2022
DOI: 10.1109/access.2022.3195241
|View full text |Cite
|
Sign up to set email alerts
|

ColonFormer: An Efficient Transformer Based Method for Colon Polyp Segmentation

Abstract: Identifying polyps is challenging for automatic analysis of endoscopic images in computeraided clinical support systems. Models based on convolutional networks (CNN), transformers, and their combinations have been proposed to segment polyps with promising results. However, those approaches have limitations either in modeling the local appearance of the polyps only or lack of multi-level feature representation for spatial dependency in the decoding process. This paper proposes a novel network, namely ColonForme… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
43
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
5
3
1

Relationship

0
9

Authors

Journals

citations
Cited by 109 publications
(43 citation statements)
references
References 48 publications
0
43
0
Order By: Relevance
“…During training auxiliary classification task for learning size-related and polyp number-related features was trained and embedded with the segmentation network alongside showing improvement of up to 2% over SOTA methods on four public datasets. Transformer-based networks have also been recently introduced, namely TransFuse 59 and ColonFormer 60 . TransFuse combined transformers with CNNs in a parallel style allowing capture of both global and low-level spatial details and demonstrated performance gain of nearly 1–2% on five public datasets compared to DL SOTA methods.…”
Section: Methodsmentioning
confidence: 99%
“…During training auxiliary classification task for learning size-related and polyp number-related features was trained and embedded with the segmentation network alongside showing improvement of up to 2% over SOTA methods on four public datasets. Transformer-based networks have also been recently introduced, namely TransFuse 59 and ColonFormer 60 . TransFuse combined transformers with CNNs in a parallel style allowing capture of both global and low-level spatial details and demonstrated performance gain of nearly 1–2% on five public datasets compared to DL SOTA methods.…”
Section: Methodsmentioning
confidence: 99%
“…This model includes an adaptive scale context (ASC) module with a lightweight attention mechanism, and feature pyramid fusion (FPF). In In recent years, Vision Transformer [22] (ViT) has achieved highly competitive performance for image analysis tasks in multiple applications by splitting an image into patches and tokenizing the patches for feature extraction in Transformer [23]. proposed a novel ColonFormer, integrating Transformer and CNN as a unified architecture for polyp segmentation.…”
Section: Related Workmentioning
confidence: 99%
“…Segmentation models 6,13,27,28 targeting polyp datasets have also been studied and early approaches adopted UNet models. These methods were designed to extract various colors, sizes, and boundary information of polyps in an image.…”
Section: Medical Image Segmentationmentioning
confidence: 99%
“…MCDALNet 36 used dual attention to utilize position and channel information. Recently, some models using Transformer with multi-head self-attention (MHSA) have been studied 9,28,37,38 . Polyp-PVT 9 and ColonFormer 28 used a modified transformer for medical segmentation tasks.…”
Section: Image Segmentation Using Attentionmentioning
confidence: 99%