Review on Computer Vision in Gastric Cancer: Potential Efficient Tools for Diagnosis

TL;DR

This review summarizes recent advances in computer vision techniques for gastric cancer diagnosis, highlighting methods for data augmentation, feature extraction, and classification to improve accuracy and efficiency in clinical settings.

Contribution

It provides a comprehensive overview of recent computer vision methods applied to gastric cancer diagnosis, including data handling, feature extraction, and segmentation approaches, with comparative analysis.

Findings

Various data augmentation techniques improve model robustness.

Deep learning methods outperform traditional approaches in accuracy.

Application of these methods reduces diagnosis time and labor.

Abstract

Rapid diagnosis of gastric cancer is a great challenge for clinical doctors. Dramatic progress of computer vision on gastric cancer has been made recently and this review focuses on advances during the past five years. Different methods for data generation and augmentation are presented, and various approaches to extract discriminative features compared and evaluated. Classification and segmentation techniques are carefully discussed for assisting more precise diagnosis and timely treatment. For classification, various methods have been developed to better proceed specific images, such as images with rotation and estimated real-timely (endoscopy), high resolution images (histopathology), low diagnostic accuracy images (X-ray), poor contrast images of the soft-tissue with cavity (CT) or those images with insufficient annotation. For detection and segmentation, traditional methods and machine learning methods are compared. Application of those methods will greatly reduce the labor and time consumption for the diagnosis of gastric cancers.