Paper Contents
Abstract
This study introduces a machine learning-based framework for classifying blood cancer (leukemia), addressing the limitations of manual diagnosis, which is often error-prone and labor-intensive. Among various machine learning methods, deep learning stands out for its ability to handle large-scale, heterogeneous medical datasets effectively. The proposed approach utilizes a feature extraction and training pipeline built on the RESNET-50 architecture, a deep neural network known for its skip connections that mitigate issues like overfitting and vanishing gradients. The systems performance was assessed using error metrics and classification accuracy, achieving an impressive 97.9% accuracy rate a notable improvement over previous benchmarks on the same dataset, which reported 91.84%. To further enhance model reliability, extensive pre-processing techniques were applied to normalize and refine input data. The architectures depth and residual learning capabilities enable it to capture intricate patterns in leukemic cell morphology. Comparative analysis with other deep learning models confirms RESNET-50s superior generalization performance. This advancement holds promise for integrating AI-driven diagnostics into routine clinical workflows, potentially accelerating early detection and treatment planning.
Copyright
Copyright © 2025 harmeet singh lubana. This is an open access article distributed under the Creative Commons Attribution License.
