Detecting and characterizing cellular responses to Mycobacterium tuberculosis from histology slides

Area of Science:

• Pathology
• Computational Biology
• Immunology

Background:

• Mycobacterium tuberculosis (M.tb) infection causes immune cell accumulation in the lungs, forming distinct granulomas and lymphocytic cuffs.
• These structures are crucial for understanding the host's immune response to tuberculosis.
• Accurate identification of these regions in histology slides is essential for research.

Purpose of the Study:

• To develop a computational framework for automated detection and classification of high cell-density regions in H&E-stained lung tissue slides.
• To accurately differentiate between granulomas and lymphocytic cuffs based on morphological characteristics, primarily cell density.

Main Methods:

• Development of DeHiDe, a computational framework utilizing internuclei geodesic distance and Dulmange Mendelsohn permutation.
• Training and validation of DeHiDe on 21 digitized H&E-stained lung tissue slides from M.tb-infected mice.
• Evaluation based on detection accuracy and classification accuracy of high cell-density regions.

Main Results:

• DeHiDe achieved a high detection accuracy of 99.39% for high cell-density regions.
• The framework correctly classified 90.87% of detected regions into granulomas and lymphocytic cuffs.
• Demonstrated ability to identify these regions in complex cellular environments with non-convex shapes.

Conclusions:

• DeHiDe offers a robust computational solution for analyzing M.tb-induced lung pathology.
• The framework facilitates precise characterization of immune cell aggregates in tuberculosis research.
• Automated analysis can improve the efficiency and accuracy of histological assessment in infectious disease studies.