Overcoming Data Limitation in Medical Visual Question Answering

TL;DR

This paper introduces a novel medical VQA framework that combines unsupervised Denoising Auto-Encoders and supervised Meta-Learning to effectively address the challenge of limited labeled data in medical image question answering.

Contribution

It proposes a new framework integrating DAE and Meta-Learning to improve medical VQA performance with scarce labeled data.

Findings

Significantly outperforms existing medical VQA methods.

Efficient training with small labeled datasets.

Leverages unlabeled images effectively.

Abstract

Traditional approaches for Visual Question Answering (VQA) require large amount of labeled data for training. Unfortunately, such large scale data is usually not available for medical domain. In this paper, we propose a novel medical VQA framework that overcomes the labeled data limitation. The proposed framework explores the use of the unsupervised Denoising Auto-Encoder (DAE) and the supervised Meta-Learning. The advantage of DAE is to leverage the large amount of unlabeled images while the advantage of Meta-Learning is to learn meta-weights that quickly adapt to VQA problem with limited labeled data. By leveraging the advantages of these techniques, it allows the proposed framework to be efficiently trained using a small labeled training set. The experimental results show that our proposed method significantly outperforms the state-of-the-art medical VQA.