← 研踪 PaperTrailarXiv:2611.21565 · v1
When Do Histology Foundation Models Need Geometric Test-Time Augmentation?
组织学基础模型何时需要几何测试时增强?
首次提交 2026-11-17 02:41
版本更新 2026-11-17 02:41
雷达观测 2026-11-17 16:10
摘要
Test-time augmentation (TTA) aggregates model predictions across multiple transformed views of an input during inference. Because histology images lack a canonical orientation, TTA---pooling across the 8 rotations and reflections of a square---is a standard default in computational pathology. However, the rise of self-supervised histology foundation models (FMs) with inherent rotational invariance calls its continued utility into question. We benchmark 9 histology FMs and 6 general-purpose models across four patch-level and two whole-slide image (WSI) datasets. We find that TTA remains a reliable post hoc improvement, but benefits diminish from percentage points (pp) for general-purpose backbones to pp for histology FMs on TCGA-UT, and shrink further for the largest models. To eliminate computational overhead, our experiments demonstrate that entropy routing---applying TTA selectively to uncertain predictions---recovers most of the gain while augmenting only of samples. Finally, we show that patch-level TTA improvements do not transfer to slide-level multiple instance learning (MIL). Code: https://github.com/lars-chen/histology-tta
作者备注:17 pages, 5 figures
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-17 02:41 · 最新观测版本
When Do Histology Foundation Models Need Geometric Test-Time Augmentation?
Test-time augmentation (TTA) aggregates model predictions across multiple transformed views of an input during inference. Because histology images lack a canonical orientation, TTA---pooling across the 8 rotations and reflections of a square---is a standard default in computational pathology. However, the rise of self-supervised histology foundation models (FMs) with inherent rotational invariance calls its continued utility into question. We benchmark 9 histology FMs and 6 general-purpose models across four patch-level and two whole-slide image (WSI) datasets. We find that TTA remains a reliable post hoc improvement, but benefits diminish from percentage points (pp) for general-purpose backbones to pp for histology FMs on TCGA-UT, and shrink further for the largest models. To eliminate computational overhead, our experiments demonstrate that entropy routing---applying TTA selectively to uncertain predictions---recovers most of the gain while augmenting only of samples. Finally, we show that patch-level TTA improvements do not transfer to slide-level multiple instance learning (MIL). Code: https://github.com/lars-chen/histology-tta
查看此版本解读 →查看此版本原文 ↗