Evaluation of Combined Artificial Intelligence and Radiologist Assessment to Interpret Screening Mammograms.

Importance: Mammography screening currently relies on subjective human interpretation. Artificial intelligence (AI) advances could be used to increase mammography screening accuracy by reducing missed cancers and false positives. Objective: To evaluate whether AI can overcome human mammography interpretation limitations with a rigorous, unbiased evaluation of machine learning algorithms. Design, Setting, and Participants: In this diagnostic accuracy study conducted between September 2016 and November 2017, an international, crowdsourced challenge was hosted to foster AI algorithm development focused on interpreting screening mammography. More than 1100 participants comprising 126 teams from 44 countries participated. Analysis began November 18, 2016. Main Outcomes and Measurements: Algorithms used images alone (challenge 1) or combined images, previous examinations (if available), and clinical and demographic risk factor data (challenge 2) and output a score that translated to cancer yes/no within 12 months. Algorithm accuracy for breast cancer detection was evaluated using area under the curve and algorithm specificity compared with radiologists' specificity with radiologists' sensitivity set at 85.9% (United States) and 83.9% (Sweden). An ensemble method aggregating top-performing AI algorithms and radiologists' recall assessment was developed and evaluated. Results: Overall, 144 231 screening mammograms from 85 580 US women (952 cancer positive ≤12 months from screening) were used for algorithm training and validation. A second independent validation cohort included 166 578 examinations from 68 008 Swedish women (780 cancer positive). The top-performing algorithm achieved an area under the curve of 0.858 (United States) and 0.903 (Sweden) and 66.2% (United States) and 81.2% (Sweden) specificity at the radiologists' sensitivity, lower than community-practice radiologists' specificity of 90.5% (United States) and 98.5% (Sweden). Combining top-performing algorithms and US radiologist assessments resulted in a higher area under the curve of 0.942 and achieved a significantly improved specificity (92.0%) at the same sensitivity. Conclusions and Relevance: While no single AI algorithm outperformed radiologists, an ensemble of AI algorithms combined with radiologist assessment in a single-reader screening environment improved overall accuracy. This study underscores the potential of using machine learning methods for enhancing mammography screening interpretation.

Radiotherapy isocenters verified by matching to bony landmarks of the canine and feline head differ when localized using volumetric versus planar imaging.

The "gold standard" for verification of patient positioning before linear accelerator-based stereotactic radiation therapy is kilovoltage cone-beam computed tomography (kV-CBCT), which is not uniformly available or utilized; planar imaging is sometimes used instead. The primary aim of this study was to determine if the position of the bony skull, when used as a surrogate for isocenter verification, is different when orthogonal megavoltage (MV) portal or kilovoltage (kV/kV) radiographs are used for image guidance, rather than kV-CBCT. A secondary aim was to determine the influence of intra-observer variability, body size and skull conformation on positioning, as determined using these three imaging modalities. Dogs and cats receiving radiotherapy of the head were recruited for this prospective analytical study. Planar (MV portal and kV/kV images) and volumetric (kV-CBCT) images were acquired before treatment, and manually coregistered with reference images. Differences in skull position when matched based on MV portal, kV/kV images and kV-CBCT were compared. A total of 65 subjects and 148 unique datasets were evaluated. The Wilcoxon rank-sum test was used to evaluate effects of transitioning between imaging modalities. When comparing magnitude of shifts in MV to kV-CBCT, MV to kV/kV and kV/kV to kV-CBCT, there were statistically significant differences. Results were not measurably impacted by body size, skull conformation or interobserver differences. Based on shift magnitude and direction, an isotropic setup margin of at least 1 mm should be incorporated within the planning target volume when MV or kV planar imaging is used for position verification.