PathViT Model for Automated Disease Classification from Skeletal Muscle Histopathology

Area of Science:

• Biomedical Engineering
• Computational Pathology
• Digital Health

Background:

• Manual analysis of skeletal muscle pathology from histological images is time-consuming and subjective.
• Inter- and intra-user variability in manual analysis impacts diagnostic accuracy and consistency.
• Current methods require manual cell counting, segmentation, and thresholding, increasing labor intensity.

Purpose of the Study:

• To develop an automated deep-learning model, PathViT, for skeletal muscle pathology analysis.
• To reduce human intervention, subjectivity, and variability in muscle fiber classification.
• To significantly decrease analysis time compared to conventional manual methods.

Main Methods:

• Utilized wheat germ agglutinin staining and digital histopathology of skeletal muscle.
• Employed a transformer-based deep-learning model (PathViT) for automated classification.
• Compared PathViT performance against state-of-the-art deep-learning models.

Main Results:

• PathViT achieved 96% accuracy in classifying healthy versus diseased muscle fibers.
• The model outperformed other deep-learning models in distinguishing muscle pathologies.
• PathViT demonstrated enhanced scalability and decreased variability in analysis.

Conclusions:

• PathViT offers a powerful, automated solution for skeletal muscle pathology analysis.
• The model improves diagnostic accuracy and consistency, reducing reliance on manual methods.
• PathViT has potential as a valuable tool for biomedical research and clinical settings.