| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Special Reports |
1 From the Department of Radiology, University of North Carolina School of Medicine, Chapel Hill, NC (E.D.P.); Center for Statistical Sciences, Brown University, Providence, RI (C.A.G.); Departments of Medical Imaging and Medical Biophysics (M.J.Y.) and Radiology (R.A.J.), University of Toronto, Ontario, Canada; Department of Radiology, Northwestern University, Feinberg School of Medicine, Chicago, Ill (R.E.H.); Department of Medicine, Dartmouth Medical School, Hanover, NH (A.N.A.T.); Department of Population Health Sciences, University of Wisconsin, Madison, Wis (D.G.F.); Department of Radiology, University of California, Los Angeles, Calif (L.W.B.); Department of Radiology, Beth Israel Deaconess Medical Center, Boston, Mass (J.K.B.); Department of Radiology, University of Pennsylvania, Philadelphia, Pa (E.F.C.); Department of Radiology, William Beaumont Hospital, Royal Oak, Mich (M.R.); and Department of Radiology, Emory University, Atlanta, Ga (C.J.D.). Received March 15, 2005; revision requested March 21; revision received April 1; accepted April 5. Supported by NCI grant U01 CA079778 and U01 CA080098. Address correspondence to E.D.P. (e-mail: etta_pisano{at}med.unc.edu).
 |
ABSTRACT |
|---|
© RSNA, 2005
|
INTRODUCTION |
|---|
Digital systems quantify x-ray photons and decouple the process of x-ray photon detection from image display. Digital images can undergo image processing and be displayed in multiple formats, on film or on a monitor. Lesion conspicuity can be affected by these contrast manipulations. Since the steps of image acquisition and display are separated, each can be optimized. In addition, image storage, transmission, and retrieval can be substantially improved and simplified. Software to assist the radiologist in interpreting the images can also be more readily utilized if images are digital at acquisition (3).
Despite this promise, published data on diagnostic accuracy have not yet demonstrated improved performance of digital mammography over screen-film mammography in the screening setting (4). Trials performed for U.S. Food and Drug Administration (FDA) approval were not designed to address the question of the relative accuracy of digital and screen-film mammography in the screening setting. Indeed, by using reader studies with large numbers of positive mammograms drawn from both diagnostic and screening populations, these trials showed only that digital mammography was substantially equivalent to screen-film mammography for FDA-approval purposes. In fact, these studies were required only to show that the area under the receiver operating characteristic (ROC) curve for digital mammography was within 0.1 of that for screen-film mammography in this mixed population. This goal was met, although digital mammography had a lower, but not significantly different, area under the ROC curve. Possible explanations for the lower performance include the potential for bias toward the traditional modality created by the recruitment of patients on the basis of their abnormal screen-film mammograms and the relative lack of experience of radiologist readers in interpreting digital images compared with traditional screen-film mammograms (5,6).
One large screening trial using the Senographe 2000D (GE Medical Systems, Milwaukee, Wis) compared the interpretations of 6736 paired mammograms (digital and screen-film) of 4489 asymptomatic women (7). There was no statistically significant difference in the sensitivity or area under the ROC curve for these mammograms. Fifty cancers were detected in this trial, 42 at imaging and eight as interval malignancies. Of the 42 cancers detected at imaging, 18 were detected with both modalities, 15 were detected only with screen-film mammography, and nine were detected only with digital mammography. While the difference in sensitivity was not statistically significant, the trend was in favor of screen-film over digital systems. Findings of this study did show a statistically significant reduction in recalls and biopsies prompted with digital compared with screen-film systems (7).
Another large screening trial performed in Oslo, Norway (the Oslo I trial), also using the Senographe 2000D unit and with all participants undergoing both digital and screen-film mammography, yielded 31 cancers and showed no statistically significant difference between the modalities in cancer detection rate per 1000 women screened, although digital mammography had significantly lower specificity (8). A subsequent trial, Oslo II, in which 25 263 women were randomized between digital and screen-film mammography showed a small but not statistically significant improvement in cancer detection rates between digital mammography compared with screen-film mammography, with digital mammography having a significantly higher recall rate (9). That trial is ongoing.
A major limitation of these prior screening studies was that they included only one type of digital detector and had insufficient statistical power to help determine if there was a small difference in diagnostic accuracy between digital and screen-film mammography. Since even a small reduction in diagnostic accuracy might yield an increase in breast cancer deaths, and since there are a number of different digital detectors, the National Cancer Institute funded the American College of Radiology Imaging Network (ACRIN) Digital Mammographic Imaging Screening Trial (DMIST) to measure the diagnostic accuracy of digital versus screen-film mammography in a screening population. This study was designed to measure the relative diagnostic accuracy of these two modalities across a larger population of women and across all available digital mammography detectors at the time the study was open for enrollment. If there are enough cancers detected by using each machine type, the study may also provide an estimate of the accuracy of each of the five digital detectors that were included in the trial versus screen-film mammography. The primary aim of the study was to compare the diagnostic performance of digital and screen-film mammography in a prospectively enrolled screening cohort of asymptomatic women across all digital mammography machine types.
|
DIGITAL TECHNOLOGY INCLUDED IN THE TRIAL |
|---|
The technical characteristics of the five digital mammography systems (Table 1) have been described in greater detail elsewhere (9).
|
|
SCREEN-FILM SYSTEMS INCLUDED IN THE TRIAL |
|---|
|
|
|
ACCEPTANCE TESTING AND QUALITY CONTROL PROTOCOL FOR DIGITAL AND SCREEN-FILM EQUIPMENT |
|---|
A single reference physicist, with 7 years of experience in mammography physics, visited all the study sites and inspected each digital system by carrying out specific tests on them. She also instructed the local physicist and the quality control technologists in how to perform the applicable tests. Because of the new technology involved, the physics testing on the digital systems was carried out semiannually rather than annually as required in the United States by the Mammography Quality Standard Act for screen-film systems. During the course of the trial, a physicist remained available by telephone and e-mail to help solve testing problems.
For the digital systems, the programs required daily imaging of the same mammographic accreditation phantom used for accreditation of screen-film systems (10). Images were inspected locally and then were sent in digital form to the ACRIN office in Philadelphia, Pa, where the central data archive was located. Subsequently, they were transmitted to the Physics Core Group in Toronto, where they were evaluated by a physicist. All clinical digital images were also sent to ACRIN for archiving and inspection for obvious artifacts or problems caused by image transmission, which was performed by a radiologic technologist with 18 years experience in medical imaging.
Many of the tests in the digital quality control program, particularly those associated with the x-ray production system (tube output, linearity, half-value layer, focal spot unsharpness), were similar to those performed with screen-film systems. Additional tests were specifically developed to evaluate the performance of the x-ray detector, soft-copy display system, and film printing, where these were used. Where possible, these tests took advantage of the fact that the images were in digital form so that quantitative results on signal levels and noise could be easily obtained. The new tests included the use of a digital mammography phantom (Misty; Sunnybrook & Women's Research Institute, Toronto, Ontario, Canada), which incorporated inserts to evaluate the conspicuity of subtle structures in the presence of image noise in regions of different x-ray attenuation, as well as a test of image contrast. In addition, it provided a measure of the amount of tissue excluded at the chest wall. The testing protocol assessed spatial resolution in terms of modulation transfer function by using a specially designed test tool and an analytic software package (DMistifier; Sunnybrook & Women's Research Institute). Tests of the linearity of the x-ray system and the detector were also performed, and a custom-designed tool was used to evaluate geometric distortion. The dependence of image noise on radiation level was assessed as an index of system performance, and the relative magnitudes of structural versus quantum noise were analyzed. To determine whether there were periodicities in the image noise, the Wiener noise power spectra were measured and analyzed by using the DMistifier software package. In addition to tests of the display system specified by the equipment manufacturers, the compliance of the hard- and soft-copy display systems with the Digital Imaging and Communications in Medicine Grayscale Standard was evaluated.
For the screen-film technical quality control tests, summaries of test results obtained by the local physicist were sent to the central physics core in Toronto. In addition, technical data on breast compression and dose for the first 100 study participants were collected by Dr Edward Hendrick's group at Northwestern University (Chicago, Ill).
Data from the technologists' quality control tests were faxed monthly to the reference physicist for review. Similarly, the physicist's annual testing report was reviewed by the physics core in Toronto. All electronic data obtained by the reference physicist or submitted by the sites to the physics core were maintained in a database (Microsoft Access; Microsoft, Redmond, Wash). A complete description of the technical quality control program used in DMIST will be published elsewhere.
|
TRAINING FOR PROTOCOL COMPLIANCE |
|---|
At the training meetings, the protocol was described in detail, including the rationale for the different aspects of compliance with the protocol's features. This included tutorials for the radiologists about the use of interpretation scales and how these scales would be used for analysis purposes, with examples of how different mammograms should be scored by the readers depending on the interpreter's individual impression of the likelihood for malignancy. There were two such meetings for radiologists and research associates prior to the opening of the trial to accrual. The lead radiologist at each site was expected to train those radiologists who could not attend these meetings, and written materials were provided to them for that purpose. ACRIN staff and the principal investigator (E.D.P.) were available, as needed, to respond to queries regarding protocol-compliance issues.
In addition, before a site was opened to accrual, a radiology technologist with 10 years of experience in breast imaging and a MPH degree-level research associate with 10 years of experience in clinical research visited every site the first 2 days that subjects were enrolled in the trial at that site. They reviewed all aspects of the protocol with the lead radiologist and the site's lead research associate. They inspected the facility to ensure that it was set up to comply with the protocol. They demonstrated the use of the online data entry system and then observed and assisted in recruitment of the first few subjects into the trial.
Subject Enrollment
Thirty-three academic and community practice participating sites in the United States and Canada completed accrual of 49 528 women during 25.5 months of accrual from October 2001 to November 2003. Two sites (Elizabeth Wende Breast Clinic, Rochester, NY, and the Mount Sinai Hospital, New York, NY) had two different digital mammography systems available. Women presenting to those sites were assigned randomly to imaging with one of the two available digital mammography systems. All other women were imaged by using the digital mammography system available at the site where they presented for screening.
Women were eligible to participate if they presented for screening mammography at the participating institutions. Women were excluded from the study if they complained of a focal dominant lump or a bloody or clear nipple discharge, had breast implants, were pregnant or believed they might be pregnant, could not undergo follow-up screen-film mammography at the participating institution or could not provide mammograms from another institution for review 1 year after study entry, or had a history of breast cancer treated with lumpectomy. After being informed about the risks and benefits of the study, including the increased radiation exposure of undergoing two screening mammograms at the same visit, participants signed a study-specific informed consent form at each site and were entered into the trial. Our study was approved by the Institutional Review Board of the ACRIN and at each participating site and by the Institutional Review Board and the Cancer Therapy Evaluation Program at the National Cancer Institute. These bodies were informed of the lack of FDA approval for some of the equipment being tested in the study. The study was monitored by an independent Data Safety and Monitoring Board, which received interim analyses of data to ensure that the study would be terminated early if indicated by trends in the outcomes. The study was compliant with the Health Insurance Portability and Accountability Act of 1996.
No company provided financial support for the completion of the study, although the Fuji and Lorad equipment was provided at the expense of those companies. While several of the coauthors (M.J.Y. for Fischer Imaging and GE Medical Systems, C.J.D. and R.E.H for GE Medical Systems) of this article have served as consultants for digital mammography equipment manufacturers, data were centrally controlled and analyzed by nonconsultant coauthors (E.D.P., C.A.G.) so that the consultants involved could not directly influence study outcomes.
To determine whether the study population was indeed representative of the entire screening population at that institution, for 2 weeks the sites collected demographic information without identifiers on all eligible women. Table 4 provides a list of the participating sites, the digital machine type, and the number of participants enrolled for each machine at each site.
|
|
IMAGING PROTOCOL |
|---|
All study participants underwent two-view mammography (both craniocaudal and mediolateral oblique views) of both breasts, or of one breast if they had undergone prior mastectomy, by using both digital and screen-film systems. For large-breasted women, as many craniocaudal and mediolateral oblique views were obtained as were deemed necessary by the technologist to include the entire breast in the screening examination, according to standard clinical practice.
The same technologist performed both screen-film and digital mammography by using as close to the same positioning, dose, and degree of compression for the two modalities as possible. Automatic exposure control systems were used for digital mammography systems, when available.
|
IMAGE INTERPRETATION |
|---|
There were a total of 153 radiologists who interpreted digital and screen-film mammograms as part of DMIST, from two to eight readers participating at each site, with each radiologist reading approximately equal number of digital and screen-film mammograms. All readers had at least 8 hours of digital mammography training, including in the use of soft-copy display, if applicable, prior to participation as readers in DMIST. In addition, all U.S. readers met qualifications as mammography interpreters under the U.S. Mammography Quality Standard Act. Canadian radiologists met the equivalent standards. No other special qualifications were required. The goal was to allow participation in the study by a population of readers who would ordinarily interpret mammograms in the course of their usual clinical work, not just those who were fellowship-trained breast imagers, or individuals who had received special training in the interpretation of either digital or screen-film mammograms.
Interpretation conditions were standardized across digital mammography systems as much as possible so that, in general, all readers at all locations read the mammograms obtained with mammography units of the same manufacturer in the same fashion. The reading modality was determined by agreement between reporting radiologists and manufacturer's suggestion. All Fuji and Lorad/Trex digital mammograms were read on hard copy only, since those units did not have soft-copy options when the trial began. Fischer Imaging digital mammograms were read in both hard- and soft-copy format at all institutions. GE Medical Systems digital mammograms were read in soft-copy format only. Hologic digital mammogram interpretations could not be standardized since not all institutions had the same display systems. Those institutions with Hologic soft-copy systems used only soft-copy display systems, and those institutions without Hologic soft-copy display systems used only printed film display. Comparison with prior examinations was made by using film hung on a nearby view box for both digital and screen-film examinations or by displaying prior digital soft-copy studies on a nearby monitor according to local protocols.
All mammograms were interpreted by using four independent scales. The radiologists estimated the probability of malignancy (from 0% to 100%) and also used American College of Radiology Breast Imaging Reporting and Data System (BI-RADS) assessment categories 0 through 5 (11). In addition, the radiologists scored the digital and screen-film mammograms by using an ordinal categoric malignancy scale with the following categories: score of 1, the finding is definitely not malignant; score of 2, the finding is almost certainly not malignant; score of 3, the finding is probably not malignant; score of 4, the finding is possibly malignant; score of 5, the finding is probably malignant; score of 6, the finding is almost certainly malignant; and score of 7, the finding is definitely malignant.
Finally, to assess where on the seven-point scale the individual radiologist's cut point between positive and negative examination findings actually was, the radiologists provided a score regarding the need for callback after screening mammography, as follows: score of 1, there is no evidence that the patient should be called back for diagnostic work-up; score of 2, there is some evidence of an abnormality but it is insufficient to justify that the patient should be called back for diagnostic work-up; score of 3, there is sufficient evidence to justify that the patient should be called back for diagnostic work-up; score of 5, there is strong evidence to justify that the patient should be called back for diagnostic work-up; and score of 6, there is overwhelming evidence to justify that the patient should be called back for diagnostic work-up.
Readers were also asked for their recommendation regarding the need for further work-up for all subjects, regardless of the other scores assigned, including the BI-RADS assessment categories (11). Findings and breast density were coded by all readers using standard BI-RADS descriptors (11), and each individual finding in each breast was assigned a probability of malignancy (0–100%) and was scored according to the seven-point malignancy and the five-point call-back scales. The digital and screen-film mammograms were interpreted prior to the interpretation of any follow-up images.
Radiologist readers at each site were familiarized with the interpretation scales used in DMIST through the training sessions described earlier and through written materials provided to them prior to their participation in the study. The scales themselves were developed on the basis of on extensive consultation with experts in the field of ROC curve analysis and the available literature on this topic (12,13).
|
WORK-UP AFTER AN ABNORMAL MAMMOGRAM |
|---|
|
DETERMINATION OF TRUTH REGARDING BREAST CANCER STATUS |
|---|
|
DATA COLLECTION AND MANAGEMENT |
|---|
|
STATISTICAL CONSIDERATIONS FOR PRIMARY AIM |
|---|
A total accrual of 49 500 participants was planned for this study. The primary determinant of the sample size was the need to accrue an adequate number of participants with cancer in this screening cohort. A cancer rate of five per 1000 women was expected in the study cohort. On the basis of prior experience and the available information in 1999, it was assumed that the area under the ROC curve for screen-film mammography would be about 0.75. The overall sample size was chosen so that if a two-sided test at a significance level of .05 was used to compare the ROC areas between the two modalities, the power to detect a difference of .06 would be at least 80% and the power to detect a difference of .07 would be at least 90%. To account for the paired nature of the design, it was assumed that the average correlation of the degree of suspicion with digital and screen-film mammography in cancer and noncancer cases will be about 0.4 (14). The final number of 49 500 was derived by adjusting the sample size to allow for the possibility of subjects with incomplete or otherwise unusable data. The robustness of the design to prior assumptions was examined by varying key parameters and assessing their impact on statistical power.
|
QUALITY OF LIFE SUBSTUDY |
|---|
At 22 participating sites, a stratified total random sample of 600 women who had been recommended for further work-up as a result of a screening mammogram (either digital or screen-film) and age- and site-matched, randomly sampled, total control group of 600 women with negative screening mammograms were surveyed twice by trained research assistants by a telephone interview. The first interview was conducted shortly after the screening mammogram, and the second was conducted approximately 12 months after DMIST enrollment. The survey timing is planned with the intent of initially interviewing women with positive mammograms after notification that additional work-up is necessary but prior to its completion. The timing of the second survey allows the full work-up experience to be reported.
To assess whether health-related quality of life and anxiety for women prior to the screening event differ from these measures 1 year after screening, both surveys were also administered to the first 42 women enrolled in the study at each participating site prior to their screening mammogram.
Survey instruments consisted of three main parts: the EuroQol (EQ-5D), an existing generic instrument for measurement of the quality of life (15), the Spielberger State-Trait Anxiety Inventory (STAI Y-6), a six-item short form of the State Scale (16), and a resource utilization patient questionnaire developed for this study.
The EQ-5D contains five questions, one each about mobility, self-care, usual activities, pain and discomfort, and anxiety and/or depression. Each question has three possible answers categorizing that particular aspect of health. Along with the EQ-5D, women assess their overall health as excellent, very good, good, fair, or poor and complete a rating scale for current health. The STAI Y-6 contains six questions that measure patient anxiety. The patient questionnaire portion of the survey, which is administered only at the follow-up interview, records information about the breast-related care women undergo following their initial screening mammograms (see Cost-Effectiveness Analysis for details on resource utilization questions). Women are asked about the anxiety and/or concern and discomfort associated with their breast-related care. In addition to information on resource utilization, the patient questionnaire includes five questions regarding attitudes toward future screening. One question asks women about an imaginary mammogram that is just as good at depicting cancer but results in fewer false-positives. Women are also asked about their willingness to travel an additional amount of time to obtain the new mammogram. Another question asks women to imagine two new types of mammograms, both as accurate as mammograms of today. One results in fewer false-positives and the other requires less breast compression. Women are asked which new type of mammogram they would choose.
|
COST-EFFECTIVENESS ANALYSIS |
|---|
These data provide the basis for estimating work-up costs following a screening mammogram. These cost estimates, along with the estimated effect of false-positive screening examinations on the quality of life (from the Quality of Life substudy), are combined with main trial end points concerning sensitivity and specificity of digital and screen-film mammography by using the Wisconsin Breast Cancer Epidemiology Simulation Model (17). This model is used to estimate the costs and quality-adjusted life years that would have accrued in the U.S. population for the decade 1991–2000, with the assumption that digital mammography had been used (under usual screening patterns and under assumed complete screening participation) throughout that period versus conventional screen-film mammography. This retrospective analysis allows comparison of costs and effects without needing to project results to an uncertain future. If digital mammography is both more costly and more effective than screen-film mammography, then the incremental cost-effectiveness ratio for digital mammography relative to screen-film mammography will be reported as the added cost per quality-adjusted life-year gained.
|
READER STUDIES USING THE DMIST IMAGE ARCHIVE |
|---|
Soft-Copy Digital, Printed Digital, and Screen-Film Comparison Studies
Four separate studies were conducted to assess the relative diagnostic accuracy of four digital machine types (Fischer Imaging, Fuji Medical Systems, GE Medical Systems, and Hologic) versus screen-film mammography. Twelve radiologist readers participated in each of these studies and interpreted the mammograms from all cases in each of the three formats: soft-copy digital, printed digital, and screen-film. The order of completion of each of the case sets was randomly assigned. A minimum of 6 weeks were planned between readings in each format so that readers would not recall the mammograms or their prior interpretations of them.
Density Study
This reader study was designed to assess the effect of breast density on diagnostic accuracy of soft-copy digital versus screen-film mammography. A predetermined mixture of overall breast densities (fatty, scattered fibroglandular, heterogeneously dense, extremely dense) is included in the set of cases for this study. Eight readers were assigned to interpret the cases by using either soft-copy digital or screen-film display.
Prevalence Study
This reader study was designed to examine whether a reader's perception of the prevalence of disease affects diagnostic performance. To achieve this goal, reader accuracy was tested by varying the percentage of cancers in the case mix to be interpreted by the readers.
A total of 400 cases were used in this reader study. Each of the 30 readers will be assigned to read a total of 300 cases of the 400 as either soft-copy digital or screen-film format. The case set for each reader will be at one of three randomly assigned cancer prevalence rates.
|
SUMMARY |
|---|
In fact, there is controversy about the appropriate follow-up period for mammograms before they can be deemed truly negative. Estimates of the diagnostic accuracy of mammography vary substantially as the follow-up interval is varied (25). The Breast Cancer Surveillance Consortium, a National Cancer Institute-funded group that studies mammography screening outcomes, and the American College of Radiology BI-RADS system audits use a 1-year interval as the follow-up period that defines the truth status for the mammograms (11,26). That is, if a cancer manifests within 365 days of a negative screening mammogram, the mammogram is considered false-negative. Likewise, a positive screening mammogram would be considered true-positive in such a case. The decision to define the follow-up interval for DMIST at 11–15 months was based on the desire to allow comparisons with the published performance of screening mammography by the Breast Cancer Surveillance Consortium and others and the available resources. Of course, once the data are collected up to a specific length of time after the screening event, the results can be reported across a variety of follow-up intervals up to that length of time (eg, 11, 12, and 15 months).
Other planned DMIST secondary analyses include an assessment of the effects of study subject characteristics and digital machine type on diagnostic accuracy. The demographic factors to be studied will include age, lesion type (ie, mass, clustered calcifications, architectural distortion), pathologic diagnosis, menopausal and hormonal status, breast density, and family history. This will yield substantial information regarding the effectiveness of screening with both digital and screen-film mammography across various population subsets. In addition, the cost-effectiveness and quality of life substudies and the reader studies may provide the answers to many other important questions that can influence the potential adoption of digital mammography, including providing better performance estimates for the individual digital mammography machines and for soft-copy versus printed film display.
Results of the primary DMIST study comparing digital to screen-film mammography should be available in 2005, after the follow-up period for all study participants and data analysis have occurred. Results of secondary studies, including the reader studies, will be analyzed after the primary study.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: ACRIN = American College of Radiology Imaging Network • BI-RADS = Breast Imaging Reporting and Data System • DMIST = Digital Mammographic Imaging Screening Trial • FDA = Food and Drug Administration • ROC = receiver operating characteristic
M.J.Y. was a consultant to Fischer Imaging and GE Medical Systems. C.J.D. and R.E.H were consultants to GE Medical Systems
Author contributions: Guarantors of integrity of entire study, E.D.P., C.A.G; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, E.D.P., C.A.G., R.E.H.; clinical studies, E.D.P., C.A.G., M.J.Y., R.E.H., L.W.B., J.K.B., E.F.C. R.A.J., M.R., C.J.D.; experimental studies, E.D.P., C.A.G. M.J.Y. L.W.B., M.R., C.J.D.; statistical analysis, C.A.G.; and manuscript editing, all authors
|
References |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. S. The, K. J. Schilling, J. W. Hoffmeister, E. Friedmann, R. McGinnis, and R. G. Holcomb Detection of Breast Cancer with Full-Field Digital Mammography and Computer-Aided Detection Am. J. Roentgenol., February 1, 2009; 192(2): 337 - 340. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. J. Heine, M. J. Carston, C. G. Scott, K. R. Brandt, F.-F. Wu, V. S. Pankratz, T. A. Sellers, and C. M. Vachon An Automated Approach for Estimation of Breast Density Cancer Epidemiol. Biomarkers Prev., November 1, 2008; 17(11): 3090 - 3097. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. J. Hillman and C. A. Gatsonis When Is the Right Time to Conduct a Clinical Trial of a Diagnostic Imaging Technology? Radiology, July 1, 2008; 248(1): 12 - 15. [Full Text] [PDF]
|
|
|
|
|
|
R. E. Hendrick, E. B. Cole, E. D. Pisano, S. Acharyya, H. Marques, M. A. Cohen, R. A. Jong, G. E. Mawdsley, K. M. Kanal, C. J. D'Orsi, et al. Accuracy of Soft-Copy Digital Mammography versus That of Screen-Film Mammography according to Digital Manufacturer: ACRIN DMIST Retrospective Multireader Study Radiology, April 1, 2008; 247(1): 38 - 48. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. W. Fletcher, S. M. Kymes, M. Gould, N. Alazraki, R. E. Coleman, V. J. Lowe, C. Marn, G. Segall, L. A. Thet, K. Lee, et al. A Comparison of the Diagnostic Accuracy of 18F-FDG PET and CT in the Characterization of Solitary Pulmonary Nodules J. Nucl. Med., February 1, 2008; 49(2): 179 - 185. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. D. Pisano, R. E. Hendrick, M. J. Yaffe, J. K. Baum, S. Acharyya, J. B. Cormack, L. A. Hanna, E. F. Conant, L. L. Fajardo, L. W. Bassett, et al. Diagnostic Accuracy of Digital versus Film Mammography: Exploratory Analysis of Selected Population Subgroups in DMIST Radiology, February 1, 2008; 246(2): 376 - 383. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. N.A. Tosteson, N. K. Stout, D. G. Fryback, S. Acharyya, B. A. Herman, L. G. Hannah, E. D. Pisano, and for the DMIST Investigators Cost-Effectiveness of Digital Mammography Breast Cancer Screening Ann Intern Med, January 1, 2008; 148(1): 1 - 10. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. D. Pisano, R. E. Hendrick, M. Yaffe, E. F. Conant, and C. Gatsonis Should Breast Imaging Practices Convert to Digital Mammography? A Response from Members of the DMIST Executive Committee Radiology, October 1, 2007; 245(1): 12 - 13. [Full Text] [PDF]
|
|
|
|
|
|
S. K. Yang, W. K. Moon, N. Cho, J. S. Park, J. H. Cha, S. M. Kim, S. J. Kim, and J.-G. Im Screening Mammography-detected Cancers: Sensitivity of a Computer-aided Detection System Applied to Full-Field Digital Mammograms Radiology, July 1, 2007; 244(1): 104 - 111. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Gur Large Field Trial for Lung Cancer Screening: Putting the Wrong Cart before the Horse?--Commentary Radiology, May 1, 2007; 243(2): 317 - 318. [Full Text] [PDF]
|
|
|
|
|
|
Y. Jiang, D. L. Miglioretti, C. E. Metz, and R. A. Schmidt Breast Cancer Detection Rate: Designing Imaging Trials to Demonstrate Improvements Radiology, May 1, 2007; 243(2): 360 - 367. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. M Kymes, K. Lee, J. W Fletcher, and SNAP (CSP 027) Study Group Assessing diagnostic accuracy and the clinical value of positron emission tomography imaging in patients with solitary pulmonary nodules (SNAP) Clinical Trials, February 1, 2006; 3(1): 31 - 42. [Abstract] [PDF]
|
|
|
|
 |
|
 E. D. Pisano, C. Gatsonis, E. Hendrick, M. Yaffe, J. K. Baum, S. Acharyya, E. F. Conant, L. L. Fajardo, L. Bassett, C. D'Orsi, et al. Diagnostic Performance of Digital versus Film Mammography for Breast-Cancer Screening N. Engl. J. Med., October 27, 2005; 353(17): 1773 - 1783. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
TOP
ABSTRACT
INTRODUCTION