Data

Tools

Indexes

Applications

Experiments

Open Source Science

https://github.com/alimzade/ai-histopathology-ovarian-cancer

Code and resources for AI-driven ovarian cancer subtype classification using histopathology images (WSI and TMA). This repository includes preprocessing scripts, MIL-based WSI classification, TMA with Encoder as classifier using external dataset and TMA majority voting approach, along with evaluation metrics and visualizations.

https://github.com/alimzade/ai-histopathology-ovarian-cancer

Science Score: 13.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ○
    CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ○
    .zenodo.json file
  • ○
    DOI references
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (9.1%) to scientific vocabulary
Last synced: 10 months ago · JSON representation

Repository

Code and resources for AI-driven ovarian cancer subtype classification using histopathology images (WSI and TMA). This repository includes preprocessing scripts, MIL-based WSI classification, TMA with Encoder as classifier using external dataset and TMA majority voting approach, along with evaluation metrics and visualizations.

Basic Info
  • Host: GitHub
  • Owner: Alimzade
  • Language: Jupyter Notebook
  • Default Branch: main
  • Size: 12.4 MB
Statistics
  • Stars: 0
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created over 1 year ago · Last pushed over 1 year ago
Metadata Files
Readme

README.md

AI for Histopathology: Ovarian Cancer Subtype Classification

Abstract

This repository contains code for ovarian cancer subtype classification using Whole Slide Images (WSI) and Tissue Microarrays (TMA). We explore multiple approaches, including a Multiple Instance Learning (MIL) classifier for WSI, encoder and tile-based majority voting for TMA. The MIL classifier demonstrates strong performance, while TMA methods show potential despite generalization challenges. This study emphasizes AI's role in advancing diagnostics and calls for larger datasets and innovative architectures to improve precision and patient outcomes.

Details

This study was conducted as part of the "Applied AI Engineering Lab" course at the University of Passau. It focuses on classifying ovarian cancer subtypes using Whole Slide Images (WSI) and Tissue Microarrays (TMA) data from the UBC-OCEAN Kaggle competition. The task is to classify ovarian cancer images into the following subtypes:

  • Clear Cell Carcinoma (CC)
  • Endometrioid Carcinoma (EC)
  • High-grade Serous Carcinoma (HGSC)
  • Low-grade Serous Carcinoma (LGSC)
  • Mucinous Carcinoma (MC)
  • Other (Note: This class is absent in the training set but present in the test set.)

The dataset consists of images from different hospitals, with the test set containing images from institutions not represented in the training set, adding a challenge of domain shift.

OCEAN-Optional-Figure

Data Description

  • WSI (Whole Slide Images): Large images (up to 100,000 x 50,000 pixels) at 20x magnification. The average file size is 1-2 GB.
  • TMA (Tissue Microarrays): Smaller images (~4,000 x 4,000 pixels) at 40x magnification, but there are relatively fewer TMA samples in the dataset.

Results

Best Approach: The MIL classifier approach, treating WSI images as bags of instances, yielded the best performance with 80% accuracy.

Owner

  • Login: Alimzade
  • Kind: user

GitHub Events

Total
  • Watch event: 1
  • Member event: 1
  • Push event: 4
  • Create event: 2
Last Year
  • Watch event: 1
  • Member event: 1
  • Push event: 4
  • Create event: 2