Hezi Hadar, Gelber Matan, Balabanov Alexander, Maruvka Yosef E, Freiman Moti
Faculty of Biomedical Engineering, Technion - Israel Institute of Technology, Haifa, Israel.
Faculty of Electrical and Computer Engineering, Technion - Israel Institute of Technology, Haifa, Israel.
Comput Methods Programs Biomed. 2025 Feb;259:108513. doi: 10.1016/j.cmpb.2024.108513. Epub 2024 Nov 19.
Treatment approaches for colorectal cancer (CRC) are highly dependent on the molecular subtype, as immunotherapy has shown efficacy in cases with microsatellite instability (MSI) but is ineffective for the microsatellite stable (MSS) subtype. There is promising potential in utilizing deep neural networks (DNNs) to automate the differentiation of CRC subtypes by analyzing hematoxylin and eosin (H&E) stained whole-slide images (WSIs). Due to the extensive size of WSIs, multiple instance learning (MIL) techniques are typically explored. However, existing MIL methods focus on identifying the most representative image patches for classification, which may result in the loss of critical information. Additionally, these methods often overlook clinically relevant information, like the tendency for MSI class tumors to predominantly occur on the proximal (right side) colon.
We introduce 'CIMIL-CRC', a DNN framework that: (1) solves the MSI/MSS MIL problem by efficiently combining a pre-trained feature extraction model with principal component analysis (PCA) to aggregate information from all patches, and (2) integrates clinical priors, particularly the tumor location within the colon, into the model to enhance patient-level classification accuracy. We assessed our CIMIL-CRC method using the average area under the receiver operating characteristic curve (AUROC) from a 5-fold cross-validation experimental setup for model development on the TCGA-CRC-DX cohort, contrasting it with a baseline patch-level classification, a MIL-only approach, and a clinically-informed patch-level classification approach.
Our CIMIL-CRC outperformed all methods (AUROC: 0.92±0.002 (95% CI 0.91-0.92), vs. 0.79±0.02 (95% CI 0.76-0.82), 0.86±0.01 (95% CI 0.85-0.88), and 0.87±0.01 (95% CI 0.86-0.88), respectively). The improvement was statistically significant. To the best of our knowledge, this is the best result achieved for MSI/MSS classification on this dataset.
Our CIMIL-CRC method holds promise for offering insights into the key representations of histopathological images and suggests a straightforward implementation.
结直肠癌（CRC）的治疗方法高度依赖于分子亚型，因为免疫疗法在微卫星不稳定（MSI）病例中显示出疗效，但对微卫星稳定（MSS）亚型无效。利用深度神经网络（DNN）通过分析苏木精和伊红（H&E）染色的全切片图像（WSI）来自动区分CRC亚型具有很大潜力。由于WSI尺寸巨大，通常会探索多实例学习（MIL）技术。然而，现有的MIL方法专注于识别用于分类的最具代表性的图像块，这可能导致关键信息丢失。此外，这些方法常常忽略临床相关信息，比如MSI类肿瘤主要发生在近端（右侧）结肠的倾向。
我们引入了“CIMIL-CRC”，这是一个DNN框架，它：（1）通过将预训练的特征提取模型与主成分分析（PCA）有效结合来汇总所有图像块的信息，从而解决MSI/MSS的MIL问题；（2）将临床先验信息，特别是肿瘤在结肠内的位置，整合到模型中以提高患者水平的分类准确性。我们使用来自TCGA-CRC-DX队列的5折交叉验证实验设置的受试者操作特征曲线下的平均面积（AUROC）来评估我们的CIMIL-CRC方法，将其与基线图像块水平分类、仅MIL方法以及临床信息丰富的图像块水平分类方法进行对比。
我们的CIMIL-CRC优于所有方法（AUROC分别为：0.92±0.002（95%CI 0.91 - 0.92），而其他方法分别为0.79±0.02（95%CI 0.76 - 0.82）、0.86±0.01（95%CI 0.85 - 0.88）和0.87±0.01（95%CI 0.86 - 0.88））。这种改进具有统计学意义。据我们所知，这是该数据集上MSI/MSS分类取得的最佳结果。
我们的CIMIL-CRC方法有望为深入了解组织病理学图像的关键特征提供见解，并表明其易于实施。
