DeepLeuko-(Deep learning applied to leukemia detection)
Ms. Jeevitha M, Abinaya K, Sharmila R, Arthi S, B Karthika · International Journal of Research Publication and Reviews · 2025
This study presents an automated approach for detecting Acute Lymphoblastic Leukemia (ALL) using deep learning.ALL is a severe blood disorder caused by the overproduction of immature lymphoblasts, making early diagnosis crucial for effective treatment.Traditional diagnostic methods like fluorescence in situ hybridization (FISH) and immunophenotyping are expensive and time-consuming, highlighting the need for a faster and more cost-effective alternative.To address this challenge, we propose a CNN-based classification model for ALL detection using microscopic blood smear images.A dataset of 108 stained blood smear images was processed using a fuzzy-based two-stage color segmentation technique to isolate leukocytes from other blood components.Feature extraction involved texture analysis, shape descriptors, and novel geometric attributes such as Hausdorff dimension and contour signature.These extracted features were used to train a CNN classifier, which achieved high accuracy in differentiating leukemia-positive and leukemia-negative samples.The proposed method significantly improves diagnostic efficiency compared to traditional techniques by reducing human error and accelerating the detection process.Experimental results demonstrate that CNN-based classification offers a reliable, cost-effective, and time-efficient solution for early leukemia screening.By automating the detection process, this approach can assist medical professionals in making faster and more accurate diagnoses, ultimately leading to better patient outcomes.This research highlights the potential of deep learning in medical imaging and hematology, emphasizing the effectiveness of CNNs in identifying leukemia from blood smear images.The integration of image processing techniques and deep learning models not only enhances diagnostic accuracy but also provides an accessible solution for leukemia screening, particularly in resource-limited settings.The findings suggest that AI-driven approaches can revolutionize hematological diagnostics, making leukemia detection more efficient, affordable, and widely available.