Weakly supervised deep learning-based classification for histopathology of gliomas: a single center experience.

in Scientific reports by Mingrong Zuo, Xiang Xing, Linmao Zheng, Hao Wang, Yunbo Yuan, Siliang Chen, Tianping Yu, ShuXin Zhang, Yuan Yang, Qing Mao, Yongbin Yu, Ni Chen, Yanhui Liu

TLDR

  • A weakly supervised deep learning model was developed to aid in glioma diagnosis using hematoxylin-eosin-stained slides, achieving promising results in differentiating grades of astrocytomas, oligodendrogliomas, and gliomas.
  • The model showed a strong ability to infer IDH status and demonstrated high AUC values for differentiating glioma types.
  • The study highlights the potential of weakly supervised deep learning models in facilitating accurate and prompt histopathological diagnosis of tumors.

Abstract

Multiple artificial intelligence systems have been created to facilitate accurate and prompt histopathological diagnosis of tumors using hematoxylin-eosin-stained slides. We aimed to investigate whether weakly supervised deep learning can aid in glioma diagnosis. We analyzed 472 whole slide images (WSIs) from 226 patients in West China Hospital (WCH) and 1604 WSIs from 880 patients in The Cancer Genome Atlas (TCGA). We utilized the OpenSlide library to load WSIs, segmented them into small patches using the DeepZoom module, and then normalized the color using the Reinhard method. A weakly supervised deep learning model was developed using ResNet-50 combined with an attention mechanism. We investigated the performance of the model by calculating area under the curve (AUC) in a ten-fold cross-validation setting. Heatmap visualizations showed the prediction mechanism of the model. The results were promising, with high AUC values for differentiating grades of astrocytomas, oligodendrogliomas, all gliomas, and glioma types in the TCGA dataset (0.9419, 0.8659, 0.9904, and 0.9298, respectively), and in the WCH cohort (0.9048, 0.7423, 0.9510, and 0.7098, respectively). The model demonstrated a strong ability to infer IDH status in the TCGA dataset (AUC = 0.9488). The weakly supervised deep learning model proved to be an effective and reliable tool for neuropathological diagnosis, making it an attractive auxiliary tool.

Overview

  • The study investigated whether weakly supervised deep learning can aid in glioma diagnosis using hematoxylin-eosin-stained slides.
  • The team analyzed whole slide images (WSIs) from 226 patients in West China Hospital (WCH) and 880 patients in The Cancer Genome Atlas (TCGA).
  • The objective of the study was to develop a weakly supervised deep learning model that can accurately differentiate grades of astrocytomas, oligodendrogliomas, and gliomas, as well as infer IDH status.

Comparative Analysis & Findings

  • The model achieved high area under the curve (AUC) values for differentiating grades of astrocytomas, oligodendrogliomas, all gliomas, and glioma types in the TCGA dataset (0.9419, 0.8659, 0.9904, and 0.9298, respectively).
  • The model showed a strong ability to infer IDH status in the TCGA dataset (AUC = 0.9488).
  • The results suggest that the weakly supervised deep learning model is an effective and reliable tool for neuropathological diagnosis, making it an attractive auxiliary tool.

Implications and Future Directions

  • The study highlights the potential of weakly supervised deep learning models in facilitating accurate and prompt histopathological diagnosis of tumors.
  • Future studies should aim to validate the model's performance on larger and more diverse datasets, as well as explore its application in other cancer types and histopathological sections.
  • The development of more advanced attention mechanisms and transfer learning techniques may further enhance the model's ability to accurately diagnose gliomas and infer IDH status.
Read Full Article