基于 DNA 超分子水凝胶的保护性 NK 细胞储库用于增强三阴性乳腺癌治疗
Protective NK Cell Reservoir Based on DNA Supramolecular Hydrogel for Enhanced Triple-Negative Breast Cancer Therapy.
CELL INTELLIGENCE · 肿瘤细胞治疗研究
肿瘤细胞治疗研究
一种用于乳腺癌 H&E 切片中基质 TILs 评分的模块化深度学习流程
英文原题：A modular deep learning pipeline for stromal TILs scoring in breast cancer H&E slides.
查看英文原题
A modular deep learning pipeline for stromal TILs scoring in breast cancer H&E slides.
分数与星级只用于站内排序 —— 不代表疗效、安全性或个人适用性。
研究概要
重要的是，该流程具有可解释性：每个阶段都能产生人类可读的输出（间质掩膜、间质内 TIL 分布图），且 SegGradCAM 可视化证实预测所依赖的是具有生物学意义的组织区域。这些发现表明，该流程有望成为一种可靠且临床适用的工具，用于乳腺癌病理中标准化、全自动的 TILs 定量。源代码和预训练模型已公开提供：https://github.com/Shrief-Abdelazeez/TILs-Scoring。
研究思路结论见上方概要
背景
TIL（肿瘤浸润淋巴细胞）（TILs）是乳腺癌免疫活性的重要指标，然而在常规病理学中，在H&E切片上对其进行一致评分仍然具有挑战性。本研究提出了一种模块化深度学习流程，能够提供符合国际免疫肿瘤学生物标志物工作组（IIOBWG）指南的全自动且连续的间质TILs（sTILs）评分。
方法
该流程结合了三个组成部分：一个通过病理学家引导的主动学习进行优化的 TIL 分割模型，一个基于增强型 DeepLabV3+ 的稳健间质分割网络，以及一个学习 TIL 在间质区域内如何分布的轻量级回归模块。一种新的自适应聚合策略将 patch 级预测整合为单一且具有临床意义的评分，同时兼顾异质性浸润。
结果
该系统在两个独立数据集（分别为 60 例和 112 例 WSI）上进行了评估，这些数据集带有专家标注的 ROI，结果与病理学家的一致性很高（Pearson 为 0.814；ICC 为 0.808）。
展开英文摘要原文
BACKGROUND AND OBJECTIVE
Tumor-infiltrating lymphocytes (TILs) are an important indicator of immune activity in breast cancer, yet scoring them consistently on H&E slides remains challenging in routine pathology. This work presents a modular deep learning pipeline that delivers fully automated and continuous stromal TILs (sTILs) scores in line with the International Immuno-Oncology Biomarker Working Group (IIOBWG) guidelines.
METHODS
The pipeline combines three components: a TIL segmentation model refined through pathologist-guided active learning, a robust stroma segmentation network based on an enhanced DeepLabV3+, and a lightweight regression module that learns how TILs distribute within stromal regions. A new adaptive aggregation strategy integrates patch-level predictions into a single, clinically meaningful score while accounting for heterogeneous infiltration.
RESULTS
The system was evaluated on two independent datasets (60 and 112 WSIs) with expert-annotated ROIs, achieving strong agreement with pathologists (Pearson of 0.814; ICC of 0.808).
CONCLUSIONS
Importantly, the pipeline is interpretable: each stage produces human-readable outputs (stroma masks, TIL-in-stroma maps), and SegGradCAM visualizations confirm that predictions rely on biologically relevant tissue regions. These findings demonstrate the pipeline's potential as a reliable and clinically adaptable tool for standardized, fully automated TILs quantification in breast cancer pathology. The source code and pretrained models are publicly available at https://github.com/Shrief-Abdelazeez/TILs-Scoring.
论文信息
- 作者
- Abdelazeez S、Ahmed F、Adalid L、Siemion K、Lopez C、Lejeune M、Rashwan H、Korzynska A
- 单位
- Laboratory of Analysis and Processing of Microscopic Image, Nalęcz Institute of Biocybernetics and Biomedical Engineering, Polish Academy of Sciences, Ks. Trojdena 4, Warsaw, 02-109, Poland; System and Biomedical Engineering Department, Faculty of Engineering, Cairo University, Giza, Cairo, Egypt. Electronic address: sabdelazeez@ibib.waw.pl.Switzerland
- 期刊
- Computer methods and programs in biomedicine2026 Aug 1
- 原文标识
- PubMed 42013559 · DOI 10.1016/j.cmpb.2026.109374