| Abstrakt: |
Leukemia, often called blood cancer, is a disease that primarily affects white blood cells (WBCs), which harms a person's tissues and plasma. This condition may be fatal when if it is not diagnosed and recognized at an early stage. The physical technique and lab procedures for Leukaemia identification are considered time-consuming. It is crucial to use a quick and unexpected way to identify different forms of Leukaemia. Timely screening of the morphologies of immature cells is essential for reducing the severity of the disease and reducing the number of people who require treatment. Various deep-learning (DL) model-based segmentation and categorization techniques have already been introduced, although they still have certain drawbacks. In order to enhance feature extraction and classification in such a practical way, Mayfly optimization with Generative Adversarial Network (MayGAN) is introduced in this research. Furthermore, Generative Adversarial System (GAS) is integrated with Principal Component Analysis (PCA) in the feature-extracted model to classify the type of blood cancer in the data. The semantic technique and morphological procedures using geometric features are used to segment the cells that makeup Leukaemia. Acute lymphocytic Leukaemia (ALL), acute myelogenous Leukaemia (AML), chronic lymphocytic Leukaemia (CLL), chronic myelogenous Leukaemia (CML), and aberrant White Blood Cancers (WBCs) are all successfully classified by the proposed MayGAN model. The proposed MayGAN identifies the abnormal activity in the WBC, considering the geometric features. Compared with the state-of-the-art methods, the proposed MayGAN achieves 99.8% accuracy, 98.5% precision, 99.7% recall, 97.4% F1-score, and 98.5% Dice similarity coefficient (DSC). [ABSTRACT FROM AUTHOR] |
