Early Indicators of the Impact of Using AI in Mammography Screening for Breast Cancer.

Background Retrospective studies have suggested that using artificial intelligence (AI) may decrease the workload of radiologists while preserving mammography screening performance. Purpose To compare workload and screening performance for two cohorts of women who underwent screening before and after AI system implementation. Materials and Methods This retrospective study included 50-69-year-old women who underwent biennial mammography screening in the Capital Region of Denmark. Before AI system implementation (October 1, 2020, to November 17, 2021), all screenings involved double reading. For screenings conducted after AI system implementation (November 18, 2021, to October 17, 2022), likely normal screenings (AI examination score ≤5 before May 3, 2022, or ≤7 on or after May 3, 2022) were single read by one of 19 senior full-time breast radiologists. The remaining screenings were read by two radiologists with AI-assisted decision support. Biopsy and surgical outcomes were retrieved between October 1, 2020, and April 15, 2023, ensuring at least 180 days of follow-up. Screening metrics were compared using the χ2 test. Reading workload reduction was measured as saved screening reads. Results In total, 60 751 and 58 246 women were screened before and after AI system implementation, respectively (median age, 58 years [IQR, 54-64 years] for both cohorts), with a median screening interval before AI of 845 days (IQR, 820-878 days) and with AI of 993 days (IQR, 968-1013 days; P < .001). After AI system implementation, the recall rate decreased by 20.5% (3.09% before AI [1875 of 60 751] vs 2.46% with AI [1430 of 58 246]; P < .001), the cancer detection rate increased (0.70% [423 of 60 751] vs 0.82% [480 of 58 246]; P = .01), the false-positive rate decreased (2.39% [1452 of 60 751] vs 1.63% [950 of 58 246]; P < .001), the positive predictive value increased (22.6% [423 of 1875] vs 33.6% [480 of 1430]; P < .001), the rate of small cancers (≤1 cm) increased (36.6% [127 of 347] vs 44.9% [164 of 365]; P = .02), the rate of node-negative cancers was unchanged (76.7% [253 of 330] vs 77.8% [273 of 351]; P = .73), and the rate of invasive cancers decreased (84.9% [359 of 423] vs 79.6% [382 of 480]; P = .04). The reading workload was reduced by 33.5% (38 977 of 116 492 reads). Conclusion In a population-based mammography screening program, using AI reduced the overall workload of breast radiologists while improving screening performance. Published under a CC BY 4.0 license. Supplemental material is available for this article. See also the editorial by Lee and Friedewald in this issue.

Robust cross-vendor mammographic texture models using augmentation-based domain adaptation for long-term breast cancer risk.

Purpose: Risk-stratified breast cancer screening might improve early detection and efficiency without comprising quality. However, modern mammography-based risk models do not ensure adaptation across vendor-domains and rely on cancer precursors, associated with short-term risk, which might limit long-term risk assessment. We report a cross-vendor mammographic texture model for long-term risk. Approach: The texture model was robustly trained using two systematically designed case-control datasets. Textural features, indicative of future breast cancer, were learned by excluding samples with diagnosed/potential malignancies from training. An augmentation-based domain adaption technique, based on flavorization of mammographic views, ensured generalization across vendor-domains. The model was validated in 66,607 consecutively screened Danish women with flavorized Siemens views and 25,706 Dutch women with Hologic-processed views. Performances were evaluated for interval cancers (IC) within 2 years from screening and long-term cancers (LTC) from 2 years after screening. The texture model was combined with established risk factors to flag 10% of women with the highest risk. Results: In Danish women, the texture model achieved an area under the receiver operating characteristic curve (AUC) of 0.71 and 0.65 for ICs and LTCs, respectively. In Dutch women with Hologic-processed views, the AUCs were not different from AUCs in Danish women with flavorized views. The AUC for texture combined with established risk factors increased to 0.68 for LTCs. The 10% of women flagged as high-risk accounted for 25.5% of ICs and 24.8% of LTCs. Conclusions: The texture model robustly estimated long-term breast cancer risk while adapting to an unseen processed vendor-domain and identified a clinically relevant high-risk subgroup.

Assessing Breast Cancer Risk by Combining AI for Lesion Detection and Mammographic Texture.

Background Recent mammography-based risk models can estimate short-term or long-term breast cancer risk, but whether risk assessment may improve by combining these models has not been evaluated. Purpose To determine whether breast cancer risk assessment improves when combining a diagnostic artificial intelligence (AI) system for lesion detection and a mammographic texture model. Materials and Methods This retrospective study included Danish women consecutively screened for breast cancer at mammography from November 2012 to December 2015 who had at least 5 years of follow-up data. Examinations were evaluated for short-term risk using a commercially available diagnostic AI system for lesion detection, which produced a score to indicate the probability of cancer. A mammographic texture model, trained on a separate data set, assessed textures associated with long-term cancer risk. Area under the receiver operating characteristic curve (AUC) analysis was used to evaluate both the individual and combined performance of the AI and texture models for the prediction of future cancers in women with a negative screening mammogram, including those with interval cancers diagnosed within 2 years of screening and long-term cancers diagnosed 2 years or more after screening. AUCs were compared using the DeLong test. Results The Danish screening cohort included 119 650 women (median age, 59 years [IQR, 53-64 years]), of whom 320 developed interval cancers and 1401 developed long-term cancers. The combination model achieved a higher AUC for interval and long-term cancers grouped together than either the diagnostic AI (AUC, 0.73 vs 0.70; P < .001) or the texture risk (AUC, 0.73 vs 0.66; P < .001) models. The 10% of women with the highest combined risk identified by the combination model accounted for 44.1% (141 of 320) of interval cancers and 33.7% (472 of 1401) of long-term cancers. Conclusion Combining a diagnostic AI system and mammographic texture model resulted in improved risk assessment for interval cancers and long-term cancers and enabled identification of women at high risk. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Poynton and Slanetz in this issue.

Preserved blood-brain barrier and neurovascular coupling in female 5xFAD model of Alzheimer's disease.

Introduction: Dysfunction of the cerebral vasculature is considered one of the key components of Alzheimer's disease (AD), but the mechanisms affecting individual brain vessels are poorly understood. Methods: Here, using in vivo two-photon microscopy in superficial cortical layers and ex vivo imaging across brain regions, we characterized blood-brain barrier (BBB) function and neurovascular coupling (NVC) at the level of individual brain vessels in adult female 5xFAD mice, an aggressive amyloid-ß (Aß) model of AD. Results: We report a lack of abnormal increase in adsorptive-mediated transcytosis of albumin and preserved paracellular barrier for fibrinogen and small molecules despite an extensive load of Aß. Likewise, the NVC responses to somatosensory stimulation were preserved at all regulatory segments of the microvasculature: penetrating arterioles, precapillary sphincters, and capillaries. Lastly, the Aß plaques did not affect the density of capillary pericytes. Conclusion: Our findings provide direct evidence of preserved microvascular function in the 5xFAD mice and highlight the critical dependence of the experimental outcomes on the choice of preclinical models of AD. We propose that the presence of parenchymal Aß does not warrant BBB and NVC dysfunction and that the generalized view that microvascular impairment is inherent to Aß aggregation may need to be revised.

An Artificial Intelligence-based Mammography Screening Protocol for Breast Cancer: Outcome and Radiologist Workload.

Background Developments in artificial intelligence (AI) systems to assist radiologists in reading mammograms could improve breast cancer screening efficiency. Purpose To investigate whether an AI system could detect normal, moderate-risk, and suspicious mammograms in a screening sample to safely reduce radiologist workload and evaluate across Breast Imaging Reporting and Data System (BI-RADS) densities. Materials and Methods This retrospective simulation study analyzed mammographic examination data consecutively collected from January 2014 to December 2015 in the Danish Capital Region breast cancer screening program. All mammograms were scored from 0 to 10, representing the risk of malignancy, using an AI tool. During simulation, normal mammograms (score < 5) would be excluded from radiologist reading and suspicious mammograms (score > recall threshold [RT]) would be recalled. Two radiologists read the remaining mammograms. The RT was fitted using another independent cohort (same institution) by matching to the radiologist sensitivity. This protocol was further applied to each BI-RADS density. Screening outcomes were measured using the sensitivity, specificity, workload, and false-positive rate. The AI-based screening was tested for noninferiority sensitivity compared with radiologist screening using the Farrington-Manning test. Specificities were compared using the McNemar test. Results The study sample comprised 114 421 screenings for breast cancer in 114 421 women, resulting in 791 screen-detected, 327 interval, and 1473 long-term cancers and 2107 false-positive screenings. The mean age of the women was 59 years ± 6 (SD). The AI-based screening sensitivity was 69.7% (779 of 1118; 95% CI: 66.9, 72.4) and was noninferior (P = .02) to the radiologist screening sensitivity of 70.8% (791 of 1118; 95% CI: 68.0, 73.5). The AI-based screening specificity was 98.6% (111 725 of 113 303; 95% CI: 98.5, 98.7), which was higher (P < .001) than the radiologist specificity of 98.1% (111 196 of 113 303; 95% CI: 98.1, 98.2). The radiologist workload was reduced by 62.6% (71 585 of 114 421), and 25.1% (529 of 2107) of false-positive screenings were avoided. Screening results were consistent across BI-RADS densities, although not significantly so for sensitivity. Conclusion Artificial intelligence (AI)-based screening could detect normal, moderate-risk, and suspicious mammograms in a breast cancer screening program, which may reduce the radiologist workload. AI-based screening performed consistently across breast densities. © RSNA, 2022 Online supplemental material is available for this article.

GENERALIZABLE MULTI-SITE TRAINING AND TESTING OF DEEP NEURAL NETWORKS USING IMAGE NORMALIZATION.

The ability of medical image analysis deep learning algorithms to generalize across multiple sites is critical for clinical adoption of these methods. Medical imging data, especially MRI, can have highly variable intensity characteristics across different individuals, scanners, and sites. However, it is not practical to train algorithms with data from all imaging equipment sources at all possible sites. Intensity normalization methods offer a potential solution for working with multi-site data. We evaluate five different image normalization methods on training a deep neural network to segment the prostate gland in MRI. Using 600 MRI prostate gland segmentations from two different sites, our results show that both intra-site and inter-site evaluation is critical for assessing the robustness of trained models and that training with single-site data produces models that fail to fully generalize across testing data from sites not included in the training.