvis with L4 and L5 vertebra andfemoral bones was retrieved from the cadaver of a70-year-old man. The specimen was defrostedfrom �20� C to room temperature and the musclesand the hip capsules were removed. The right ac-etabulum was prepared in the same manner as thatfor the acetabulum from the bovine, except that 46-mm, 48-mm, 50-mm, and 51-mm outer diameterreamers were used consecutively.

For the human pelvis, one acetabular defect inthe ilium was created. A 52-mm outer diameterHarris-Galante-II (Zimmer Inc) uncemented, metal-backed acetabular component was implanted in 40�

operative inclination and 25� operative antever-sion.17 The component was fixed with two self-tapping TiA1V screws 6.5-mm diameter and 30-mmlong (Zimmer Inc). A polyethylene liner (ZimmerInc) was inserted into the component. An unce-mented Versys fiber metal midcoat femoral stem(Zimmer Inc) was implanted into the right femur,using a standard uncemented technique. A 28-mmCoCr femoral head (Zimmer Inc) was inserted. Thepelvis was placed in a supine position so that thespinous tubercles of the sacrum and posterior-superior spines of the ilium rested on the base of acontainer. The line joining the midpoints of thesymphysis pubis and the L5 vertebra was repre-sented as a longitudinal axis of the specimen andwas oriented parallel to the midline of the containerin the coronal plane. The pelvis was fixed to thecontainer using plastic ties. Both specimens werecovered with tap water, which was used to simulatesoft tissues during scanning.

To investigate the effects on CT scanningcaused by a contralateral total hip replacement, anuncemented metal-backed Trilogy acetabular com-ponent (Zimmer Inc) and a cementless PCA CoCralloy femoral stem (Howmedica International, Inc,

Number 412July, 2003 Measurement of Acetabular Bone Defects 119

Fig 1. A bovine hemipelvis totalhip replacement model is shown.The diagram shows the largebone defect adjacent to the ac-etabular component and thesmall defect around a lateral fix-ation screw. Computed tomog-raphy scans at several levelsshow the following defects: (A)1, large bone defect; 2, lateralscrew; (B) 1, large bone defect;2, bone defect around lateralscrew; 3, tip of medial screw thatpenetrated medial cortex; and(C) 1, large bone defect; 2, lat-eral screw with artifact interfer-ence from top of acetabular com-ponent; and 3, medial screw.

London, United Kingdom) and 28-mm CoCr head(Howmedica, International, Inc) were implantedinto the contralateral hip and the CT scans of theoriginal study hip were repeated.

Quantitative Computed TomographyComputed tomography scanning was done using aPicker PQ6000 spiral CT (Picker International Inc,Cleveland, OH) scanner with Voxel Q FALCONmetal artifact reduction software (Picker Interna-tional Inc). Computed tomography scans weretaken at 120 kV and 200 mA to maximize the res-olution and contrast of the trabecular and corticalbone. To allow standardization of measurements,the top of the sacroiliac joint was used as theanatomic reference point and scans were taken tothe end of the ischium. A metal pin was insertedinto the ilium of the bovine hemipelvis to mark thereference point.

The simulated bone defects were measured inconsecutive CT slices displayed on the computerscreen by tracing the inner border of the defect us-ing a computer mouse. Volume measurement wasdone using algorithms coded into the commercialimage-analysis software (PQ6000 Voxel Q FAL-CON, Picker International Inc).

To determine the optimal CT scanning operat-ing conditions for measuring the volume of bonedefects adjacent to metal-backed acetabular com-ponents, two parameters were investigated in thecow hemipelvis model. These were the angle of theCT beam, �25� (cephalad), 0� and �25� (caudad),and the two thicknesses of CT slice routinely usedin clinical practice, 3 and 4 mm.

Once the protocol and operating conditionswere determined, the impact of metal artifact usingthis protocol then was examined by comparing vol-umetric measurements of the large defect from CTscans taken with and without the prosthesis in situ.The radiodensity of the bovine bone was markedlyhigher than that of the bone from an elderly man.The operating conditions and protocol tested in thebovine hemipelvis model therefore were investi-gated again using the human pelvis model im-planted with a unilateral total hip replacementprosthesis and containing a large superior defectadjacent to the acetabular component. To examinethe interference effect of a contralateral total hip re-placement prosthesis on the volumetric measure-ments, scans were repeated on the human pelvisafter implantation of a total hip replacement pros-thesis on the contralateral side.

Statistical AnalysisThe accuracy of the volumetric measurements fromthe CT scans was determined by calculation of thepercent error between the manual measurement ofthe simulated defects, done using water and a sy-ringe and the measurements from the CT scans.Volumetric measurements from the CT scans wererepeated five times in each instance by two ob-servers, an orthopaedic research registrar and a re-search officer. The reliability of the quantitativemeasurements was determined by calculation ofthe intraobserver and interobserver errors. The in-traobserver error was reported as the percent coef-ficient of variation and the interobserver error wasmeasured by the percent variance.

Interferential analysis was done using analy-sis of variance (ANOVA) and the Fisher’s least-significant-difference test. Data summary statisticswere calculated using Microsoft Excel (MicrosoftCorporation, Redmund, WA) and interferentialanalyses were done using the SAS statistical soft-ware package (SAS® System for Windows™ Re-lease 6.12 1989–1996, SAS Institute Inc, Cary,NC). A probability value less than 0.05 was con-sidered to indicate statistical significance.

RESULTS

Bovine Acetabulum StudyThe actual volume of the large simulated ac-etabular defect adjacent to the component,measured using a syringe and water, was5.03 mL (standard deviation, 0.04 mL). Theactual volume of the small defect around thelateral screw was 1.24 mL (standard devia-tion, 0.02 mL).

The estimated volumes of the large simu-lated defect measured from each of the CTscans taken using the different operating con-ditions are shown in Figure 2. There was a sig-nificant effect of beam angle and slice thick-ness on the volumetric measurement of thelarge defect (p � 0.0001). The most accuratemeasurement of defect volume was obtainedusing a beam angle of 0� and a slice thicknessof 3 mm (error, �1.7%). The volume was sig-nificantly greater than the volumes measuredfrom scans taken with a beam angle of �25�and �25� (p � 0.0001) and significantly

Clinical Orthopaedics120 Stamenkov et al and Related Research

greater, and closer to the true volume, than thatdetermined from scans taken using a slicethickness of 4 mm and a beam angle of 0�(p � 0.0001). The intraobserver error and in-terobserver error of the volume measurementsof the large defect were 2.27% and �2.73%,respectively, using the operating parametersof 3-mm slice thickness and a beam angle of0�. Using a scan thickness of 3 mm and a beamangle of 0�, there was no significant differencebetween the volumes measured from the scanstaken with and without the prosthesis in situ (p � 0.944).

For the small simulated defect around thelateral screw, measurement of the CT scanstaken using a beam angle of �25� gave signif-icantly more accurate measures of volumethan when using scans taken at 0� (p � 0.0001)(Fig 3). The CT scan taken using a beam angleof �25� resulted in the least accurate mea-surement of volume. There also was a signifi-cant difference in the volumes measured fromthe CT scans taken at 3- and 4-mm slice thick-ness (p � 0.0072). The mean intraobserverand interobserver errors of the volume mea-surements of the small defect were 0.62% and0.9%, respectively, using the operating para-

meters of 3-mm slice thickness and a beamangle of 0�.

Cadaver StudyThe actual volume of the simulated acetabulardefect adjacent to the component, measuredusing a syringe and water, was 3.42 mL (stan-dard deviation, 0.03 mL). There was a signifi-cant effect of beam angle (p � 0.0001) but noeffect of slice thickness (p � 0.647) on thevolume of the defect measured when a unilat-eral total hip replacement prosthesis was im-planted (Fig 4). With a 3-mm slice thickness,the most accurate measurements of the defectvolume were obtained with scans taken usinga beam angle of 0� (error, �1.9�) and �25�(error, �2.5%), which were both significantlymore accurate than measurements taken usinga beam angle of �25� (p � 0.0005). The in-traobserver and interobserver errors for thescans taken at 0� were 2.8% and 2.15%, re-spectively.

With bilateral total hip replacement pros-theses implanted, the volumetric measure-ments taken using a slice thickness of 3 mmand a beam angle of 0� were less accurate (er-ror, �4.2%) than when a unilateral total hip

Number 412July, 2003 Measurement of Acetabular Bone Defects 121

Fig 2. This graph shows the effect of beam an-gle and slice thickness on the volume of a largeacetabular defect measured from CT scans in abovine hemipelvis model. The dashed line indi-cates the defect volume measured using waterand a syringe. SD � standard deviation.

Fig 3. The effect of beam angle and slice thick-ness is shown on the apparent volume of a smalldefect around a fixation screw measured from CTscans in a bovine hemipelvis model. The dashedline indicates the defect volume measured usingwater and a syringe. SD � standard deviation

replacement prosthesis alone was implanted(error, �1.9%) (Fig 5). The difference in themeasured volume of the defect in the acetabu-lum when bilateral total hip replacement pros-theses were implanted was not statisticallysignificant (p � 0.164) to that measured when

only unilateral total hip replacement prosthe-ses were implanted. The intraobserver (4.2%)and interobserver (�2.3%) errors were similarto those obtained for the unilateral total hipreplacement model.

DISCUSSION

In the current study, spiral CT scanning wasable to accurately and reliably measure smalland large bone defects in the ilium adjacent toTi metal-backed acetabular components im-planted in a bovine hemipelvis and a pelvisfrom a cadaver. Volumetric measurements ofsimulated defects were accurate to within 96%for small and large defects and precise to 99%and 98% for small and large defects, respec-tively. There was a small loss of accuracy inthe estimate of the volume of the acetabulardefect when a contralateral total hip replace-ment prosthesis was present, but this was un-likely to be clinically significant. This smallloss likely was caused by the artifact from theopposite acetabular component.

The findings described here show that, inexperimental models of total hip replacement-associated bone loss, it is feasible to use CTscanning to measure the volume of defects inthe cancellous bone superior to metal-backedacetabular implants. Using the scanning para-meters that were determined, including CTslice thickness and beam angle, together withmetal artifact reduction software, accurate andreliable volumetric measurements can be ob-tained. The ability to accurately measure dif-ferent types of defects in the current study of alarge acetabular defect adjacent to the compo-nent and a small defect around a fixationscrew, is likely to be useful in measuring andmonitoring similar types of osteolytic lesionsin humans.

There were differences in the accuracy ofvolumetric estimation between CT scans takenwith different angles of the gantry. The likelyexplanation of this is the angle of the gantry.The CT volume measurement of the large de-fect using a 0� angle of beam had better accu-racy than that using �25� or �25�. This was

Clinical Orthopaedics122 Stamenkov et al and Related Research

Fig 4. The effect of slice thickness is shown onthe volume of bone defect adjacent to the ac-etabular component measured from CT scans ina pelvis from a cadaver. The beam angle was 0�.The dashed line indicates the defect volumemeasured using water and a syringe. SD � stan-dard deviation

Fig 5. The effect of bilateral total hip replace-ment is shown on the volume of bone defect ad-jacent to the acetabular component measuredfrom CT scans of a pelvis from a cadaver. Thebeam angle was 0�. The dashed line indicatesthe defect volume measured using water and asyringe. SD � standard deviation

probably because the longitudinal axis of thelarge defect was parallel to the longitudinalaxis of the specimen resulting in more of thelarge defect being seen by the axial view at 0�.

The CT volume measurement of the smalldefect using a �25� angle of beam had betteraccuracy than that using 0� or �25�. That wasprobably because the longitudinal axis of thesmall defect was orientated posteriorly fromthe tip of the acetabular component. In this po-sition, most of the small defect could be de-tected only by a posteriorly tilted CT beam,which was �25� cephalad.

This CT technique is likely to be superior toanteroposterior radiographs of the pelvis inmeasuring osteolysis around total hip replace-ment acetabular components. Studies usingplain radiographs have been restricted to us-ing two-dimensional measurements of osteol-ysis16,24 or volumetric measurements usingplanimetric measurements and mathematicformulas assuming ellipsoidal cavities.7 Be-cause of the recognized limitations of plain ra-diographs for identification and localization ofosteolytic lesions,5,12,19,26 these techniques willremain, at best, semiquantitative. Computedtomography scanning, using the techniquesdescribed, offers significant advantages and, ifthe level of accuracy in vivo approaches thatin vitro, it will allow definitive studies of thecause and outcomes of treatment of osteolysis.

A criticism of the current study is that sim-ulated defects were used to investigate theusefulness of CT scanning in this application.Osteolytic defects may be filled with granulo-matous tissue and fluid and therefore the fluid-filled defects in the current study may notexactly represent the clinical situation. Pre-liminary studies in patients, however, have re-sulted in similar CT images as seen in thecurrent model. Clearly this is an area for addi-tional study.

Other imaging methods currently availablecannot be used to determine the volume of oste-olytic defects adjacent to metallic componentsof total hip replacement prostheses, but dual en-ergy xray absorptiometry (DEXA)1,2,11,18 can beused to assess overall bone density. It will be

of interest to apply DEXA in parallel with CTscanning to investigate associations betweenCT measurements and overall bone densitymeasurements. This would be of particular in-terest in exploring the potential benefit ofpharmacologic treatment of osteolysis, includ-ing the use of bisphosphonates, in which treat-ment would be expected to influence bonedensity generally, and inhibit bone resorptionat the specific site of bone loss.

A precise and reliable CT quantitativemethod for measuring the volume of peripros-thetic osteolytic lesions has been described.Although currently a research tool, it is pro-posed that this technique could be further de-veloped to allow assessment of osteolysis,monitoring of the natural history of peripros-thetic osteolysis and its relationship to wear,and the potential benefits of surgical treatment,including bone grafting without prosthesis re-moval,21 or pharmacologic treatment.

AcknowledgmentsThe authors thank Mishelle Korlaet, PhillipaSharpe, and John Anderson for assistance in studydesign and development of CT scan protocols.

References1. Adolphson P, von Sivers K, Dalen N, Jonsson U,

Dahlborn M: Bone and muscle mass after hip arthro-plasty: A quantitative computed tomography study in20 arthrosis cases. Acta Orthop Scand 164:181–184,1993.

2. Adolphson P, von Sivers K, Jonsson U, Dalen N,Dahlborn M: Bone and muscle mass after hiprearthroplasty: Controlled CT study of 12 patients.Acta Orthop Scand 64:282–284, 1993.

3. Barmeir E, Dubowitz B, Roffman M: Computed to-mography in the assessment and planning of compli-cated total hip replacement. Acta Orthop Scand53:597–604, 1982.

4. Berman AT, McGovern KM, Paret RS, Yanichko JrDR: The use of preoperative computed tomographyscanning in total hip arthroplasty. Clin Orthop222:190–196, 1987.

5. Berry DJ: Management of osteolysis around total hiparthroplasty. Orthopedics 22:805–808, 1999.

6. Clark HJ, Jinnah RH, Cox QG, Curtis MJ: Comput-erized templating in uncemented total hip arthro-plasty to assess component fit and fill. J Arthroplasty7:235–239, 1992.

7. Garcia-Cimberelo E, Madero R, Blasco-Alberdi A,Munuera L: Femoral osteolysis after low-frictionarthroplasty: A planimetric study and volumetric es-timate. J Arthroplasty 12:624–634, 1997.

Number 412July, 2003 Measurement of Acetabular Bone Defects 123

8. Harris WH: The problem is osteolysis. Clin Orthop311:46–53, 1995.

9. Howie DW: Tissue response in relation to type ofwear particles around failed hip arthroplasties. JArthroplasty 5:337–348, 1990.

10. Jerosch J, Steinbeck J, Fuchs S, Kirchhoff C: Radio-logic evaluation of acetabular defects on acetabularloosening of hip alloarthroplasty. Unfallchirurg99:727–733, 1996.

11. Kiralti BJ, Heiner JP, McBeat AA, Wilson MA: De-termination of bone mineral density by dual x-rayabsorptiometry in patients with uncemented total hiparthroplasty. J Orthop Res 10:836–844, 1992.

12. Learmonth ID, Hussell JG, Grobler GP: Unpre-dictable progression of osteolysis following ce-mentless hip arthroplasty: 24 femoral componentsfollowed for 6–10 years. Acta Orthop Scand67:245–248, 1996.

13. Learmonth ID, Smith EJ, Cunningham JL: The patho-genesis of osteolysis in two different cementless hipreplacements. Proc Inst Mech Eng 211:59–63, 1997.

14. Link TM, Berning W, Scherf S, et al: CT of metalimplants: Reduction of artefacts using an extendedCT scale technique. J Comput Assist Tomogr24:165–172, 2000.

15. Maloney WJ, Galante JO, Anderson M, et al: Fixation,polyethylene wear, and pelvic osteolysis in primary to-tal hip replacement. Clin Orthop 369:157–164, 1999.

16. Maloney WJ, Herzwurm P, Paprosky W, RubashHE, Engh C: Treatment of pelvic osteolysis associ-ated with a stable acetabular component insertedwithout cement as part of a total hip replacement. JBone Joint Surg 79A:1628–1634, 1997.

17. Murray DW: The definition and measurement of ac-etabular orientation. J Bone Joint Surg 75B:228–232,1993.

18. Neander G, von Silvers K, Adolphson P, DahlbornM, Dalen N: An evaluation of bone loss after totalhip arthroplasty for femoral head necrosis after fem-oral neck fracture: A quantitative CT study in 16 pa-tients. J Arthroplasty 14:64–70, 1999.

19. Robertson DD, Sutherland CJ, Lopes T, Yuan J: Pre-operative description of severe acetabular defectscaused by failed hip replacement. J Comput AssistTomogr 22:444–449, 1998.

20. Robertson DD, Yuan J, Wang G, Vannier MW: To-tal hip prosthesis metal-artefact suppression using it-erative deblurring reconstruction. J Comput AssistTomogr 21:293–298, 1997.

21. Schmalzried TP, Fowble VA, Amstutz HC: The fateof pelvic osteolysis after reoperation: No recurrencewith lesional treatment. Clin Orthop 350:128–137,1998.

22. Soto MO, Rodriguez JA, Ranawat CS: Clinical andradiographic evaluation of the Harris-Galante cup:Incidence of wear and osteolysis at 7 to 9 yearsfollow-up. J Arthroplasty 15:139–145, 2000.

23. Thanner J, Karrholm J, Malchau H, Herberts P: Pooroutcome of PCA and Harris-Galante hip prostheses:Randomised study of 171 arthroplasties with 9-yearfollow-up. Acta Orthop Scand 70:155–162, 1999.

24. Wan Z, Dorr LD: Natural history of femoral focal os-teolysis with proximal ingrowth smooth stem im-plant. J Arthroplasty 11:718–725, 1996.

25. Wellings RM, Davies AM, Pyncent PB, Carter SR,Grimer RJ: The value of computed tomographicmeasurements in osteosarcoma as a predictor ofresponse to adjuvant chemotherapy. Clin Radiol49:19–23, 1994.

26. Zimlich RH, Fehring TK: Underestimation of pelvicosteolysis: The value of the iliac oblique radiograph.J Arthroplasty 6:796–801, 2000.

Clinical Orthopaedics124 Stamenkov et al and Related Research