Impact of the number of readers on mammography interpretation

Purpose: To evaluate the impact of the number of readers on sensitivity and specificity, and compare it with conference consensus reading.

Material and Methods: Eight readers read mammography films of 200 women (including 35 false-negative and 16 screen-detected cancers). The sensitivities and specificities of the two methods were calculated: either at least a single cancer-positive opinion within the group (summarized independent reading) or the cancer-positive opinion of the reader majority (conference consensus reading) was considered decisive.

Results: The mean sensitivity for summarized independent readings of different groups was 64.7% as compared to the 43.1% mean sensitivity of conference consensus readings. The mean specificities were 92.4% and 97.7%, respectively. The greatest sensitivity of 74.5% was achieved when the readings of the four best-performing readers were combined.

Conclusion: The sensitivity of reading is maximal when any positive opinion within a pair or a group of readers is taken into consideration. Discordant double reading may best be judged as screening positive, and the value of a third opinion should be questioned.     

Factors affecting radiologist inconsistency in screening mammography

RATIONALE AND OBJECTIVES: Although research has successfully documented variability in radiologists' interpretation of mammograms, it has failed to determine the relative contributions of case-specific features and reader inconsistency. Training interventions to improve consistency will be ineffectual if they do not target the principal determinants of disagreement among radiologists. The current study assessed the relative contributions of the case and the interpreter to the problem of inconsistent interpretation. MATERIALS AND METHODS: One hundred ten radiologists independently interpreted mammograms from the same 148 screening cases (43% with biopsy-proved cancers) and reported the presence or absence of calcifications, mass, architectural distortion, and asymmetric density in each of 296 breasts. The radiologists were blinded to disease status (established at biopsy or follow-up). RESULTS: Case-related differences accounted for a greater proportion of interpretation disagreement than did differences between interpreters. The presence of cancer was associated with increased disagreement, perhaps because of the multiplicity of findings. Patient age was also associated with increased disagreement in the reporting of calcifications. CONCLUSION: For screening mammography, increased consistency between radiologists in their recognition and reporting of clinically important findings will best be achieved by reducing disagreement in difficult cases. Current training in the United States addresses difficult cases only as they have been defined intuitively or experientially. The authors' population-based method provides an objective metric to measure case difficulty and basis from which to identify difficult cases for targeted training.     

数字乳腺断层摄影及其合成二维影像技术在乳腺癌筛查中的研究进展

目前乳腺X线检查仍是乳腺癌早期诊断的有效检查方法之一,主要包括全视野数字化乳腺摄影（FFDM）、数字乳腺断层摄影（DBT）、合成乳腺X线摄影（SM）以及3种技术的联合应用（FFDM联合DBT、SM联合DBT）。对DBT、SM和SM联合DBT在乳腺筛查中诊断效能、影像质量及辐射剂量等进行比较。SM联合DBT可有效平衡辐射剂量和诊断效能,但仍然在判读时间、信息的存储与传输和检查成本方面存在局限性。就以上3种检查技术在乳腺癌筛查中的研究进展予以综述。     

Computer-aided detection in mammography: an assessment of performance on current and prior images

RATIONALE AND OBJECTIVES: The authors assessed and compared the performance of a computer-aided detection (CAD) scheme for the detection of masses and microcalcification clusters on a set of images collected from two consecutive ("current" and "prior") mammographic examinations. MATERIALS AND METHODS: A previously developed CAD scheme was used to assess two consecutive screening mammograms from 200 cases in which the current mammogram showed a mass or cluster of microcalcifications that resulted in breast biopsy. The latest prior examinations had been initially interpreted as negative or definitely benign findings (Breast Imaging Reporting and Data System rating, 1 or 2). The study involved images of 400 examinations acquired in 200 patients. Radiologists identified 172 masses and 128 clusters of microcalcifications on the current images. The performance of the CAD scheme was analyzed and compared for the current and latest prior images. RESULTS: There were significant differences (P < .01) between current and prior images in many feature values. The performance of the CAD scheme was significantly lower for prior than for current images (P < .01). At 0.5 and 0.2 false-positive mass and cluster cues per image, the scheme detected 78 malignant masses (78%) and 63 malignant clusters (80%) on current images. Only 42% of malignant cases were detected on prior images, including 40 masses (40%) and 36 microcalcification clusters (46%). CONCLUSION: CAD schemes can detect a substantial fraction of masses and microcalcification clusters depicted on prior images. To improve performance with prior images, the scheme may have to be adaptively reoptimized with increasingly more subtle abnormalities.     

数字乳腺X线摄影技术的研究进展

随着数字化乳腺X线检查技术的发展,屏/片系统乳腺X线摄影(SFM)技术已逐步被取代,一些技术改进的优势已在致密型乳腺的女性中得到证实。比较数字乳腺体层摄影(DBT)技术、对比增强双能数字乳腺X线摄影(CEDM)技术及常规数字乳腺X线摄影(DM)技术,并综述数字乳腺摄影技术发展中一些新技术的特点,分析其优势、局限性及对病人的潜在影响等,从而预测它们的发展前景。     

"Memory effect" in observer performance studies of mammograms

RATIONALE AND OBJECTIVE: To evaluate breast radiologists' recognition of mammograms showing cancers that they correctly detected or "missed" during clinical interpretations. MATERIALS AND METHODS: Two similar experiments were conducted. In the first, 33 bilateral screening mammograms were reviewed by four breast imagers. These included five cancers that each radiologist had detected, two cancers that each radiologist had "missed," and five mammograms recalled by other radiologists that were not cancer. Radiologists were asked if they had interpreted the mammogram in clinic and if the mammogram was suspicious for cancer. In the second experiment, four different breast imagers reviewed 48 mammograms that included five cancers that each radiologist had detected, two cancers that each radiologist had "missed," and five mammograms that were recalled by each radiologist but were not cancer. Using chi-square analysis, the performance of the radiologists on screening mammograms they had read in clinic was compared with their performance on mammograms read in clinic by other radiologists. RESULTS: Seven of eight radiologists did not remember interpreting any of the mammograms in clinic. One radiologist correctly remembered interpreting one mammogram in clinic, but interpreted it incorrectly. Average performance showed no significant difference (P = .60) between mammograms they had interpreted in clinic and those interpreted by others. CONCLUSION: Radiologists do not remember most mammograms showing cancer that they have interpreted, either correctly or incorrectly, after they are mixed with mammograms showing cancer that were interpreted by other radiologists. Screening mammograms can be used in observer performance studies in which the interpreting radiologist participates as an observer.     

数字化乳腺摄影与普通钼靶摄影在小乳癌诊断中的对比分析

目的：探讨数字化乳腺摄影在小乳癌诊断中的优越性。方法：经手术病理证实的小乳癌57例，采用数字化乳腺摄影及普通钼靶摄影方法。结果：数字化乳腺摄影诊断51例，其敏感性、特异性、准确性分别为89．4％、95．O％、90.9％；普通钼靶摄影术前诊断46例，其敏感性、特异性、准确性分别为80．7％、85．O％、81．8％。结论：在小乳癌诊断中。数字化乳腺摄影在敏感性、特异性、准确性方面均优于普通钼靶摄影。     

Breast cancer detection: Comparison of digital mammography and digital breast tomosynthesis across non-dense and dense breasts

IntroductionBreast density is associated with an increase in breast cancer risk and limits early detection of the disease. This study assesses the diagnostic performance of mammogram readers in digital mammography (DM) and digital breast tomosynthesis (DBT).MethodsEleven breast readers with 1–39 years of experience reading mammograms and 0–4 years of experience reading DBT participated in the study. All readers independently interpreted 60 DM cases (40 normal/20 abnormal) and 35 DBT cases (20 normal/15 abnormal). Sensitivity, specificity, ROC AUC, and diagnostic confidence were calculated and compared between DM and DBT.ResultsDBT significantly improved diagnostic confidence in both dense breasts (p = 0.03) and non-dense breasts (p = 0.003) but not in other diagnostic performance metrics. Specificity was higher in DM for readers with >7 years' experience (p = 0.03) in reading mammography, non-radiologists (p = 0.04), readers who had completed a 3–6 months training fellowship in breast imaging (p = 0.04), and those with ≤2 years’ experience in reading DBT (p = 0.02), particularly in non-dense breasts.ConclusionDiagnostic confidence was higher in DBT when compared to DM. In contrast, other performance metrics appeared to be similar or better with DM and may be influenced by the lack of experience of the reader cohort in reading DBT.Implications for practiceThe benefits of DBT may not be entirely accrued until radiologists attain expertise in DBT interpretation. Specificity of DBT varied according to reader characteristics, and these characteristics may be useful for optimising pairing strategies in independent double reading of DBT as practiced in Australia to reduce false positive diagnostic errors.     

DIMA enlargement mammography in microcalcifications: a prospective study with ROC analysis

The purpose of this study was to investigate whether the four-fold magnification mammography (direct magnification, DIMA) technique would perform better than conventional 1.5-fold magnification mammography in the differentiation of breast microcalcifications into benign and malignant. Fifty patients with non-palpable microcalcifications detected by mammography were examined immediately prior to surgical biopsy using both a conventional (1.5-fold) and the DIMA (fourfold) magnification mammography techniques. The microcalcifications were classified by five experienced radiologists using morphological criteria. A receiver operating characteristics curve (ROC) analysis of the sensitivity and specificity of both techniques in assessing malignancy was then carried out. The DIMA mammography technique was slightly but non-significantly superior to the conventional method in detecting malignancy (p > 0.05). Coarse granular and pleomorphic calcifications were detected more frequently with the DIMA technique. Coarse calcifications were significantly more frequently associated with histologically benign findings, whereas fine granular calcifications were significantly more likely to be malignant lesions. Assessment of malignancy associated with microcalcifications using morphological criteria is not significantly improved by mammography techniques with higher magnification.     

Interpolation algorithms for digital mammography systems with multiple detectors

RATIONALE AND OBJECTIVES: In some full-field digital mammography systems, multiple detectors are abutted together, and the physical gaps between adjacent detectors produce seams between the resultant subimages. In this study, a variety of interpolation algorithms for estimating the missing information in the seams were compared, and their effect on image quality was evaluated. MATERIALS AND METHODS: Eight representative interpolation algorithms were selected, including nearest neighbor, one-dimensional and two-dimensional weighting, mean value, one-dimensional and two-dimensional polynomial, and one-dimensional and two-dimensional cubic spline interpolation methods. These methods were applied to digital mammograms and phantom images. The effectiveness of each algorithm was evaluated for accuracy and geometric distortion. RESULTS: These interpolation algorithms offered similar accuracy in estimating missing image information. The weighting, polynomial, and cubic spline interpolation algorithms introduced less geometric distortion than the nearest neighbor and mean value interpolation algorithms. All algorithms were more effective in estimating larger, lower-contrast features (such as breast masses) than in estimating smaller, higher-contrast features (such as breast microcalcifications). Small microcalcifications within the seams cannot be recovered with interpolation. The probability of a microcalcification in a seam is small, however, and the failure to image a few microcalcifications of a cluster generally does not substantially alter diagnostic performance. CONCLUSION: In the development of full-field digital breast imaging systems, appropriate interpolation algorithms can satisfactorily fill in narrow gaps between adjacent detectors. The one-dimensional weighting interpolation method seems an effective and efficient choice.     

MR mammography in the pre-operative staging of breast cancer in patients with dense breast tissue: comparison with mammography and ultrasound

The aim of this study was to determine whether pre-operative MR mammography could predict the extent of breast cancer in patients with dense breasts or whether dense parenchyma will lead to false-positive or inconclusive examinations. Sixty-seven patients with dense breasts with a malignant breast tumor planned for conservative surgery were reviewed. Detection rates of mammography, ultrasound, and MR mammography were studied, and the diameters of the lesions were measured and compared with pathological examination. Pathology revealed breast cancer in 65 patients. Sensitivity for detection of index lesions was 83% for mammography, 70.8% for ultrasound, and 98% for MR mammography. Mammography underestimated tumor extent in 37%, ultrasound in 40%, and MR in 12.5%. Of the 20 patients (31%) with multifocal or multicentric carcinoma, mammography detected the lesions in 35%, ultrasound in 30%, and MR in 100%, with a false-positive rate of 12.5, 14, and 23%. The MR mammography is more accurate in assessing tumor extent and multifocality in patients with dense breasts, but benign changes may lead to false-positive examinations.     

Observer variability in screen-film mammography versus full-field digital mammography with soft-copy reading

Full-field digital mammography (FFDM) with soft-copy reading is more complex than screen-film mammography (SFM) with hard-copy reading. The aim of this study was to compare inter- and intraobserver variability in SFM versus FFDM of paired mammograms from a breast cancer screening program. Six radiologists interpreted mammograms of 232 cases obtained with both techniques, including 46 cancers, 88 benign lesions, and 98 normals. Image interpretation included BI-RADS categories. A case consisted of standard two-view mammograms of one breast. Images were scored in two sessions separated by 5 weeks. Observer variability was substantial for SFM as well as for FFDM, but overall there was no significant difference between the observer variability at SFM and FFDM. Mean kappa values were lower, indicating less agreement, for microcalcifications compared with masses. The lower observer agreement for microcalcifications, and especially the low intraobserver concordance between the two imaging techniques for three readers, was noticeable. The level of observer agreement might be an indicator of radiologist performance and could confound studies designed to separate diagnostic differences between the two imaging techniques. The results of our study confirm the need for proper training for radiologists starting FFDM with soft-copy reading in breast cancer screening. Presented at ECR, Wien 2006.     

Follow-up and final results of the Oslo I Study comparing screen-film mammography and full-field digital mammography with soft-copy reading

Purpose: To compare cancer detection rates of screen-film (SFM) and full-field digital mammography (FFDM) with soft-copy reading in a screening program including the initial positive scores for interval cancers and cancers in the subsequent screening round, and to analyze the false-negative FFDM interpretations.

Material and Methods: Using a paired study design, 3683 women underwent SFM and FFDM in a population-based screening program. Two standard views of each breast were acquired. The images were interpreted without previous films for comparison. Independent double reading using a 5-point rating scale for probability of cancer was used for each modality. An examination was defined as positive if at least one of the two independent readers scored 2 or higher on the 5-point rating scale. SFM-positive cases were discussed in a SFM consensus meeting and FFDM-positive cases in a separate FFDM consensus meeting before recall. The study population was followed for more than 2 years so that interval cancers and screen-detected cancers in the subsequent screening round could be included. Cancer detection rates were compared using the McNemar test for paired proportions. The kappa statistic and Wilcoxon signed-rank test for matched pairs were used for comparing rating scores. The reading time was recorded for all FFDM interpretations.

Results: A total of 31 cancers (detection rate 0.84%) were diagnosed initially, of which SFM detected 28 and FFDM 23 (McNemar test P = 0.23, discordant pair 8 and 3). Two cancers with a positive score at initial SFM reading and three with a positive score at initial FFDM reading were dismissed at SFM and FFDM consensus meetings, respectively. The difference in cancer detection after recall (discordant pair 11 and 5) was not significant (McNemar test, P = 0.21). Of the 10 interval cancers and 16 screen-detected cancers in the subsequent round, 3 had true-positive SFM scores while 4 had true-positive FFDM scores in the initial reading session. A total of 38 cancers therefore had a positive result at double reading at one or both modalities, 31 at SFM and 27 at FFDM (McNemar test, P = 0.48). Comparison of SFM and FFDM interpretations using the mean score for each case revealed no statistically significant difference between the two modalities (Wilcoxon signed-rank test for matched pairs; P-value = 0.228). Two initial round cancers (one tumor found incidentally at work-up for a mass proved to be a simple cyst with a positive score at FFDM but a negative score at SFM, and one tumor with positive score at SFM but negative score at FFDM due to positioning failure) were excluded from the further analysis. Excluding these two cancers from comparison, there were 31% (22 of 72) false-negative SFM and 47% (34 of 72) false-negative FFDM individual interpretations. The overall mean interpretation time for normal FFDM examinations was 45 s. For most false-negative FFDM results, the reading time was shorter or longer than for normal examinations. The recorded FFDM interpretation time was noticeably short for several overlooked cancers manifesting as microcalcifications (ductal carcinoma in situ).

Conclusion: There is no statistically significant difference in cancer detection rate between SFM and FFDM with soft-copy reading in a mammography screening program. Analysis of cancers missed at FFDM with soft-copy reading indicates that close attention has to be paid to systematic use of image display protocols.     

Clinical evaluation of wavelet-compressed digitized screen-film mammography

RATIONALE AND OBJECTIVES: The authors compared diagnostic accuracy and callback rates with conventional screen-film mammograms and wavelet-compressed digitized images. MATERIALS AND METHODS: Sixty sets of mammograms (four views per case) were digitized at a spatial resolution of 100 microm. The images were wavelet compressed to a mean compression ratio of 8:1 and reviewed by three mammographers. Five regions were evaluated in each breast. Suspicion of malignancy was graded on a scale of 0% to 100%, and receiver operating characteristic (ROC) analysis was performed. Callback rates were calculated by using the American College of Radiology's Breast Imaging Reporting and Data System lexicon scale. RESULTS: The mean diagnostic accuracy with compressed and conventional images was 0.832 and 0.860, respectively. The upper 95% confidence bound for the difference in ROC areas was 0.061. The mean false-positive rate at a fixed sensitivity of 0.90 was 0.041 for compressed images and 0.059 for conventional images. The mean callback rates for normal, benign, and malignant regions were 0.023, 0.305, and 0.677, respectively, for compressed images and 0.036, 0.447, and 0.750, respectively, for conventional images. The upper 95% confidence bound for the (absolute) differences in callback rates was 0.012 for normal regions, 0.163 for benign regions, and 0.138 for malignant regions. CONCLUSION: Diagnostic accuracies were equivalent for both compressed and conventional images. The mean false-positive rate at fixed sensitivity was much better with the compressed images. However, the callback rates for malignant lesions were lower when the compressed images were used.     

Subtraction of in-phase and opposed-phase images in dynamic MR mammography

PURPOSE: To develop and to evaluate an advanced image acquisition and analysis method for collecting T(1)-weighted dynamic 3D MR mammography data sets by using a combined in-phase (IP) and opposed-phase (OP) imaging procedure. MATERIALS AND METHODS: 3D MR mammography data sets were acquired by applying an interleaved gradient-echo OP and IP imaging sequence during administration of contrast agent. A phantom data set, two volunteer breast data sets, and six patient breast data sets were recorded. Subtraction of dynamic in-phase magnitude images was performed for clinical assessment. In addition, the magnitude subtraction (SIPOP) as well as the complex subtraction (cSIPOP) of the IP and OP magnitude and phase images were considered. RESULTS: The detection of small lesions, lesion boundaries, and tumor offshoots in fatty tissue was improved by the subtraction of IP and OP images without the risk of signal cancellation due to partial volume effects. CONCLUSION: Dynamic MR mammography acquisition of IP and OP images in combination with appropriate data processing yields important supplementary information that can support routinely applied diagnostics of breast lesions that are fully embedded in fatty tissue by only marginally increasing acquisition time.     

Tomosynthesis and contrast-enhanced digital mammography: recent advances in digital mammography

Digital mammography is more and more replacing conventional mammography. Initial concerns about an inferior image quality of digital mammography have been largely overcome and recent studies even show digital mammography to be superior in women with dense breasts, while at the same time reducing radiation exposure. Nevertheless, an important limitation of digital mammography remains: namely, the fact that summation may obscure lesions in dense breast tissue. However, digital mammography offers the option of so-called advanced applications, and two of these, contrast-enhanced mammography and tomosynthesis, are promising candidates for improving the detection of breast lesions otherwise obscured by the summation of dense tissue. Two techniques of contrast-enhanced mammography are available: temporal subtraction of images acquired before and after contrast administration and the so-called dual-energy technique, which means that pairs of low/high-energy images acquired after contrast administration are subtracted. Tomosynthesis on the other hand provides three-dimensional information on the breast. The images are acquired with different angulations of the X-ray tube while the object or detector is static. Various reconstruction algorithms can then be applied to the set of typically nine to 28 source images to reconstruct 1-mm slices with a reduced risk of obscuring pathology. Combinations of both advanced applications have only been investigated in individual experimental studies; more advanced software algorithms and CAD systems are still in their infancy and have only undergone preliminary clinical evaluation.