arXiv:2606. 07368v1 Announce Type: cross Abstract: Automated mitosis detection is a well-established task in computational pathology.
Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.
Mitotic figure (MF) analysis supports tumor grading and prognostic assessment, but automated models remain sensitive to differences in tissue type and image acquisition. We present MiTHras, a task-spe...
The study introduces MitPro, an AI tool that guides pathologists to regions with high mitotic activity and highlights mitotic figures, aiming to reduce variability and time in tumour grading. In a retrospective paired reader study of 385 whole‑slide images across seven tumour types, AI assistance raised the intraclass correlation coefficient from 0.589 to 0.949 and cut median assessment time from 286.4 to 127.8 seconds. The tool also produced a slight increase in mitotic counts, aligning with the detection of more active hotspots, while the frequency of score changes matched typical inter‑pathologist variation.
The paper introduces CoPath, a lightweight framework for diagnosing peripheral neuroblastic tumors (pNTs) from whole-slide images. CoPath combines CoHisNet, a multi‑scale feature‑fusion network that replaces traditional MLPs with Kolmogorov‑Arnold Network layers for efficient nonlinear modeling, and PathVote, which aggregates patch‑level predictions using pathology‑informed priors. Experiments on a private pNT cohort and the public BreakHis dataset show that CoPath matches or surpasses existing classifiers while reducing computational complexity.
arXiv:2609.24736v1 Announce Type: cross Abstract: Mitotic figure (MF) analysis supports tumor grading and prognostic assessment, but automated models remain sensitive to differences in tissue type an...
arXiv:2606. 12346v1 Announce Type: cross Abstract: Hematoxylin and eosin (H&E) staining is the cornerstone of histopathology, yet scalable, quantitative analysis of H&E whole-slide images (WSIs) remains a central challenge in computational pathology.
HERO (Histology Encoder for Robust Representation in Oncology) is a ViT‑G/14 pathology foundation model trained with DINO and iBOT objectives and refined using high‑resolution Gram anchoring on a 500‑million‑tile corpus from about 575,000 clinical whole‑slide images. It demonstrates superior robustness to center, scanner, and stain variation compared to other state‑of‑the‑art foundation models, while maintaining competitive performance on tile‑level classification, segmentation, and gene‑expression prediction. Across 39 slide‑level clinical tasks, HERO ranks first on average and achieves the best average rank across six benchmark frameworks under an equal‑weighted analysis.