Learning Deep Representations of Medical Images using Siamese CNNs with Application to Content-Based Image Retrieval

TL;DR

This paper introduces a deep Siamese CNN architecture for medical image representation learning that requires less supervision and performs comparably to traditional supervised CNNs in content-based image retrieval tasks.

Contribution

A novel deep Siamese CNN approach that learns medical image representations with minimal supervision using binary image pairs.

Findings

Comparable retrieval performance to state-of-the-art CNNs

Requires significantly less labeled data for training

Effective in content-based medical image retrieval

Abstract

Deep neural networks have been investigated in learning latent representations of medical images, yet most of the studies limit their approach in a single supervised convolutional neural network (CNN), which usually rely heavily on a large scale annotated dataset for training. To learn image representations with less supervision involved, we propose a deep Siamese CNN (SCNN) architecture that can be trained with only binary image pair information. We evaluated the learned image representations on a task of content-based medical image retrieval using a publicly available multiclass diabetic retinopathy fundus image dataset. The experimental results show that our proposed deep SCNN is comparable to the state-of-the-art single supervised CNN, and requires much less supervision for training.