Related Experiment Video
Updated: Sep 5, 2025
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
627
Colorectal polyp region extraction using saliency detection network with neutrosophic enhancement
Keli Hu1, Liping Zhao2, Sheng Feng2
1Cancer Center, Department of Gastroenterology, Zhejiang Provincial People's Hospital (Affiliated People's Hospital, Hangzhou Medical College), Hangzhou, 310014, PR China; Department of Computer Science and Engineering, Shaoxing University, Shaoxing, 312000, PR China.
Computers in Biology and Medicine
|July 8, 2022
Summary
This study introduces NeutSS-PLP, a new method for colorectal polyp detection. It uses neutrosophic enhancement and a saliency detection network to improve polyp region extraction in colonoscopy images.
Area of Science:
- Medical Imaging
- Computer Vision
- Artificial Intelligence
Background:
- Colorectal polyp recognition is vital for early cancer detection.
- Colonoscopy misses up to 25% of polyps due to visual similarity with normal tissue.
- Accurate polyp detection is essential for effective colorectal cancer screening and treatment.
Purpose of the Study:
- To present NeutSS-PLP, a novel method for polyp region extraction in colonoscopy images.
- To enhance polyp detection accuracy by addressing challenges like specular reflections.
- To improve the performance of automated polyp recognition systems.
Main Methods:
- Utilized neutrosophic theory for enhancing specular reflection detection and suppression in colonoscopy images.
- Developed local and global threshold criteria within the single-valued neutrosophic set (SVNS) domain.
- Introduced a saliency detection network with two-level short connections for multi-level and multi-scale feature extraction.
Main Results:
- Achieved mean Intersection over Union (mIoU) scores of 0.877 and 0.9135 on two public colorectal polyp datasets.
- Demonstrated superior performance compared to several state-of-the-art saliency and semantic segmentation networks.
- Validated the effectiveness of the proposed neutrosophic enhancement and saliency detection approach.
Conclusions:
- NeutSS-PLP effectively extracts colorectal polyp regions from colonoscopy images.
- The integration of neutrosophic enhancement and short-connected saliency networks improves polyp detection accuracy.
- This method shows significant potential for advancing automated polyp recognition in colorectal cancer screening.
