Deep Model with Siamese Network for Viability and Necrosis Tumor Assessment in Osteosarcoma

TL;DR

This paper introduces a deep learning model using Siamese networks to automatically classify osteosarcoma histology images, aiming to assist pathologists in diagnosis by improving accuracy and efficiency.

Contribution

The study presents a novel deep Siamese network architecture specifically designed for osteosarcoma image classification, addressing morphological variability.

Findings

Achieved high classification accuracy on osteosarcoma histology images

Reduced diagnostic time for pathologists

Demonstrated effectiveness of Siamese networks in medical image analysis

Abstract

Osteosarcoma is the most common primary malignant bone tumor, which has high mortality due to easy lung metastasis. Osteosarcoma is a highly anaplastic, pleomorphic tumor with a variety of tumor cell morphology, including fusiform, oval, epithelial, lymphocyte like, small round, transparent cells, etc. Due to the multiple patterns of osteosarcoma cell morphology, pathologists have differences in the classification (viable tumor, necrotic tumor, non-tumor) of osteosarcoma. Therefore, automatic and accurate recognition algorithms can help pathologists greatly reduce time and improve diagnostic accuracy. In recent years, deep learning technology has made great progress in the field of natural images and medical images, and has achieved excellent results beyond human performance in classification. In this paper, we propose a Deep Model with Siamese Network (DS-Net) for automatic classification in Hematoxylin and Eosin (H&E) stained osteosarcoma histology images.