Diagnosing Microscopic Blood Samples for Early Detection of Leukemia by Deep and Hybrid Learning Techniques
Ebrahim Mohammed Senan, Mukti E. Jadhav, Ramesh R. Manza, Vandana C. Bagal · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2023
Blood is an important component of the human, which consists of many important components including White Blood Cells (WBC).Leukaemia is one of the dangerous kinds of cancer that affect the blood and bone marrow, affecting children and adults.Acute lymphoblastic Leukaemia (ALL) is dangerous and deadly type of blood cancer.Hematologists and experts work on diagnosing blood by taking patient samples and analyzing them with a high-quality magnifying lens.However, manual diagnosis is boring, time-consuming, and more prone to errors and differing expert views.Therefore, artificial intelligence techniques solve this problem and support the opinions of highly experienced experts.This research aims to develop diagnostic systems using a Convolutional Neural Network (CNN) and a hybrid CNN and SVM to diagnose the ALL_IDB2 dataset for early diagnosis of Leukaemia.CNN models and a hybrid technique consisting of two blocks were implemented, the first block of CNN models to extract feature and the second block, the SVM algorithm, to classify the feature.All the proposed systems achieved superior results in diagnosing the ALL_IDB2 dataset for early diagnosis of Leukaemia.