Article: Deep learning architectures for detection of acute myeloid leukemia Journal: International Journal of Computational Science and Engineering (IJCSE) 2025 Vol.28 No.3 pp.278 - 291 Abstract: Acute myeloid leukaemia (AML) involves rapid growth of immature blood cells, impairing normal immune functions. AML represents nearly 80% of adult blood cancers. Despite various treatments and therapies, prognosis remains poor and there are critical but unmet needs. Artificial intelligence (AI) offers potential diagnostic options yet existing models often rely on traditional algorithms and small datasets, limiting effectiveness in AML detection. We propose two deep learning architectures using large blood cell image datasets to detect AML. These models use diverse kernel shapes to identify complex patterns, requiring up to 97% fewer parameters while achieving high accuracy (99.7% and 99.6%). Compared to other networks like AlexNet, MobileNet, ResNet50, and InceptionV3, which show accuracy of 99.5%, 96.4%, 54.6%, and 95.9%, our models are better or competitive. Improved generalisation is confirmed by learning curves and feature maps. These models can help diagnosis of AML more accurately and efficiently. Inderscience Publishers - linking academia, business and industry through research

Title: Deep learning architectures for detection of acute myeloid leukemia

Authors: Niteesh Kempusagara Ramesh; Vijayakumar Kadappa; Rajeshwari Devi Doddapoojari Veerabhaskar; Divijendranatha Reddy Sirigiri; Pooja Tekal Sreedhar

Addresses: Department of Computer Applications, PES University, Bengaluru, 560085, India ' Department of Computer Applications, B.M.S. College of Engineering, Bengaluru, 560019, India ' Department of Electronics and Communication Engineering, Global Academy of Technology, Bengaluru, 560098, India ' Department of Biotechnology, B.M.S. College of Engineering, Bengaluru, 560019, India ' Department of Computer Applications, PES University, Bengaluru, 560085, India

Abstract: Acute myeloid leukaemia (AML) involves rapid growth of immature blood cells, impairing normal immune functions. AML represents nearly 80% of adult blood cancers. Despite various treatments and therapies, prognosis remains poor and there are critical but unmet needs. Artificial intelligence (AI) offers potential diagnostic options yet existing models often rely on traditional algorithms and small datasets, limiting effectiveness in AML detection. We propose two deep learning architectures using large blood cell image datasets to detect AML. These models use diverse kernel shapes to identify complex patterns, requiring up to 97% fewer parameters while achieving high accuracy (99.7% and 99.6%). Compared to other networks like AlexNet, MobileNet, ResNet50, and InceptionV3, which show accuracy of 99.5%, 96.4%, 54.6%, and 95.9%, our models are better or competitive. Improved generalisation is confirmed by learning curves and feature maps. These models can help diagnosis of AML more accurately and efficiently.

Keywords: artificial intelligence; deep learning; artificial neural networks; acute myeloid leukemia; healthcare.

DOI: 10.1504/IJCSE.2025.146070

International Journal of Computational Science and Engineering, 2025 Vol.28 No.3, pp.278 - 291

Received: 29 Oct 2023
Accepted: 23 May 2024
Published online: 06 May 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: Deep learning architectures for detection of acute myeloid leukemia

Keep up-to-date