Scholarlink

Colonoscopy is widely regarded as the most efficient method for identifying and eliminating colorectal polyps. Early colonoscopy has led to a massive reduction in the occurrence of colorectal cancer. This paper proposes SegFormer, a model which unifies refined Vision Transformers with a compact decoder containing multilayer perceptron layers only. This model is able to segments polyp regions at pixel-level, furthermore classifies those into benign or malignant lables with high accuracy. The study is conducted on BKAI-IGH NeoPolyp-Small, a public dataset released by BKAI, Hanoi University of Science and Technology incorporation with Institute of Gastroenterology and Hepatology (IGH), Vietnam. The model achieves up to 93.3% mDice and 87.7% mloU on this dataset.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Polyps Segmentation in Colonoscopy Images Using SegFormer Transformer

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Truong Hoang Son,
  • Phan Duy Hung

摘要

Colonoscopy is widely regarded as the most efficient method for identifying and eliminating colorectal polyps. Early colonoscopy has led to a massive reduction in the occurrence of colorectal cancer. This paper proposes SegFormer, a model which unifies refined Vision Transformers with a compact decoder containing multilayer perceptron layers only. This model is able to segments polyp regions at pixel-level, furthermore classifies those into benign or malignant lables with high accuracy. The study is conducted on BKAI-IGH NeoPolyp-Small, a public dataset released by BKAI, Hanoi University of Science and Technology incorporation with Institute of Gastroenterology and Hepatology (IGH), Vietnam. The model achieves up to 93.3% mDice and 87.7% mloU on this dataset.