LeukoDeepNet: A Novel Approach in Automated Identification of Leukemia Diseases in Microscopic Human Blood Cells
Md Monwar Hossain, Najmus Sakib, Insan Ara Milu, Sanjeda Sara Jennifer, Md. Adnan Morshed, Ahmed Wasif Reza, Mohammad Shamsul Arefin · 2024
Leukemia, a critical hematologic malignancy, demands early and accurate diagnosis for effective treatment. In this research, we present a comprehensive approach to the detection of leukemia based on microscopic human blood cell images. Our research is dedicated to the early detection of leukemia by leveraging deep learning techniques on microscopic blood cell images. The study introduces a new modified VGG16 model, LeukoDeepNet, and compares it with conventional Convolutional Neural Network (CNN) and Inception (GoogLeNet) models. LeukoDeepNet stands out with remarkable training and validation accuracies of 99.96% and 92.63%, along with low training and validation losses of 0.0013 and 0.6295, respectively. These findings underscore the success of the approach and its potential to enhance both leukemia diagnosis and treatment.