Machine Learning-Based Detection of Blood Cancer from Microscopic Images: A Comparative Study
Milan Manoj, Vishnu Sreekumar, Surya Reghuram, Tricha Anjali, Nandakishor Prabhu Ramlal · 2024
Early and accurate detection of blood cancer is critical for improving patient outcomes. This article is proposed to develop and evaluate the effectiveness of using machine learning techniques for automation in blood cancer detection based on microscopic images of blood cells. A comparison will be conducted using three of the most used models: Support Vector Machine, Decision Tree, and Random Forest on their capability in the classification process of distinguishing blood cells from benign and malig-nant groups. The methodology is divided into the following steps: image preprocessing, extraction of color histogram and texture features through the Gray-Level Co-occurrence Matrix (GLCM), and optimization of hyperparameters for model training. Evaluation metrics used are accuracy, precision, recall, and F1-score. Random Forest has been seen to perform with a high accuracy of 99.08% with all the categories showing robust classification. It was followed by SVM with an accuracy of 96.15 % and Decision Tree with 94.92%. The results present Random Forest as a valuable resource for the automated diagnosis of blood cancer. This paper reiterates the promise that machine learning holds in boosting diagnostic efficiency and encourages future studies into more advanced approaches such as deep learning for further improvement in clinical applications.