Related Experiment Video
Updated: Oct 2, 2025
07:40
Visualizing Scar Development Using SCAD Assay - An Ex-situ Skin Scarring Assay
Published on: April 28, 2022
3.0K
Automated Structural Analysis and Quantitative Characterization of Scar Tissue Using Machine Learning.
Luluil Maknuna1, Hyeonsoo Kim1, Yeachan Lee1
1Industry 4.0 Convergence Bionics Engineering and Marine-Integrated Biomedical Technology Center, Pukyong National University, Busan 48513, Korea.
Diagnostics (Basel, Switzerland)
|February 25, 2022
Summary
This study introduces machine learning for rapid scar tissue analysis. AI models accurately characterize collagen density and features in HE-stained tissues, improving pathological assessment.
Area of Science:
- Histopathology
- Computational pathology
- Biomedical imaging analysis
Background:
- Scar tissue analysis is crucial for understanding wound healing pathology.
- Conventional Hematoxylin and eosin (HE) staining lacks automated whole-slide analysis capabilities.
- Objective histological assessment of scar tissue requires advanced methods.
Purpose of the Study:
- To develop rapid, automated methods for scar lesion characterization in HE-stained tissues.
- To apply supervised and unsupervised machine learning algorithms for scar tissue analysis.
- To enable objective quantification of pathological features in scar tissue.
Main Methods:
- Utilized Mask region-based convolutional neural network (RCNN) for supervised learning with MMDetection tools.
- Employed K-means clustering for unsupervised characterization of tissue features.
- Validated model performance using various backbone networks (ResNet, ResNeSt).
Main Results:
- Mask RCNN achieved high accuracy in predicting scar images.
- K-means successfully separated collagen fibers and dermal components (glands, follicles, nuclei).
- Quantitative analysis revealed significant differences (50%) in collagen density and variance between normal and scar tissues.
Conclusions:
- The proposed machine learning methods provide an objective and time-efficient approach to scar tissue analysis.
- Automated characterization aids in understanding scar pathology and informs treatment strategies.
- This AI-assisted analysis enhances histological assessment compared to manual methods.
