MedVisionLlama: Leveraging Pre-Trained Large Language Model Layers to Enhance Medical Image Segmentation

TL;DR

This paper introduces MedVisionLlama, a novel approach that integrates pre-trained Large Language Model transformer layers into Vision Transformers to significantly improve medical image segmentation accuracy across various modalities.

Contribution

It proposes a hybrid model combining frozen LLM transformer blocks with Vision Transformers, along with a new attention mechanism and multi-scale fusion for enhanced segmentation performance.

Findings

Dice score increased from 0.74 to 0.79

Significant improvements in accuracy and precision

Effective integration of LLMs into vision models

Abstract

Large Language Models (LLMs), known for their versatility in textual data, are increasingly being explored for their potential to enhance medical image segmentation, a crucial task for accurate diagnostic imaging. This study explores enhancing Vision Transformers (ViTs) for medical image segmentation by integrating pre-trained LLM transformer blocks. Our approach, which incorporates a frozen LLM transformer block into the encoder of a ViT-based model, leads to substantial improvements in segmentation performance across various medical imaging modalities. We propose a Hybrid Attention Mechanism that combines global and local feature learning with a Multi-Scale Fusion Block for aggregating features across different scales. The enhanced model shows significant performance gains, including an average Dice score increase from 0.74 to 0.79 and improvements in accuracy, precision, and the Jaccard Index. These results demonstrate the effectiveness of LLM-based transformers in refining medical image segmentation, highlighting their potential to significantly boost model accuracy and robustness. The source code and our implementation are available at: https://github.com/AS-Lab/Marthi-et-al-2025-MedVisionLlama-Pre-Trained-LLM-Layers-to-Enhance-Medical-Image-Segmentation