Related Experiment Video
Updated: Oct 20, 2025
07:10
In Vivo Vascular Permeability Detection in Mouse Submandibular Gland
Published on: August 4, 2022
2.2K
Two-phase non-invasive multi-disease detection via sublingual region
Jianhang Zhou1, Qi Zhang2, Bob Zhang2
1PAMI Research Group, Dept. of Computer and Information Science, University of Macau, Taipa, Macau, China; Shenzhen Research Institute of Big Data, Shenzhen, 518172, China.
Computers in Biology and Medicine
|September 14, 2021
Summary
This study introduces a novel two-phase framework for non-invasive multi-disease detection using sublingual vein imaging. The method achieved high accuracy, demonstrating the potential of sublingual vein analysis for automated health status assessment.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Non-invasive multi-disease detection is crucial for early diagnosis and health monitoring.
- The sublingual vein serves as a key indicator of health status, yet quantitative analysis is underexplored.
- Current methods lack comprehensive quantitative approaches for sublingual vein-based disease detection.
Purpose of the Study:
- To propose and evaluate a novel two-phase framework for non-invasive multi-disease detection using sublingual vein images.
- To investigate the effectiveness of multi-feature representations for improved disease detection accuracy.
- To establish a quantitative method for analyzing sublingual vein characteristics for health status assessment.
Main Methods:
- A two-phase framework involving sublingual vein region segmentation and feature extraction.
- Generation of multi-feature representations including color, texture, shape, and latent representations.
- Application of multi-class classification on extracted features for disease detection using 1103 sublingual vein images.
Main Results:
- The proposed framework successfully segmented sublingual vein regions.
- Multi-feature representation, particularly combining color, texture, and latent features, significantly enhanced detection.
- The best performing model achieved a high accuracy of 98.05% in multi-disease detection.
Conclusions:
- The developed two-phase framework offers a promising quantitative approach for non-invasive multi-disease detection.
- Sublingual vein imaging combined with AI-powered feature analysis can accurately reflect patient health status.
- This method holds potential for automated, early disease detection and health monitoring.
