ConvSegNet: Automated Polyp Segmentation From Colonoscopy Using Context Feature Refinement With Multiple Convolutional Kernel Sizes

Abstract

Colorectal cancer occurs in the rectal of humans, and early detection has been proved to reduce its mortality rate. Colonoscopy is the standard used in detecting the presence of polyps in the rectal, and accurate segmentation of the polyps from colonoscopy images often provides helpful information for early diagnosis and treatment. Although existing deep learning models often achieve high segmentation performance when tested on the same dataset used in model training; still, their performance often degrades when applied to out-of-distribution datasets, leading to low model generalization or overfitting. This challenge is often associated with the quality of the features learnt from the input images. In this work, a novel Context Feature Refinement (CFR) module is proposed to address the challenge of low model generalization and segmentation performance. The CFR module is built to extract contextual information from the incoming feature map by using multiple parallel convolutional layers with progressively increasing kernel sizes. Using multiple parallel convolutions with different kernel sizes helped to extract more efficient multi-scale contextual information and thus enabled the network to effectively identify and segment small and fine details, as well as larger and more complex structures in the input images. Extensive experiments on three public benchmark datasets in CVC-ClinicDB, Kvasir-SEG, and BKAI-NeoPolyp showed that the proposed ConvSegNet model achieved jaccard, dice and F2 scores of 0.8650, 0.9177, and 0.9328 on CVC-ClinicDB, 0.7936, 0.8618, and 0.8855 on Kvasir-SEG, and 0.8045, 0.8747 and 0.8909 on BKAI-NeoPolyp datasets respectively. Also, an improved generalization performance was achieved by the ConvSegNet model, compared to the benchmark polyp segmentation models.

Lightweight colon polyp segmentation algorithm based on improved DeepLabV3+

“…To test the superiority of the improved algorithm, Kvasir-SEG, and CVC-ClinicDB datasets are used to compare three segmentation algorithms, Unet 26 , SegFormer 27 , and ConvSegNet 11 . At the same time, two cross-dataset experiments are carried out to explore the generalization ability of the improved model.…”

“…Ige et al proposed a Context Feature Refinement (CFR) module to solve the challenges of model generalization and low segmentation performance. By using multiple parallel convolution layers to extract context information from the incoming feature map and gradually increasing the kernel size, the network can effectively identify and segment fine details in the input image 11 . To address the challenge of having no obvious boundary between polyps and their surroundings, Zhou et al proposed a cross-level feature aggregation network (CFA-Net) for polyp segmentation, which consists of a boundary prediction network and a polyp segmentation network.…”

Automated Detection of Colorectal Polyp Utilizing Deep Learning Methods With Explainable AI

“…The need for immediate action emphasizes the requirement for aggressive and assertive steps to address this situation directly. The growth stage, depicted in Figure 1, initiates as polyps, which are diminutive, non-cancerous growths that have the potential to transform into cancer over time [3].…”

Polyp Segmentation With the FCB-SwinV2 Transformer

“…[16] [17] [18] [19] [20] [21] [22][23]. These models often incorporate advanced ImageNet pre-trained CNNs into the encoder, significantly improving the efficiency and accuracy of feature extraction by leveraging pre-learned image representations.…”