Breast Screening Radiologists Research Articles

Exploring the impact of work experience on the detection of specific cancers across different breast density levels on screening mammograms

PurposeThis study explores the attributes of true positive and false positive rates in screening mammogram test sets assessed by breast screening radiologists in order to identify the combined impact of prior images, breast density and lesion features with experience factors linked to diagnostic performance. Methods869 radiologists’ first-time readings across nine mammogram BREAST test sets with 361 normal and 179 cancer mammograms were collected between 2014 and 2023. Participants viewed digital mammograms on diagnostic monitors and localized abnormal lesions. The performances of readers in normal and cancer cases were compared with the ground truth and analyzed in four quartiles of breast density, lesion types and the availability of prior images using Man-Whitney U and Kruskal Wallis tests. The general linear model was applied to determine independent and significant covariates that affected the true positives and false positives. The correlation of the readers’ experience with the performances in different case and lesion features was explored using Spearman test. ResultsThe combination of lesion appearance and the availability of prior images had a significant impact on false positive results (P=0.033). The model that included lesion appearance, breast density, and no prior image status significantly influenced case true positives of readers (P=0.026). Meanwhile, the model considering only lesion appearance and breast density (P=0.002) had a significant effect on lesion true positives. There was a positive correlation observed between the number of cases read per week and readers’ performances, including TP rates and lesion sensitivity across various lesion types (P<0.05). Radiologists reading over 100 cases weekly achieved 80 % true positive rate for architectural distortion, asymmetric density, and masses, while this threshold increased to over 150 cases for calcifications. Detecting mixed lesion types required reading more than 60 cases weekly. Radiologists with over 5 years’ experience achieved 70–75 % accuracy in localizing cancer lesions. ConclusionFindings highlight the significant combined impact of breast density, prior image availability, lesion characteristics, and breast screening readers’ experience on breast cancer detection.

Impact of Artificial Intelligence Decision Support Using Deep Learning on Breast Cancer Screening Interpretation with Single-View Wide-Angle Digital Breast Tomosynthesis.

Background The high volume of data in digital breast tomosynthesis (DBT) and the lack of agreement on how to best implement it in screening programs makes its use challenging. Purpose To compare radiologist performance when reading single-view wide-angle DBT images with and without an artificial intelligence (AI) system for decision and navigation support. Materials and Methods A retrospective observer study was performed with bilateral mediolateral oblique examinations and corresponding synthetic two-dimensional images acquired between June 2016 and February 2018 with a wide-angle DBT system. Fourteen breast screening radiologists interpreted 190 DBT examinations (90 normal, 26 with benign findings, and 74 with malignant findings), with the reference standard being verified by using histopathologic analysis or at least 1 year of follow-up. Reading was performed in two sessions, separated by at least 4 weeks, with a random mix of examinations being read with and without AI decision and navigation support. Forced Breast Imaging Reporting and Data System (categories 1-5) and level of suspicion (1-100) scores were given per breast by each reader. The area under the receiver operating characteristic curve (AUC) and the sensitivity and specificity were compared between conditions by using the public-domain iMRMC software. The average reading times were compared by using the Wilcoxon signed rank test. Results The 190 women had a median age of 54 years (range, 48-63 years). The examination-based reader-averaged AUC was higher when interpreting results with AI support than when reading unaided (0.88 [95% CI: 0.84, 0.92] vs 0.85 [95% CI: 0.80, 0.89], respectively; P = .01). The average sensitivity increased with AI support (64 of 74, 86% [95% CI: 80%, 92%] vs 60 of 74, 81% [95% CI: 74%, 88%]; P = .006), whereas no differences in the specificity (85 of 116, 73.3% [95% CI: 65%, 81%] vs 83 of 116, 71.6% [95% CI: 65%, 78%]; P = .48) or reading time (48 seconds vs 45 seconds; P = .35) were detected. Conclusion Using a single-view digital breast tomosynthesis (DBT) and artificial intelligence setup could allow for a more effective screening program with higher performance, especially in terms of an increase in cancers detected, than using single-view DBT alone. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Chan and Helvie in this issue.

How does image quality affect radiologists\u2019 perceived ability for image interpretation and lesion detection in digital mammography?

ObjectivesTo study how radiologists’ perceived ability to interpret digital mammography (DM) images is affected by decreases in image quality.MethodsOne view from 45 DM cases (including 30 cancers) was degraded to six levels each of two acquisition-related issues (lower spatial resolution and increased quantum noise) and three post-processing-related issues (lower and higher contrast and increased correlated noise) seen during clinical evaluation of DM systems. The images were shown to fifteen breast screening radiologists from five countries. Aware of lesion location, the radiologists selected the most-degraded mammogram (indexed from 1 (reference) to 7 (most degraded)) they still felt was acceptable for interpretation. The median selected index, per degradation type, was calculated separately for calcification and soft tissue (including normal) cases. Using the two-sided, non-parametric Mann-Whitney test, the median indices for each case and degradation type were compared.ResultsRadiologists were not tolerant to increases (medians: 1.5 (calcifications) and 2 (soft tissue)) or decreases (median: 2, for both types) in contrast, but were more tolerant to correlated noise (median: 3, for both types). Increases in quantum noise were tolerated more for calcifications than for soft tissue cases (medians: 3 vs. 4, p = 0.02). Spatial resolution losses were considered less acceptable for calcification detection than for soft tissue cases (medians: 3.5 vs. 5, p = 0.001).ConclusionsPerceived ability of radiologists for image interpretation in DM was affected not only by image acquisition-related issues but also by image post-processing issues, and some of those issues affected calcification cases more than soft tissue cases.Key Points• Lower spatial resolution and increased quantum noise affected the radiologists’ perceived ability to interpret calcification cases more than soft tissue lesion or normal cases.• Post-acquisition image processing-related effects, not only image acquisition-related effects, also impact the perceived ability of radiologists to interpret images and detect lesions.• In addition to current practices, post-acquisition image processing-related effects need to also be considered during the testing and evaluation of digital mammography systems.

The professional quality criteria of Italian breast screening radiologists: results from a national survey comparing the programmes started in 2000-2012 versus the ones started in 1990-1999.

In Italy, due to increasing healthcare budget and staff shortages, the recently created regional mammography screening programmes were established under worse radiology practice quality criteria than the previously created programmes. Using available data from a national questionnaire survey conducted at the end of 2013 and involving 222 responder radiologists, we compared the main professional quality standards of radiologists working in the screening programmes established during the period 2000-2012 with those working in the screening programmes created from 1990 to 1999. The former reported more years of clinical experience in breast imaging and a greater clinical mammogram reading volume than the latter. Conversely, they dedicated less working time to breast imaging, were less likely to participate in the diagnostic assessment of screen-detected lesions, to work in large-staffed screening centres, and to have a screening and a total mammogram reading volume (SMRV and TMRV) ≥ 5000 per year. The level of most professional quality criteria of Italian mammography screening radiologists has decreased over time. As SMRV and TMRV are important predictors of diagnostic accuracy, we can expect a lower interpretation performance of radiologists working in the recently created screening programmes.

Factors associated with breast screening radiologists' annual mammogram reading volume in Italy.

Screening mammogram reading volume (SMRV) and total (screening and clinical) mammogram reading volume (TMRV) per year are strongly associated with the radiologist's diagnostic performance in breast cancer screening. The current article reports the prevalence and correlates of a SMRV and a TMRV ≥5000 among Italian breast screening radiologists. A questionnaire survey was carried out in 2013-2014 by the Italian Group for Mammography Screening (GISMa). The questionnaire included items of information for radiologist's experience-related characteristics and for facility-level factors supposedly associated with SMRV and TMRV. Multivariate analysis was performed using backward stepwise multiple logistic regression models. Data for 235 radiologists from 51 local screening programmes were received. Of the 222 radiologists who were eligible, 133 (59.9%) reported a SMRV ≥5000 and 163 (73.4%) a TMRV ≥5000. Multivariate factors positively associated with both characteristics included: the number of years of experience reading mammograms; the percentage of total working time dedicated to breast imaging and breast care; the participation in diagnostic assessment; and the availability of digital tomosynthesis at facility. Full-time dedication to breast imaging and breast care was associated with the highest odds ratio for a SMRV and a TMRV ≥5000, i.e. 11.80 and 46.74, respectively, versus a percentage of time ≤50%. An early (<2000) year of implementation of the screening programme and the availability of vacuum-assisted biopsy at facility were associated with a SMRV and, respectively, a TMRV ≥5000. Increasing the proportion of radiologists with full-time dedication to breast imaging and breast care qualified as the most effective approach to improve SMRV and TMRV.

Towards personalized compression in mammography: A comparison study between pressure- and force-standardization

ObjectiveTo compare a conventional 14 decanewton (daN) force-standardized compression protocol with a personalized 10kilopascal (kPa) pressure-standardized protocol. MethodsA new add-on contact area detector, which enables pressure-standardized compression, is validated in a double-blinded intra-individual comparison study. Breast screening participants (433) received one craniocaudal (CC) and one mediolateral oblique (MLO) compression for both breasts. Three of these compressions were force-standardized, and one, blinded and randomly assigned, was pressure-standardized. Participants scored their pain experience on an 11-point numerical rating scale (NRS). Three experienced breast-screening radiologists, blinded for compression protocol, indicated which images required retakes. ResultsAn unanticipated under-compression issue that occurred at forces below 5daN was effectively solved with minimal extra radiographer training during the study. For pressure-standardized compressions obtained at 5daN or more, the compressed breasts thickness increased on average 4.2% (MLO)—6.3% (CC), average pain scores were reduced by 10% (MLO)—17% (CC) and the proportion of women experiencing severe pain (NRS≥7) was reduced by 27% (MLO)—32% (CC), compared with force-standardized compressions (all p-values <0.05). Average glandular dose (AGD) and proportions of retakes were similar for both protocols. ConclusionPressure-standardized compressions resulted in AGD values and a retake proportion similar to force-standardized compressions, while pain was significantly reduced.

Experiences with a self-test for Dutch breast screening radiologists: lessons learnt

To evaluate a self-test for Dutch breast screening radiologists introduced as part of the national quality assurance programme. A total of 144 radiologists were invited to complete a test-set of 60 screening mammograms (20 malignancies). Participants assigned findings such as location, lesion type and BI-RADS. We determined areas under the receiver operating characteristics (ROC) curves (AUC), case and lesion sensitivity and specificity, agreement (kappa) and correlation between reader characteristics and case sensitivity (Spearman correlation coefficients). A total of 110 radiologists completed the test (76%). Participants read a median number of 10,000 screening mammograms/year. Median AUC value was 0.93, case and lesion sensitivity was 91% and case specificity 94%. We found substantial agreement for recall (κ = 0.77) and laterality (κ = 0.80), moderate agreement for lesion type (κ = 0.57) and BI-RADS (κ = 0.45) and no correlation between case sensitivity and reader characteristics. Areas under the ROC curve, case sensitivity and lesion sensitivity were satisfactory and recall agreement was substantial. However, agreement in lesion type and BI-RADS could be improved; further education might be aimed at reducing interobserver variation in interpretation and description of abnormalities. We offered individual feedback on interpretive performance and overall feedback at group level. Future research will determine whether performance has improved. • We introduced and evaluated a self-test for Dutch breast screening radiologists. • ROC curves, case and lesion sensitivity and recall agreement were all satisfactory. • Agreement in BI-RADS interpretation and description of abnormalities could be improved. • These are areas that should be targeted with further education and training. • We offered individual feedback on interpretative performance and overall group feedback.

Cross-Atlantic differences in reading the same breast screening cases

American breast screening radiologists typically read a lower annual volume and recall more cases than their UK counterparts. This study investigated what happens when experienced breast screening radiologists from both countries examined the same FFDM case set, albeit using different resolution displays.

Breast biopsy pathology request forms: are the radiologists completing them correctly and legibly?

Radiologists rightly feel that imaging request forms should be completed correctly. Pathologists also require a minimum dataset to be completed on all histopathology requests. The breast screening radiologists and pathologists initiated an audit of wide-bore needle breast biopsy pathology request forms.

Breast screening radiographers and radiologists: performance and confidence levels on the PERFORMS film sets

Specially trained radiographers are increasingly double-reading screening cases with radiologists. Each year the majority of these film-readers interpret a 'PERFORMS' set of difficult cases as a means of self-assessing their skills. We examined the radiographers' confidence in their interpretation of these difficult cases and investigated whether they performed as well as radiologists. We also studied radiographers' performance and confidence ratings over a number of years, comparing those with different levels of experience. Additionally, by closely matching individual radiographers and radiologists, we investigated real-life experience, performance and confidence levels on PERFORMS, fractionating by both volume of cases read, and years of experience. Data were examined for approximately 250 breast-screening radiologists and 90 specially trained radiographers over 3 years. Overall, radiographers did not perform as well as radiologists on this particular task, nor were they as confident of their diagnosis. However, when closely matched on real-life factors these differences in performance levels disappeared and radiographers performed as well as radiologists. Although the more experienced radiographers performed at a level similar to the radiologists, differences in confidence levels remained. This supports the current UK practice of radiographers double-reading with radiologists and implies that experienced radiographers, like radiologists, may be able to be single-readers.

A 3-year audit of radiographer screen film reading

A team of five radiographers assists with film reading in our large combined screening and symptomatic breast unit. The first were trained in 1999, and now have 5 years experience. For the last 3 years their screen reading performance has been intensively audited. Film reading numbers have ranged from 4.8 to 11,000/year. Cancer detection rates average 6/1000, with the best radiographer achieving 7/1000. Recall rates are kept low by consensus review of all recalls, less than 5%. Film reading performance in real life does reflect the 'PERFORMS' score. Lobular cancers are more likely to be missed by new film readers and are disproportionately over-represented in the 'minority report' cancers detected. In conclusion, radiographers perform as well as average breast-screening radiologists after 3 years supervised experience.

Are there demographic differences in film reading skill?

Since 1991, the majority of UK breast screening radiologists have annually taken part in the Personal Performance in Mammographic Screening (PERFORMS) self-assessment scheme. In addition to reporting a set of difficult mammographic cases, they have also completed a questionnaire where they indicated their normal screening practices (e.g. number of cases read per week, mammographic views used and the time of day they usually read screening cases) and demographic details (e.g. age, gender and experience). The purpose of this study was to determine whether there were any correlations between these questionnaire data and the various performance measures provided by the self-assessment film sets. Additionally, any change in response to the questionnaire over time was examined. All data presented are anonymous. There were no significant differences for gender, but significant correlations were found for some of the questionnaire responses and performance. For example, experience and the number of cases read per week were positively correlated with some performance measures. In conclusion, different screening practices and some demographic differences can have an effect on an individual's performance in interpreting a self-assessment film set. This finding may also be applicable to actual breast screening performance.

A Pilot Evaluation of the R2 Image Checker System and Users' Response in the Detection of Interval Breast Cancers on Previous Screening Films

AIM: Breast screening radiologists are under considerable pressure to maximize cancer detection and minimize the number of interval cancers. This study used screening films, previously reported as normal, taken before the development of an interval cancer. The aim was to determine the sensitivity and specificity of the R2 Image Checker System and the responses of users in detecting abnormalities in these previous films. MATERIALS AND METHODS: The first R2 Image Checker System in the U.K. was installed in the Canterbury Assessment Centre for evaluation purposes. System performance was compared with the standard protocol for assessing interval cancers used in the NHS Breast Screening Programme (NHSBSP) Three readers examined the previous films blind before switching on R2 and compared the prompts generated with the location of the subsequent interval cancer. RESULTS: The previous screening films for 104 cases generated 134 R2 prompts on the subsequent cancer side and 109 on the non-cancer side. Readers classified 29 cancers as false-negative or minimal signs and R2 correctly prompted in 15 of these cases (52%). At least one reader rejected the correct prompt in five cases. CONCLUSION: From this preliminary study it appears that R2 overprompts normal areas and underprompts some cancers in these difficult cases. A full multi-centre evaluation is needed to assess the possible contribution of R2 to the NHSBSP. Garvican, L. and Field, S. (2001).Clinical Radiology56 , 833–837.