Papers › AG-CUResNeSt: A Novel Method for Colon Polyp Segmentation

AG-CUResNeSt: A Novel Method for Colon Polyp Segmentation

2 May 2021arXiv:2105.00402archive 2025-07-28

Dinh Viet Sang, Tran Quang Chung, Phan Ngoc Lan, Dao Viet Hang, Dao Van Long, Nguyen Thi Thuy

Colorectal cancer is among the most common malignancies and can develop from high-risk colon polyps. Colonoscopy is an effective screening tool to detect and remove polyps, especially in the case of precancerous lesions. However, the missing rate in clinical practice is relatively high due to many factors. The procedure could benefit greatly from using AI models for automatic polyp segmentation, which provide valuable insights for improving colon polyp detection. However, precise segmentation is still challenging due to variations of polyps in size, shape, texture, and color. This paper proposes a novel neural network architecture called AG-CUResNeSt, which enhances Coupled UNets using the robust ResNeSt backbone and attention gates. The network is capable of effectively combining multi-level features to yield accurate polyp segmentation. Experimental results on five popular benchmark datasets show that our proposed method achieves state-of-the-art accuracy compared to existing methods.

PaperPDFCode

Code

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Medical Image SegmentationSegmentation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Medical Image Segmentation CVC-ClinicDB AG-CUResNeSt mean Dice 0.9170 #37 of 48 Archive leaderboard report
Medical Image Segmentation Kvasir-SEG AG-CUResNeSt mIoU 0.845 #41 of 58 Archive leaderboard report
Medical Image Segmentation Kvasir-SEG AG-CUResNeSt mean Dice 0.902 #41 of 58 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

1x1 ConvolutionAverage PoolingBatch NormalizationConvolutionDense ConnectionsGlobal Average PoolingReLUResNeStResidual ConnectionSoftmaxSplit Attention

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections