Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

KAN-Mamba FusionNet: Redefining Medical Image Segmentation with Non-Linear Modeling

Akansh Agrawal, Akshan Agrawal, Shashwat Gupta, Priyanka Bagade

TL;DR

The paper tackles the challenge of accurate medical image segmentation by bridging non-linear modeling and long-range dependency handling. It introduces KAN-Mamba FusionNet, which integrates KAN-based non-linear blocks with a Mamba-inspired KAMBA block that includes BoA, SSM, and spatial attention, within a U-KAN-inspired pipeline. Key contributions include the novel KAMBA block, the BoA activation scheme, and comprehensive ablations and evaluations on BUSI, Kvasir-Seg, and GlaS. Results show consistent improvements in IoU and F1 across datasets with manageable computational costs, suggesting practical impact for clinical image analysis.

Abstract

Medical image segmentation is essential for applications like robotic surgeries, disease diagnosis, and treatment planning. Recently, various deep-learning models have been proposed to enhance medical image segmentation. One promising approach utilizes Kolmogorov-Arnold Networks (KANs), which better capture non-linearity in input data. However, they are unable to effectively capture long-range dependencies, which are required to accurately segment complex medical images and, by that, improve diagnostic accuracy in clinical settings. Neural networks such as Mamba can handle long-range dependencies. However, they have a limited ability to accurately capture non-linearities in the images as compared to KANs. Thus, we propose a novel architecture, the KAN-Mamba FusionNet, which improves segmentation accuracy by effectively capturing the non-linearities from input and handling long-range dependencies with the newly proposed KAMBA block. We evaluated the proposed KAN-Mamba FusionNet on three distinct medical image segmentation datasets: BUSI, Kvasir-Seg, and GlaS - and found it consistently outperforms state-of-the-art methods in IoU and F1 scores. Further, we examined the effects of various components and assessed their contributions to the overall model performance via ablation studies. The findings highlight the effectiveness of this methodology for reliable medical image segmentation, providing a unique approach to address intricate visual data issues in healthcare.

KAN-Mamba FusionNet: Redefining Medical Image Segmentation with Non-Linear Modeling

TL;DR

The paper tackles the challenge of accurate medical image segmentation by bridging non-linear modeling and long-range dependency handling. It introduces KAN-Mamba FusionNet, which integrates KAN-based non-linear blocks with a Mamba-inspired KAMBA block that includes BoA, SSM, and spatial attention, within a U-KAN-inspired pipeline. Key contributions include the novel KAMBA block, the BoA activation scheme, and comprehensive ablations and evaluations on BUSI, Kvasir-Seg, and GlaS. Results show consistent improvements in IoU and F1 across datasets with manageable computational costs, suggesting practical impact for clinical image analysis.

Abstract

Medical image segmentation is essential for applications like robotic surgeries, disease diagnosis, and treatment planning. Recently, various deep-learning models have been proposed to enhance medical image segmentation. One promising approach utilizes Kolmogorov-Arnold Networks (KANs), which better capture non-linearity in input data. However, they are unable to effectively capture long-range dependencies, which are required to accurately segment complex medical images and, by that, improve diagnostic accuracy in clinical settings. Neural networks such as Mamba can handle long-range dependencies. However, they have a limited ability to accurately capture non-linearities in the images as compared to KANs. Thus, we propose a novel architecture, the KAN-Mamba FusionNet, which improves segmentation accuracy by effectively capturing the non-linearities from input and handling long-range dependencies with the newly proposed KAMBA block. We evaluated the proposed KAN-Mamba FusionNet on three distinct medical image segmentation datasets: BUSI, Kvasir-Seg, and GlaS - and found it consistently outperforms state-of-the-art methods in IoU and F1 scores. Further, we examined the effects of various components and assessed their contributions to the overall model performance via ablation studies. The findings highlight the effectiveness of this methodology for reliable medical image segmentation, providing a unique approach to address intricate visual data issues in healthcare.

Paper Structure

This paper contains 12 sections, 10 equations, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. KAN Block
  4. KAMBA Block
  5. Convolution Block
  6. Datasets
  7. Experiments and Results
  8. Implementation Details
  9. Evaluation Details
  10. Performance Comparisons With State-of-Art Methods
  11. Ablation Studies
  12. Conclusion

Figures (2)

  • Figure 1: An overview of the proposed KAN-Mamba FusionNet Architecture with the detailed framework of KAMBA block.
  • Figure 2: Visualized segmentation results on three distinct datasets: BUSI, Kvasir-Seg, and GlaS. For each dataset, an input image, the corresponding ground truth mask, and the outputs from SOTA methods and the proposed KAN-Mamba FusionNet model are shown.