Automated Decision Support System for Detection of Leukemia from Peripheral Blood Smear Images

Area of Science:

• Hematology
• Medical Image Analysis
• Computational Pathology

Background:

• Peripheral blood smear analysis is crucial for diagnosing various diseases, including cancers like leukemia.
• Leukemia is characterized by an abnormal increase in white blood cells (WBCs) in peripheral blood.
• Visual identification of leukemic WBCs can be challenging and subjective.

Purpose of the Study:

• To develop and evaluate an automated method for detecting leukemia using image processing techniques.
• To classify white blood cells (WBCs) as normal or abnormal and identify leukemic cells.
• To further classify normal WBCs into their subtypes.

Main Methods:

• Acquisition of 1159 Leishman-stained peripheral blood smear images with varying brightness and color.
• Utilizing Support Vector Machine (SVM) for classifying WBCs into normal and abnormal categories and detecting leukemic cells.
• Employing a Neural Network (NN) classifier for subtyping normal WBCs.

Main Results:

• Achieved an overall classification accuracy of 98.8% for WBC classification.
• Demonstrated the effectiveness of combining SVM and NN classifiers for accurate leukemia detection.
• Successfully differentiated normal WBC subtypes from abnormal, potentially leukemic cells.

Conclusions:

• The proposed automated image processing method shows high accuracy in detecting leukemia from peripheral blood smears.
• The combination of SVM and NN classifiers provides a robust approach for hematological analysis.
• This automated system has the potential to aid in the early and accurate diagnosis of leukemia.