An extensive comparison of cutting-edge image processing and machine
learning methods for leukemia identification and classification is presented in this
chapter. The proposed approach incorporates a thorough, extensive analytical approach
that includes morphological operations, watershed segmentation, Wiener filtering, Kmeans clustering, and Gaussian filtration. By fine-tuning microscopic medical images,
these preprocessing techniques enable accurate feature extraction and categorization.
Using a multi-class Support Vector Machine (SVM) model, a total of 20 sub-features,
including morphological, visual, and statistical characteristics, are retrieved and
categorized. The proposed approach effectively classifies Acute Lymphoblastic
Leukemia (ALL), Acute Myeloid Leukemia (AML), Chronic Lymphocytic Leukemia
(CLL), Chronic Myeloid Leukemia (CML), and normal cells with a detection accuracy
of 97.06% when tested on a dataset of 250 samples. To demonstrate the efficacy of the
proposed approach, it is compared to state-of-the-art methods. The current method
achieves 91.2% accuracy for ALL. By providing a scalable, effective, and precise
method for leukemia identification and categorization, this study discusses advanced
computer-aided diagnostic tools. The findings highlight how machine learning and
image processing technology may enhance medical diagnostics by helping medical
practitioners identify leukemia subtypes early and accurately.
Keywords: Classification, Deep learning, Leukemia, Lymphocytes, Neoplasm, Segmentation, White blood cells.
