Abstract—Blood cancer, resulting from the uncontrolled growth of white blood cells (leukocytes), is still an important problem in medical diagnosis. In the past, cancerous blood cells were only detected through a manual examination by trained professionals, which took up considerable time and was prone to mistakes. Deep learning algorithms are the new development towards the automation and advancement in the diagnosis of blood cancer. This article talks about how deep learning techniques are making diagnosis of blood cancer more effective and accurate. Deep learning models, such as Convolutional Neural Networks (CNNs), have progressed this further by automatically identifying key patterns in microscopic blood images and thereby making early cancer prediction more precise. The research aims to find the best architectural design which will produce the highest diagnostic accuracy through its assessment of three separate deep learning systems that include Swin Transformer and ConvNeXt and EfficientNetV2. The models achieved high performance through their combination with efficient pre-processing and image enhancement techniques which reduced misclassification rates and improved early cancer detection. The tested models successfully identified the samples according to their actual class distribution which showed slight deviation from the original classification system. The 3 models achieved overall accuracy rates of 99.84% for ConvNeXt, 99.69% for Swin Transformer, and 99.38% for EfficientNetV2, demonstrating that the modernized CNN architecture (ConvNeXt) slightly outperformed the transformer-based model (Swin Transformer) in this specific evaluation. Medical procedures now use deep learning technology to provide quick and accurate patient diagnoses which leads to improved treatment choices and better patient outcomes.
 
Keywords—blood cancer detection, leukemia classification, deep learning, Convolutional Neural Networks (CNNs), swin transformer, feature extraction, medical image analysis
 
Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).