Portal de Periódicos da CAPES

New Radiographic Technique May Renew Credibility of Bone Densitometry

1988; Elsevier BV; Volume: 63; Issue: 11 Linguagem: Inglês

10.1016/s0025-6196(12)65513-x

1942-5546

David J. Sartoris, Donald Resnick,

Radiomics and Machine Learning in Medical Imaging

In this issue of the Proceedings (pages 1075 to 1084), Wahner and associates present a comprehensive and much-needed evaluation of the performance of the first commercially available system for dual-energy x-ray absorptiometry (DEXA).1Sartoris DJ Sommer FG Marcus R Madvig P Bone mineral density in the femoral neck: quantitative assessment using dual-energy projection radiography.AJR. 1985; 144: 605-611Crossref PubMed Scopus (26) Google Scholar, 2Brody WR Butt G Hall A Macovski A A method for selective tissue and bone visualization using dual energy scanned projection radiography.Med Phys. 1981; 8: 353-357Crossref PubMed Scopus (177) Google Scholar, 3Sartoris DJ Sommer FG Kosek J Gies A Carter D Dual-energy projection radiography in the evaluation of femoral neck strength, density, and mineralization.Invest Radiol. 1985; 20: 476-485Crossref PubMed Scopus (31) Google Scholar At the University of California in San Diego (UCSD), our musculoskeletal imaging section has worked with the Hologic QDR-1000 bone densitometer for an approximately equivalent duration of time in a variety of phantom, cadaver, and patient studies, and we share the authors' enthusiasm for the new technology. The improved precision and shorter examination time of DEXA in comparison with dual photon absorptiometry (DPA) should ultimately render bone density measurement more acceptable to referring physicians, patients, and third-party reimbursement organizations and should result in widespread distribution and broad application. Another system for dual-energy bone densitometry that uses an x-ray tube photon source has been developed by Lunar Radiation Corporation, the leading manufacturer of DPA equipment.4Mazess RB Sorenson J Hanson JA Performance of an x-ray dual-photon scanner.in: Dequeker JV Geusens D Wahner HW Bone Mineral Measurements by Photon Absorptiometry: Methodological Problems. Leuven University Press, Leuven, Belgium1988: 415-418Google Scholar The Lunar DPX device differs from the Hologic unit in its approach to energy separation: it uses selective constant-potential K-edge filtration, as opposed to alternate tube pulsing at different voltages. The DPX also differs in the resultant effective beam energies (40 and 70 keV for the DPX versus 43 and 110 keV for the Hologic system). An inherently stable, low-scatter (operator exposure is less than 0.25 mR at 1 m), relatively monochromatic beam exposes each pixel uniformly to both energies, and the results are affected only negligibly by tissue thickness, composition, or both. An integral-line scintillation detector and photomultiplier tube are used for photon counting. Lunar software is designed to provide densitometric analysis of the spine, proximal femur, tibia, proximal humerus, radius, and total body, including percentage lean/fat soft-tissue characterization. Regional scanning takes less than 4 minutes; total body measurements are obtained in about 20 minutes. A regional scan and analysis program permits evaluation of up to five regions of interest within a user-specified scan area. The total-body program includes automatic location of nine anatomic regions and also local region inclusion or exclusion options. A special high-resolution mode is available for scanning pediatric patients or local regions of interest. Lunar reports that the DPX system is being investigated for possible monitoring of dissolution of renal stones, quantification of pulmonary nodules, imaging of contrast material in organs, and examination of responses adjacent to fractures, prostheses, and pagetic lesions. Lunar claims direct data comparability with DPA results and offers an upgrade program for current DPA users. Experience with this system for a 6-month period at UCSD has been extremely favorable. Norland Corporation (Fort Atkinson, Wisconsin) has also recently announced the development of a work-in-progress x-ray bone densitometer (model XR26). The device features whole-body scanning capability with use of a K-edge filtered x-ray source. The computer system includes an IBM PS/2 model 60 with a 44-megabyte hard disk, 1.44-megabyte floppy disk, 1-megabyte random-access memory, and video graphics array high-resolution monitor. Preliminary specifications include a 5-minute scan time, 1-mm image spatial resolution, and precision error of less than 1.0%. In general, no major criticisms of the report by Wahner and colleagues can be advanced. It should be noted, however, that many investigators believe that DEXA affords a lower radiation dose than DPA, which contrasts with the results of their study. A somewhat surprising finding was that DEXA exhibited greater dependence on position of the spine within the radiation beam than DPA in this investigation, a result that should be verified at other institutions. It has also been the experience of our technologists that both the hardware and the software of DEXA are more user-friendly than those of DPA, an advantage not emphasized in this study. Short- and long-term precisions of the Hologic system and the Lunar DP3 DPA unit were also examined at UCSD by using two phantom types: three mineral-equivalent cylinders of variable diameter (simple geometry) and the Hologic calcium hydroxyapatite standard modeled as four lumbar vertebrae (complex geometry). Each instrument had a smaller coefficient of variation for data obtained with the simple versus the complex geometry phantom. The short-term precision of the QDR-1000 was 0.88%; the long-term precision was 0.97%. For the Lunar DP3, the short- and long-term precisions were 2.9% and 2.6%, respectively. In long-term precision, DPA was one-third to one-fifth as precise as DEXA for both phantoms. The accuracy of DEXA, as assessed by repetitive scanning of a phantom containing known dipotassium hydrogen phosphate concentrations, was high (r = 0.99; slope = 1.005).5Hunter JA, Sartoris DJ, Stein JA, Ramos E, Moscona A, Andre M, Resnick D: Quantitative dual-energy digital radiography versus dual-photon absorptiometry: in vitro precision and accuracy. Presented at the Association of University Radiologists 36th Annual Meeting, New Orleans, Louisiana, April 17 to 21, 1988Google Scholar DEXA has also been used to measure the mineral content and density of lumbar vertebrae (L-1 through L-3) in 11 cadavers (4 men and 7 women with a mean age of 75&½; years and a range from 56 to 98 years). For determining the accuracy of DEXA, the results were compared with ash weight, dry weight, ash density, and dry density. Excellent correlation was obtained between bone mineral content measured by DEXA and ash weight (r = 0.963; P>0.0001). Likewise, the correlation between bone mineral content by DEXA and dry weight was also excellent (r = 0.965; P>0.0001). The accuracy error in determining mineral mass in lumbar vertebrae was about 9%. In addition, a strong correlation was observed between bone mineral density determined by DEXA and both ash density (r = 0.881; P>0.0001) and dry density (r = 0.903; P>0.0001). Moreover, the relationship between the actual bulk volume of lumbar vertebrae and the projected area determined by DEXA was significant (r = 0.853; P>0.0001). In an attempt to develop a normative data base rapidly for DEXA, examinations of the lumbar spine and proximal femur are currently offered free of charge at UCSD to any volunteer. To date, calculated best-fit regression equations for age-related bone loss in an unselected population of more than 1,000 volunteers have been as follows: density = −0.00246 (age) − 1.09842 (R2Brody WR Butt G Hall A Macovski A A method for selective tissue and bone visualization using dual energy scanned projection radiography.Med Phys. 1981; 8: 353-357Crossref PubMed Scopus (177) Google Scholar = 0.05; SEE = 0.16) in women and density = −0.00146 (age) − 1.11793 (R2Brody WR Butt G Hall A Macovski A A method for selective tissue and bone visualization using dual energy scanned projection radiography.Med Phys. 1981; 8: 353-357Crossref PubMed Scopus (177) Google Scholar = 0.02; SEE = 0.20) in men. The mean interlevel variation was 10.43 for women and 6.73 for men.6Sartoris DJ, Ramos E, Stein JA, Resnick D: Quantitative dual-energy digital projection radiography of the lumbar spine: normative database and interlevel variation. Presented at the Association of University Radiologists 36th Annual Meeting, New Orleans, Louisiana, April 17 to 21, 1988Google Scholar A comparison of DEXA with more conventional densitometric methods in vivo has also been performed at UCSD.7Ho C Sartoris DJ Stein JA Andre M Resnick D Dual-energy projection radiographic bone densitometry: comparison to existing methods.in: Dequeker JV Geusens D Wahner HW Bone Mineral Measurements by Photon Absorptiometry: Methodological Problems. Leuven University Press, Leuven, Belgium1988: 460Google Scholar, 8Sartoris DJ, Stein JA, Ramos E, Lambiase R, Ho C, Andre M, Resnick D: Quantitative dual-energy digital radiography of the spine: comparison to dual-photon absorptiometry and quantitative computed tomography. Presented at the Sixth International Workshop on Bone and Soft Tissue Densitometry, Buxton, Derbyshire, England, September 22 to 25, 1987Google Scholar Dual-energy x-ray results correlated well with prior DPA scans. Among more than 300 patients who underwent both examinations in the lumbar spine, linear regression showed an excellent correlation (r = 0.9845). The regression coefficient was 0.9092, with a negative offset of −0.068 for a net correction factor of about 0.84.9Borders J, Sartoris DJ, Stein JA, Ramos E, Resnick D: Quantitative dual-energy digital projection radiography of the lumbar spine: comparison to quantitative dual-photon absorptiometry in patients. Radiology (in press)Google Scholar In contrast, DEXA results apparently cannot be used as reliable predictors of quantitative computed tomographic trabecular bone densities, although this conclusion is based on a relatively small and heterogeneous group of patients. In a limited study in which images of highly variable quality were used, lateral DEXA images were used to calculate central vertebral body trabecular bone density. The correlation between this approach and quantitative computed tomographic measurements was improved. This relationship could potentially show substantial improvement with increased x-ray exposure and positioning precision.10Borders J, Sartoris DJ, Stein JA, Ramos E, Resnick D: Quantitative dual-energy digital projection radiography of the lumbar spine: comparison to quantitative computed tomography in patients. Invest Radiol (in press)Google Scholar In another study conducted at UCSD, DEXA of the lumbar spine was compared with DPA of the lumbar spine and proximal femur in 62 patients (33 women and 29 men) with rheumatoid arthritis. Of the two vertebral measurements, DEXA correlated less well with femoral data than did DPA in both sexes. A high correlation between DEXA and DPA of the lumbar spine was substantiated. No significant relationship between bone density and age was observed. In comparison with control subjects, bone density in the spine and proximal femur of rheumatoid patients was significantly reduced for both sexes.11O'Malley M, Kenrick AJ, Sartoris DJ, Hochberg AM, Weisman MH, Ramos E, Zvaifler N, Resnick D: Quantitative dual-energy digital projection radiography of the spine versus dual-photon absorptiometry of the spine and proximal femur in rheumatoid arthritis: relationship to clinical parameters. Presented at the Association of University Radiologists 36th Annual Meeting, New Orleans, Louisiana, April 17 to 22, 1988Google Scholar In addition to the studies described, numerous clinical investigations using DEXA are in progress at UCSD. The technology is being used for assessment of patients with chronic renal failure on dialysis or after transplantation as well as for oncology and pediatric patients, including normal subjects and those with risk factors for diminished bone density. Experience to date has suggested broadening horizons of clinical application for DEXA. Regional bone density measurements may ultimately have utility in monitoring disuse osteopenia, reflex sympathetic dystrophy syndrome, and available bone stock for orthopedic procedures. In conclusion, the concept of bone density measurement has yet to gain broad support among clinicians, many of whom question the accuracy and precision of existing methods. DEXA may reverse the tide of opinion, however. This radiographic technique has already demonstrated certain clinical and economic advantages over quantitative computed tomography and DPA. Just as importantly, the new method may restore credibility to bone density measurement. Because the examination time for DEXA is short, performance of bone densitometry in general is easier to justify–whether it is for screening or for longitudinal follow-up. Patients, especially older persons who cannot lie supine for extended periods, can tolerate it better than other lengthier procedures. It is also an ideal technique for children; the radiation dose is low, and the examination time is commensurate with their attention span. We predict that more DEXA evaluations will be done in pediatric patients in the future. In comparison with other methods, DEXA affords imaging of many more patients per hour; thus, it is more economic and attractive to third-party payers. The higher precision renders it more useful to referring physicians, inasmuch as a subtle change in bone density can be observed and believed with greater confidence. This technique may offer a potential for screening that does not exist with quantitative computed tomography or DPA. The concept of screening has been criticized in the literature as being unwarranted with existing methods. DEXA may not be the ideal technique, but it seems much better than previously available procedures. Future research should be directed toward continued improvement in the precision of DEXA and attempts to increase its sensitivity. One realistic possibility with this technology is the capability of doing lateral projections of the spine, which would allow selective evaluation of the vertebral bodies. The fact that they have a relatively higher trabecular bone content in comparison with the posterior elements is important and encourages their independent measurement as a site of greater sensitivity. With this technique, one might expect DEXA to approach quantitative computed tomography in sensitivity. DEXA is being well received among imaging specialists. Systems have been selling enthusiastically in both domestic and international markets. The method is also currently being used in a variety of important research endeavors funded by the National Institutes of Health and other granting agencies. The DEXA technology is an exciting advance in the field of noninvasive bone densitometry. Future prospects include mobile versions of the new equipment, which both Lunar and Hologic are developing. The method represents an important opportunity to salvage the credibility of bone mass measurements. In many respects, it is an answer to the criticisms that have been voiced about the previously existing methods.