Automated Leukemia Diagnosis from Microscopic Blood Smear Images Using Adapted Inception V3

Leukemia, a cancer characterized by the rapid overproduction of abnormal blood cells, presents notable diagnostic and treatment challenges. Identifying leukemia subtypes promptly and accurately is vital for devising effective treatments and improving patient outcomes. This study proposes an advanced Inception V3 model optimized for the automatic identification of leukemia from microscopic images of blood smears. By integrating cutting-edge deep learning methods and a customized model architecture, this system differentiates between Acute Lymphoblastic Leukemia (ALL), Acute Myeloid Leukemia (AML), Chronic Lymphocytic Leukemia (CLL), Chronic Myeloid Leukemia (CML), and healthy samples. The model benefits from preprocessing techniques like image resizing, normalization, and augmentation to enhance its performance. Through transfer learning, the Inception V3 model is fine-tuned on a rich dataset covering various leukemia subtypes. Experimental results demonstrate the model's high effectiveness, achieving 100% accuracy during training and 99.72% accuracy in testing, proving the system's reliability for automated leukemia detection.