Identification of Leukemia Subtypes from Microscopic Images Using ResNeXt Algorithm

Abstract: Leukaemia is a severe haematological malignancy characterized by the abnormal proliferation of white blood cells in bone marrow and peripheral blood. Early and accurate identification of leukaemia subtypes is essential for effective diagnosis and treatment planning. This paper proposes a deep learning–based microscopic blood image classification framework using the ResNeXt architecture for automated leukaemia subtype detection. The system formulates leukaemia classification as a multi-class medical image analysis problem, where complex morphological features are extracted from microscopic blood smear images. The proposed framework incorporates preprocessing techniques such as image resizing, normalization, and data augmentation to improve robustness and generalization performance. A ResNeXt backbone with grouped convolutions and residual learning is employed to capture discriminative cellular patterns while maintaining computational efficiency. The model is trained and evaluated on a publicly available leukaemia microscopic image dataset for accurate subtype classification. Experimental results demonstrate improved classification accuracy, feature representation, and prediction reliability compared with conventional convolutional neural network approaches. The proposed system also supports real-time prediction through a Flask-based web interface, enabling accessible and efficient computer-aided diagnosis. The results indicate that the ResNeXt-based framework provides an effective and scalable solution for intelligent leukaemia detection and automated healthcare assistance.

Keywords: Leukaemia Subtype Classification; Microscopic Blood Smear Images; Deep Learning; ResNeXt Architecture; Medical Image Analysis; Automated Disease Detection; Computer-Aided Diagnosis .