Texture-based Feature Extraction and Machine Learning Model for the Detection of Acute Lymphoblastic Leukemia

Acute Lymphocytic Leukaemia (ALL) is a dangerous malignancy that not only attacks adults but also contributes to roughly 25% of children. A precise and on-time diagnosis of cancer is a critical prerequisite for effective therapy to enhance survival rates. It is challenging to identify cancer cells from normal cells because leukemic cells seem extremely identical to normal cells beneath the microscope. Normally, morphological analysis for the detection of ALL is conducted physically on blood cells by qualified medical individuals, which has various drawbacks, like a lack of medical experts, slow analysis, and prognosis which depends on the knowledge of medical professionals. For this reason, we provide a robust method for improving the reliability of ALL identification using feature extraction and a Machine Learning (ML) model to discriminate between cancerous and healthy cells in images. To extract texture features for analysis, we employ the First Order Histogram (FOH) and the Gray-Level Co-occurrence Matrix (GLCM) in our technique. Logistic Regression (LR), Naive Bayes (NB), and Artificial Neural Network (ANN) are used in the categorizing method. The simulation results demonstrate that combining GLCM feature extraction with the ANN model yields the highest accuracy, correctly predicting 595 out of 600 data samples. The accuracy was 99.17%, which is quite remarkable.