Multi-tissue and multi-scale approach for nuclei segmentation in H&E stained images

Area of Science:

• Digital pathology
• Computational biology
• Medical imaging analysis

Background:

• Accurate nuclei detection and segmentation in histological images are vital for clinical applications.
• Manual annotation is labor-intensive and prone to variability.
• Existing automated methods often lack versatility, being specific to particular organs or tissues.

Purpose of the Study:

• To develop and validate a fully automated, multiscale method for nuclei segmentation applicable across diverse tissues and magnifications.
• To address the limitations of existing automated segmentation techniques.

Main Methods:

• Development of the Multiscale Adaptive Nuclei Analysis (MANA) algorithm.
• Testing MANA on a large dataset of H&E stained tissue images from six organs (colon, liver, bone, prostate, adrenal gland, thyroid) at 10×, 20×, and 40× magnifications.
• Comparison of MANA's performance against manual segmentations and three open-source nuclei detection software.

Main Results:

• MANA achieved an F1-score consistently above 0.91 for each organ, with an overall average F1-score of 0.9305 ± 0.0161.
• The algorithm demonstrated high efficiency, with an average computational time of approximately 20 seconds, regardless of the number of nuclei (over 1000).
• MANA's performance was comparable to or better than state-of-the-art algorithms optimized for single tissues.

Conclusions:

• MANA represents the first fully automated, multi-scale, and multi-tissue algorithm for nuclei detection.
• The method's robustness and versatility enable high performance across different organs and magnifications.
• MANA offers a significant advancement in automated nuclei segmentation for digital pathology.