Background Histopathological assessment of lymph nodes is essential in breast cancer care. Artificial intelligence (AI) has shown promise in supporting this process, yet successful clinical adoption requires attention to human-AI interaction and user interface design. This study explored the user experience of an AI-based clinical decision support tool assisting pathologists in diagnosing breast cancer metastases. Methods We conducted a within-subjects mixed-methods study in which ten pathologists evaluated two AI-based clinical decision support user interface prototypes for breast cancer metastasis detection in sentinel nodes. ANNO displayed threshold based on-image annotations with minimal automation, whereas RAVS presented four likelihood regions of interest (ROI) per slide with automated measurement and case summaries. Quantitative measures included review time, diagnostic outcomes, and System Usability Scale (SUS) scores. Qualitative data were gathered via semi-structured interviews and analysed thematically. Results ANNO achieved higher usability scores (mean SUS 90.3) and was associated with a sense of control and resembled current diagnostic workflow, whereas RAVS (mean SUS 80.0) was preferred by participants valuing rapid navigation and automated measurements. Interview findings highlighted trade-offs: ANNO preserved diagnostic overview and autonomy, while RAVS improved navigational efficiency but reduced whole-slide context. ROI presentation strategy seemed to influence preference. Trust in AI was generally cautious and linked to needs for transparency and long-term clinical exposure. Conclusion Both interfaces were perceived as potentially clinically valuable but differed in their usability and workflow integration. Our exploratory findings indicates that for future development of AI-based decision aiding tools for this diagnostic task, a combination of high human control with targeted automation is advisable, combined with visual clarity, and an effective ROI presentation strategy.

Current hardware limitations make it impossible to train convolutional neural networks on gigapixel image inputs directly. Recent developments in weakly supervised learning, such as attention-gated multiple instance learning, have shown promising results, but often use multi-stage or patch-wise training strategies risking suboptimal feature extraction, which can negatively impact performance. In this paper, we propose to train a ResNet-34 encoder with an attention-gated classification head in an end-to-end fashion, which we call StreamingCLAM, using a streaming implementation of convolutional layers. This allows us to train end-to-end on 4-gigapixel microscopic images using only slide-level labels. We achieve a mean area under the receiver operating characteristic curve of 0.9757 for metastatic breast cancer detection (CAMELYON16), close to fully supervised approaches using pixel-level annotations. Our model can also detect MYC-gene translocation in histologic slides of diffuse large B-cell lymphoma, achieving a mean area under the ROC curve of 0.8259. Furthermore, we show that our model offers a degree of interpretability through the attention mechanism.

Deep learning (DL) has shown great potential in digital pathology applications. The robustness of a diagnostic DL-based solution is essential for safe clinical deployment. In this work we evaluate if adding uncertainty estimates for DL predictions in digital pathology could result in increased value for the clinical applications, by boosting the general predictive performance or by detecting mispredictions. We compare the effectiveness of model-integrated methods (MC dropout and Deep ensembles) with a model-agnostic approach (Test time augmentation, TTA). Moreover, four uncertainty metrics are compared. Our experiments focus on two domain shift scenarios: a shift to a different medical center and to an underrepresented subtype of cancer. Our results show that uncertainty estimates increase reliability by reducing a model’s sensitivity to classification threshold selection as well as by detecting between 70 and 90% of the mispredictions done by the model. Overall, the deep ensembles method achieved the best performance closely followed by TTA.

Poor generalizability is a major barrier to clinical implementation of artificial intelligence in digital pathology. The aim of this study was to test the generalizability of a pretrained deep learning model to a new diagnostic setting and to a small change in surgical indication. A deep learning model for breast cancer metastases detection in sentinel lymph nodes, trained on CAMELYON multicenter data, was used as a base model, and achieved an AUC of 0.969 (95% CI 0.926–0.998) and FROC of 0.838 (95% CI 0.757–0.913) on CAMELYON16 test data. On local sentinel node data, the base model performance dropped to AUC 0.929 (95% CI 0.800–0.998) and FROC 0.744 (95% CI 0.566–0.912). On data with a change in surgical indication (axillary dissections) the base model performance indicated an even larger drop with a FROC of 0.503 (95%CI 0.201–0.911). The model was retrained with addition of local data, resulting in about a 4% increase for both AUC and FROC for sentinel nodes, and an increase of 11% in AUC and 49% in FROC for axillary nodes. Pathologist qualitative evaluation of the retrained model´s output showed no missed positive slides. False positives, false negatives and one previously undetected micro-metastasis were observed. The study highlights the generalization challenge even when using a multicenter trained model, and that a small change in indication can considerably impact the model´s performance.