Convolutional Deep Learning for Automated Blood Cancer Detection: A Novel Approach to Enhance Diagnostic Accuracy

Abstract

Acute Lymphoblastic Leukemia (ALL) and Acute Myeloid Leukemia (AML) are aggressive hematologic malignancies that necessitate early and accurate detection for improved patient outcomes. Traditional diagnostic methods, relying on manual microscopic examination of blood smears, are time-consuming, labor-intensive, and prone to inter-observer variability. This paper introduces a novel convolutional deep learning (CNN) framework designed for automated blood cancer detection and classification from peripheral blood smear (PBS) images. Our approach leverages advanced CNN architectures, including customized models like ALLNET and DeepLeuk, and state-of-the-art models such as ConvNeXt, to achieve high diagnostic accuracy. We detail the preprocessing techniques, model training strategies, and rigorous evaluation metrics employed. The proposed system demonstrates superior performance compared to traditional methods and existing AI models, achieving accuracies up to 99.5% for leukemia staging and 98.00% for leukocyte subtyping. Furthermore, we incorporate explainability techniques like Grad-CAM to provide biological interpretability, bridging the gap between AI-driven diagnostics and clinical decision-making. This research underscores the potential of deep learning to revolutionize hematopathology, offering a scalable, efficient, and reliable solution for early blood cancer detection and classification, ultimately aiming to minimize diagnostic errors and improve patient survival rates.

Introduction

Blood cancers such as Acute Lymphoblastic Leukemia (ALL) and Acute Myeloid Leukemia (AML) are serious diseases that require early and accurate diagnosis. Traditional diagnosis relies on manual examination of peripheral blood smear (PBS) images by hematologists, a process that is time-consuming, subjective, and dependent on specialized expertise. Recent advances in deep learning, particularly Convolutional Neural Networks (CNNs), offer a promising solution by automating blood cancer detection and improving diagnostic accuracy.

The literature review highlights the evolution of artificial intelligence in hematopathology. Early machine learning methods relied on manual feature extraction and achieved limited success. The introduction of deep learning and CNNs significantly improved performance, with specialized models such as ALLNET and DeepLeuk achieving accuracies above 95%. More advanced architectures, including ConvNeXt and Hybrid Involutional-Convolutional Neural Networks (HICNN), have further enhanced classification accuracy and subtype detection. Explainable AI techniques such as Grad-CAM have also improved model transparency and clinical trust.

The proposed methodology uses publicly available blood smear image datasets. Images undergo preprocessing steps such as scaling, region-of-interest extraction, noise reduction, and data augmentation. Multiple CNN architectures are evaluated, including customized models, pre-trained networks, advanced models like ConvNeXt, and hybrid architectures such as HICNN. Training is performed using GPU-enabled platforms with optimization techniques like Stochastic Weight Averaging (SWA). Grad-CAM is integrated to generate interpretable visualizations of model decisions.

The models are evaluated on binary and multi-class classification tasks, including leukemia detection, subtype classification, staging, and leukocyte subtyping. Performance is measured using accuracy, precision, recall, F1-score, specificity, and Brier score, and compared against traditional machine learning methods and other deep learning models.

Results demonstrate exceptional performance:

• ALLNET achieved accuracies of up to 95.54%.
• DeepLeuk reached a precision of 98.9%.
• ConvNeXt achieved 95% accuracy for AML subtype detection, outperforming ResNet50 and Vision Transformers.
• HICNN achieved up to 99.5% accuracy in leukemia staging and 98% accuracy in leukocyte subtyping, with excellent reliability and calibration.
• Grad-CAM visualizations confirmed that models focus on biologically relevant cellular features.

The study concludes that deep learning-based blood cancer detection systems can significantly improve diagnostic precision, reduce clinician workload, and provide more accessible diagnostic services, especially in underserved regions. Future research should focus on real-time clinical integration, validation on larger and more diverse datasets, federated learning for privacy-preserving training, and expanding detection capabilities to a broader range of blood cancers and genetic mutations.

Conclusion

This research successfully demonstrates the power of convolutional deep learning in revolutionizing blood cancer detection and classification.By leveraging advanced architectures like ConvNeXt and hybrid models, and incorporating explainability techniques, we have achieved unprecedented diagnostic accuracy and reliability.The proposed framework offers a robust, scalable, and interpretable solution that has the potential to significantly improve patient outcomes by enabling earlier and more precise diagnoses.This work represents a critical step towards integrating AI into routine clinical practice, paving the way for a new era of precision hematopathology.

References

[1] Agaian, S., Madhukar, M., & Chronopoulos, A. T. (2014). Automated screening system for acute myelogenous leukaemia detection in blood microscopic images. IEEE Systems Journal, 8(3), 995-1004. [2] Alshehri, O. M., Shaf, A., Shakeel, U., Ali, T., Irfan, M., Jalal, M. M., Altayar, M. A., Abu-Alghayth, M. H., & Al Shmrany, H. (2025). Dynamic kernel generation through hybrid involution and convolution neural networks for leukemia and white blood cell classification. Nature Scientific Reports. (Preprint/Unverified Publication Date) [3] Dhurve, H., Yenurkar, G. K., Mal, S., Thakur, N., Dhomne, S., Patel, M., & Kulmeti, K. (2024). DeepLeuk: a convolutional neural network pre-trained model for microscopic cell images-Based leukemia Cancer analysis. Multimedia Tools and Applications, 84(8), 13809–13842. [4] Fujita, T. C., Sousa-Pereira, N., Amarante, M. K., & Watanabe, M. A. E. (2021). Acute lymphoid leukaemia etiopathogenesis. Molecular Biology Reports, 48(1), 817–822. [5] Labati, R. D., Piuri, V., & Scotti, F. (2011). All-IDB: the acute lymphoblastic leukaemia image database for image processing. In Proceedings of the 18th IEEE International Conference on Image Processing (pp. 2045-2048). [6] Mohamed, H., Omar, R., & Saeed, N. (2018). Automated detection of white blood cells cancer diseases. In Proceedings of the 1st International Workshop on Deep and Representation Learning (pp. 48-54). [7] Mohapatra, S., Patra, D., & Satpathy, S. (2014). An ensemble classifier system for early diagnosis of acute lymphoblastic leukaemia in blood microscopic images. Neural Computing and Applications, 24(7-8), 1887-1900. [8] Mustapha, M. T., & Ozsahin, D. U. (2025). Morphological Analysis and Subtype Detection of Acute Myeloid Leukemia in High-Resolution Blood Smears Using ConvNeXT. Artificial Intelligence, 6(3), 45. [9] Sampathila, N., Chadaga, K., Goswami, N., Chadaga, R. P., Pandya, M., Prabhu, S., Bairy, M. G., Katta, S. S., Bhat, D., & Upadya, S. P. (2022). Customized Deep Learning Classifier for Detection of Acute Lymphoblastic Leukemia Using Blood Smear Images. Healthcare, 10(10), 1812. [10] Shafique, S., & Tehsin, S. (2018). Acute lymphoblastic leukaemia detection and classification of its subtypes using pretrained deep convolutional neural networks. Technology in Cancer Research & Treatment, 17. [11] Singhal, V., & Singh, P. (2014). Local binary pattern for automatic detection of acute lymphoblastic leukaemia. In Proceedings of the 20th National Conference on Communications (pp. 1-5).

Copyright

Copyright © 2026 Dr. Saranya Vijayan. 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.