发表机构
Flensburg University of Applied Sciences; University of Veterinary Medicine, Vienna; Freie Universität Berlin; Technische Hochschule Ingolstadt; Medical University of Vienna; Julius-Maximilians-Universität Würzburg; Friedrich-Alexander-Universität Erlangen-Nürnberg（弗伦斯堡应用科学大学; 维也纳兽医大学; 柏林自由大学; 英戈尔施塔特应用技术大学; 维也纳医科大学; 维尔zburg大学; 埃尔朗根-纽伦堡弗里德里希-亚历山大大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究将多种病理学基础模型作为骨干网络，结合不同类型检测器在MIDOG++和TUPAC16数据集上验证其用于有丝分裂图检测的性能，发现H-optimus-0和Virchow模型表现具竞争力，相关代码已公开。
AI 中文摘要
病理学基础模型（FMs）是在大量通常未标注数据上训练得到的模型，已被证实能生成正则化的潜在空间，可有效用于下游分类任务，在有丝分裂图与其他细胞的分类中亦是如此。然而，目前尚不明确当前FMs的潜在空间是否提供了具有判别性且空间分辨率合适的特征，以作为密集目标检测范式的骨干网络。本研究针对当前常见的病理学FMs（UNI、UNI2-h、Virchow、Virchow2、H-optimus-0、H-optimus-1）探究该问题，并将其性能与基于ResNet50架构的全端到端训练基线进行对比。我们将FM骨干网络与单阶段检测器RetinaNet、双阶段检测器Faster R-CNN、基于自注意力的检测器Deformable DETR分别结合，在多领域MIDOG++数据集及作为域外案例的TUPAC16数据集上开展实验。结果显示，H-optimus-0和Virchow模型取得了具有竞争力的性能，表明所有在图像级自监督下训练的当前FMs的潜在空间适用于直接进行有丝分裂图检测，且在域外测试案例中可能具有更强的鲁棒性。所有代码已公开于该https URL。
英文摘要
Pathology foundation models (FMs) are models trained on vast amounts of typically unlabeled data and have been shown to yield regularized latent spaces that can be used effectively in downstream classification tasks. This is also true for the classification of mitotic figures vs. other cells. However, it is so far unclear if the latent space of current FMs provides features that are discriminant and spatially suitably resolved to also serve as a backbone for dense object detection paradigms. In this work, we investigate this question for common current pathology FMs (UNI, UNI2-h, Virchow, Virchow2, H-optimus-0, H-optimus-1) and compare their performance against a fully end-to-end trained baseline based on a ResNet50 architecture. We combine FM backbones with representatives of single stage, dual stage and self-attention-based detectors (RetinaNet, Faster R-CNN, Deformable DETR respectively) on the multi-domain MIDOG++ dataset, and on the TUPAC16 dataset as an out-of-domain case. We show that the H-optimus-0 and Virchow models yielded competitive performance, indicating that the latent spaces of current FMs, all trained on image-level self-supervision, are suitable for direct mitotic figure detection and may be slightly more robust on our out-of-domain test case. All code is made available publicly at https://github.com/DeepMicroscopy/FM4MFdet.
