Medical visual question answering using joint self-supervised learning

TL;DR

This paper introduces a joint self-supervised learning framework for medical visual question answering, effectively leveraging large-scale image-caption data to improve performance on small medical VQA datasets.

Contribution

It proposes an encoder-decoder model that uses self-supervised pre-training on large medical image-caption data and fine-tuning on small VQA datasets, addressing data scarcity and modality diversity.

Findings

Outperforms baseline methods in medical VQA accuracy

Effective use of self-supervised learning on large-scale data

Improved generalization on small medical VQA datasets

Abstract

Visual Question Answering (VQA) becomes one of the most active research problems in the medical imaging domain. A well-known VQA challenge is the intrinsic diversity between the image and text modalities, and in the medical VQA task, there is another critical problem relying on the limited size of labelled image-question-answer data. In this study we propose an encoder-decoder framework that leverages the image-text joint representation learned from large-scaled medical image-caption data and adapted to the small-sized medical VQA task. The encoder embeds across the image-text dual modalities with self-attention mechanism and is independently pre-trained on the large-scaled medical image-caption dataset by multiple self-supervised learning tasks. Then the decoder is connected to the top of the encoder and fine-tuned using the small-sized medical VQA dataset. The experiment results present that our proposed method achieves better performance comparing with the baseline and SOTA methods.