GPT4MIA: Utilizing Generative Pre-trained Transformer (GPT-3) as A Plug-and-Play Transductive Model for Medical Image Analysis

TL;DR

GPT4MIA leverages GPT-3 as a plug-and-play transductive inference tool for medical image analysis, enhancing prediction accuracy and error detection by integrating language models with vision-based models.

Contribution

This work introduces a novel method to use GPT-3 as a transductive inference model for medical image analysis, with specific techniques to improve its efficiency and effectiveness.

Findings

Effective in detecting prediction errors in medical images

Improves prediction accuracy when combined with vision models

Validated through experiments on real medical image datasets

Abstract

In this paper, we propose a novel approach (called GPT4MIA) that utilizes Generative Pre-trained Transformer (GPT) as a plug-and-play transductive inference tool for medical image analysis (MIA). We provide theoretical analysis on why a large pre-trained language model such as GPT-3 can be used as a plug-and-play transductive inference model for MIA. At the methodological level, we develop several technical treatments to improve the efficiency and effectiveness of GPT4MIA, including better prompt structure design, sample selection, and prompt ordering of representative samples/features. We present two concrete use cases (with workflow) of GPT4MIA: (1) detecting prediction errors and (2) improving prediction accuracy, working in conjecture with well-established vision-based models for image classification (e.g., ResNet). Experiments validate that our proposed method is effective for these two tasks. We further discuss the opportunities and challenges in utilizing Transformer-based large language models for broader MIA applications.