Multi Kernel Positional Embedding ConvNeXt for Polyp Segmentation

TL;DR

This paper introduces a novel ConvNeXt-based framework with Multi Kernel Positional Embedding for improved polyp segmentation in medical images, achieving higher accuracy and better generalization than existing methods.

Contribution

The paper proposes a new framework combining ConvNeXt and a Multi Kernel Positional Embedding block to enhance polyp segmentation performance.

Findings

Achieved Dice coefficient of 0.8818 on Kvasir-SEG dataset.

Attained IOU score of 0.8163, outperforming some previous methods.

Demonstrated strong generalization across multiple datasets.

Abstract

Medical image segmentation is the technique that helps doctor view and has a precise diagnosis, particularly in Colorectal Cancer. Specifically, with the increase in cases, the diagnosis and identification need to be faster and more accurate for many patients; in endoscopic images, the segmentation task has been vital to helping the doctor identify the position of the polyps or the ache in the system correctly. As a result, many efforts have been made to apply deep learning to automate polyp segmentation, mostly to ameliorate the U-shape structure. However, the simple skip connection scheme in UNet leads to deficient context information and the semantic gap between feature maps from the encoder and decoder. To deal with this problem, we propose a novel framework composed of ConvNeXt backbone and Multi Kernel Positional Embedding block. Thanks to the suggested module, our method can attain better accuracy and generalization in the polyps segmentation task. Extensive experiments show that our model achieves the Dice coefficient of 0.8818 and the IOU score of 0.8163 on the Kvasir-SEG dataset. Furthermore, on various datasets, we make competitive achievement results with other previous state-of-the-art methods.