Features based on the percolation theory for quantification of non-Hodgkin lymphomas

Area of Science:

• Oncology
• Computational Pathology
• Image Analysis

Background:

• Non-Hodgkin lymphomas affect over 70,000 people annually in the US.
• Early diagnosis and accurate classification are crucial for effective treatment.
• Classifying non-Hodgkin lymphomas presents significant challenges for experts and computer vision methods.

Purpose of the Study:

• To introduce a novel method for quantifying and classifying non-Hodgkin lymphoma tissue samples.
• To leverage percolation theory for improved diagnostic accuracy.

Main Methods:

• A multiscale, multidimensional approach was used to segment images into smaller regions.
• Color similarity between pixels was verified within these regions.
• A cluster labeling algorithm quantified clusters, percolation, and largest cluster coverage.

Main Results:

• The method achieved high Area Under the Curve (AUC) rates, ranging from 0.940 to 0.993.
• Results were comparable to established methods in the literature.
• Percolation theory proved effective in differentiating three specific lymphoma types.

Conclusions:

• Percolation theory is a viable and effective approach for classifying non-Hodgkin lymphomas.
• The proposed method shows promise for improving the diagnosis of mantle cell lymphoma, follicular lymphoma, and chronic lymphocytic leukemia.