MedKAN: An Advanced Kolmogorov-Arnold Network for Medical Image Classification

TL;DR

MedKAN introduces a novel neural network architecture based on Kolmogorov-Arnold Networks for improved medical image classification, effectively capturing complex textures and context, outperforming CNNs and Transformers across multiple datasets.

Contribution

This work presents MedKAN, a new KAN-based framework with specialized modules and scalable variants, advancing medical image analysis beyond existing CNN and Transformer models.

Findings

MedKAN outperforms CNN and Transformer models on nine datasets.

The LIK and GIK modules effectively capture fine-grained and global features.

Scalable variants adapt to different computational requirements.

Abstract

Recent advancements in deep learning for image classification predominantly rely on convolutional neural networks (CNNs) or Transformer-based architectures. However, these models face notable challenges in medical imaging, particularly in capturing intricate texture details and contextual features. Kolmogorov-Arnold Networks (KANs) represent a novel class of architectures that enhance nonlinear transformation modeling, offering improved representation of complex features. In this work, we present MedKAN, a medical image classification framework built upon KAN and its convolutional extensions. MedKAN features two core modules: the Local Information KAN (LIK) module for fine-grained feature extraction and the Global Information KAN (GIK) module for global context integration. By combining these modules, MedKAN achieves robust feature modeling and fusion. To address diverse computational needs, we introduce three scalable variants--MedKAN-S, MedKAN-B, and MedKAN-L. Experimental results on nine public medical imaging datasets demonstrate that MedKAN achieves superior performance compared to CNN- and Transformer-based models, highlighting its effectiveness and generalizability in medical image analysis.