EUVIMEDEUVIMED entdecken
Crossref

Deep Computational Intelligence Frameworks for Automated Leukemia Cell Detection and Classification: A Systematic Survey of Advanced Deep Learning Architectures

P. Prasad, C. Madhusudhana Rao

International Journal of Drug Delivery Technology · 08.06.2026 · 16 (48s)

DOI: 10.25258/ijddt.16.48s.73 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Drug Delivery Technology, online erschienen am 08.06.2026. Verfasst von P. Prasad und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Hematologic malignancies, especially in leukemia, is remaining as a major global health concern with its progression and diagnostic complexity. The microscopic examination of the peripheral blood smears is continuing in serving as a standard diagnostic procedure; however, it is depending heavily on expert interpretation, and it often tends to introduce the variability. In recent years, the computational intelligence, which is driven by deep learning (DL), is showing a stronger potential in automating the leukemia detection with an improved consistency. Despite significant progress, existing approaches is suffering from the limitations such as the dataset imbalance across the imaging conditions, and insufficient combination of the multimodal clinical information. Further, the absence of an unified synthesis of the DL architectures as it is restricting the identification of the optimal design strategies for a stronger leukemia classification. Conventional manual diagnosis is requiring an extensive clinical expertise and it is often experiencing a reduced reliability in large-scale screening environments. Recent developments in computational intelligence and deep learning significantly improve automated leukemia detection; however, several unresolved issues persist, including class imbalance, poor cross-dataset generalization, limited interpretability, computational overhead, and constrained deployment in real-time clinical settings. This survey systematically is reviewing 45 peer-reviewed studies that is published in leading journals. The paper is categorizing the existing works into the preprocessing techniques, segmentation methods, feature extraction strategies, and classification architectures, which is including Convolutional Neural Networks, transfer learning models, Vision Transformers, and hybrid frameworks. It further is examining the evaluation protocols, and performance metrics in establishing an analytical perspective. A taxonomy-driven framework is developed in organizing the methodological trends. The analysis is indicating that DL -based approaches consistently is performing better than the conventional machine learning (ML) techniques, which is achieving the classification accuracies in the range of 92.3% to 99.7%. CNN-based models are reporting the average accuracies above 95.4%, while the hybrid CNN-Transformer architectures are showing improvements of the approximately 2.8% in F1-score over standalone CNNs. Segmentation-based pipelines are improving the classification sensitivity by nearly 3.2%, especially in Acute Lymphoblastic Leukemia detection tasks. Transfer learning models such as ResNet and EfficientNet are reducing the training time by upto 40% while it is maintaining a higher diagnostic precision.

Zitieren

APA

Prasad, P., & Rao, C. M. (2026). Deep Computational Intelligence Frameworks for Automated Leukemia Cell Detection and Classification: A Systematic Survey of Advanced Deep Learning Architectures. International Journal of Drug Delivery Technology, 16(48s). https://doi.org/10.25258/ijddt.16.48s.73

Vancouver

Prasad P, Rao CM. Deep Computational Intelligence Frameworks for Automated Leukemia Cell Detection and Classification: A Systematic Survey of Advanced Deep Learning Architectures. International Journal of Drug Delivery Technology. 2026;16(48s). doi:10.25258/ijddt.16.48s.73

BibTeX
@article{Prasad2026,
  author = {Prasad, P. and Rao, C. Madhusudhana},
  title = {Deep Computational Intelligence Frameworks for Automated Leukemia Cell Detection and Classification: A Systematic Survey of Advanced Deep Learning Architectures},
  journal = {International Journal of Drug Delivery Technology},
  year = {2026},
  volume = {16},
  number = {48s},
  doi = {10.25258/ijddt.16.48s.73}
}
Kostenlos starten →