Wild Horse Optimizer with Deep Learning Based Leukemia Diagnosis on Microscopic Blood Smear Images

Leukemia is generally analysed by inspecting the smears of blood and bone marrow utilizing microscopes and difficult Cytochemical tests. Then these approaches are expensive, slow, and affected by knowledge and ability of the professionals concerned. Leukemia is identified with the use of image processing-related approaches to analyse microscopic smear images for detecting the occurrence of leukemic cells and such approaches are easy, quick, low-cost, and not biased by professionals. Several researchers are concentrated on machine learning (ML) based automatic classification and detection of acute leukaemia and its subtypes for promoting allow extremely accurate analysis. This paper presents a Wild Horse Optimizer with Deep Learning based Leukemia Diagnosis on Microscopic Blood Smear Images (WHODL-LDBSI) technique. The aim of the WHODL-LDBSI technique is to diagnose different kinds of leukemia on blood smear images using DL model. In the initial stage, the contrast enhancement process is carried out as a preprocessing step. Next, MobileNetv2 model was utilized for deriving features in the blood smear images and its efficacy can be improvised by WHO based hyperparameter optimizer. For leukemia classification, the deep residual network (DRN) model is applied with whale optimization algorithm (WOA) as hyperparameter tuning technique. The experimental outcome of the WHODL-LDBSI system takes place on open access dataset and the obtained values portrayed outperformed other DL models.