Detection and Estimation of Leukemia Using Artificial Intelligence and Fuzzy Logic

Leukemia often leads to bone marrow failure, resulting in a shortage of healthy white blood cells. Detecting changes in white blood cell count serves as a critical diagnostic indicator for leukemia. Authors ongoing study involves the classification of blood smear micrographs, employing various techniques such as pre-processing, segmentation, and a feature-based classification algorithm. Authors research primarily focuses on early leukemia detection and prognosis assessment. Authors explore existing contributions by multiple authors and propose a promising solution with potential future applications. Acute myeloid leukemia (AML), the most prevalent form of blood cancer, can now be diagnosed using artificial intelligence, greatly enhancing diagnostic accuracy. Authors methodology is centered on gene expression analysis in blood samples to mitigate the potential catastrophic effects of leukemia. Deep learning techniques, specifically convolutional neural networks, play a pivotal role in eliminating human error from the process. The model is trained using cell images, enabling effective data pre-processing and extraction. In authors study, authors employed ResNet and pre-trained models, as well as a hybrid model combining VGG19 and VGG16. Additionally, authors introduced a hybrid model merging artificial neural networks with fuzzy logic. To address the low accuracy of ResNet, which was approximately 97%, authors developed a hybrid approach utilizing artificial neural networks for classification and fuzzy logic for feature extraction, ultimately achieving an impressive accuracy rate of 99%.