An approach for leukemia classification based on cooperative game theory

Area of Science:

• Hematology
• Computational Biology
• Medical Diagnostics

Background:

• Hematological malignancies, including leukemia, lymphoma, and multiple myeloma, affect blood, bone marrow, and lymph nodes.
• Leukemia is a prevalent blood cancer originating in bone marrow, necessitating advanced diagnostic techniques for improved patient prognosis.
• Accurate classification of leukemia subtypes is crucial for effective treatment strategies.

Purpose of the Study:

• To develop an automatic diagnosis recommender system for classifying leukemia using cooperative game theory.
• To analyze flow cytometry data for the classification of leukemia into eight distinct subtypes.
• To evaluate the proposed system's performance against existing classification methods.

Main Methods:

• Utilized a cooperative game framework to classify leukemia based on weighted markers from flow cytometry data.
• Analyzed a dataset of 400 human leukemic bone marrow samples from the Iran Blood Transfusion Organization (IBTO).
• Compared the cooperative game classification accuracy with the decision tree (C4.5) algorithm.

Main Results:

• The cooperative game system achieved a classification accuracy of 93.12%.
• This accuracy surpasses the decision tree (C4.5) method, which yielded 90.16% accuracy.
• The proposed method demonstrated versatility and applicability to various data classification tasks.

Conclusions:

• Cooperative game theory is a promising approach for direct leukemia classification from flow cytometry data.
• The developed system can serve as an active medical decision support tool for interpreting flow cytometry readouts.
• This automated system can assist clinical hematologists in leukemia diagnosis, potentially enhancing patient treatment and outcomes.