Scholarlink

This research paper addresses the imperative task of nuclei segmentation within H&E-stained Histopathological Images through the application of a Deep Learning approach. Utilizing the “2018 Data Science Bowl” dataset, we amalgamated individual mask images, resized both images and masks, and implemented data augmentation techniques to enhance the diversity of the train set. Our study encompasses the construction and training of various deep learning models, including U-Net, Segmentation Network (SegNet), Pyramid Scene Parsing Network (PSPNet), Efficient Neural Network (Enet), and DeepLab Version 3 (DeepLabV3). Through this experiment involving hyperparameter tuning, we evaluated each model's performance having metrics such as Accuracy, Mean Intersection over Union (IoU), and Dice Coefficient. Notably, our conclusion shows that the Seg-Net model attains a remarkable accuracy of 96%. This study advances automated histopathology image processing and shows that deep learning is effective for better nuclei segmentation.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Segmentation of Nuclei from H&E Stained Histopathological Images Using Deep Learning Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Aparna Kaushik,
  • Himanshu Mittal

摘要

This research paper addresses the imperative task of nuclei segmentation within H&E-stained Histopathological Images through the application of a Deep Learning approach. Utilizing the “2018 Data Science Bowl” dataset, we amalgamated individual mask images, resized both images and masks, and implemented data augmentation techniques to enhance the diversity of the train set. Our study encompasses the construction and training of various deep learning models, including U-Net, Segmentation Network (SegNet), Pyramid Scene Parsing Network (PSPNet), Efficient Neural Network (Enet), and DeepLab Version 3 (DeepLabV3). Through this experiment involving hyperparameter tuning, we evaluated each model's performance having metrics such as Accuracy, Mean Intersection over Union (IoU), and Dice Coefficient. Notably, our conclusion shows that the Seg-Net model attains a remarkable accuracy of 96%. This study advances automated histopathology image processing and shows that deep learning is effective for better nuclei segmentation.