Deep Learning Applications in Leukemia Detection

Deep learning algorithms have become integral to the automated analysis of microscopic blood images for the early detection and classification of leukaemia. Convolutional neural networks enable the extraction of hierarchical features from cell images, while transfer learning leverages pre-trained models to improve performance on limited datasets. Optimisation techniques such as bio-inspired swarm algorithms refine feature selection to reduce computational cost without sacrificing accuracy. Explainable AI approaches and attention mechanisms reveal the diagnostic criteria behind model decisions, fostering clinician trust. Moreover, the integration of deep learning with Internet of Medical Things platforms supports real-time diagnostic workflows and remote monitoring. Collectively, these innovations have enhanced diagnostic speed, reproducibility and global access to accurate leukaemia screening.

Research from Nature Portfolio

Recent studies have combined deep feature extraction with advanced optimisation to achieve high accuracy in white blood cell classification. One approach employs a pre-trained CNN architecture to derive thousands of features from peripheral blood smear images and then applies a statistically enhanced salp swarm algorithm to identify the most relevant 1 000 features. This streamlined pipeline demonstrates competitive accuracy on public datasets while substantially reducing computational requirements. Another line of research introduces an intelligent decision support system for acute lymphoblastic leukaemia diagnosis. It uses a novel clustering algorithm to segment nucleus and cytoplasm regions, extracts shape, texture and colour features, and applies ensemble classifiers to distinguish lymphoblasts from healthy cells with over 96 per cent accuracy. These methods exemplify the translation of deep learning into robust clinical tools.

Research from all publishers

In a recent advance, an explainable multiple-instance learning framework was developed to classify genetic subtypes of acute myeloid leukaemia directly from single-cell images. The model highlights diagnostically relevant cells with high concordance to expert annotations and achieves near-perfect discrimination between leukaemia subtypes and healthy controls. A comprehensive review of acute lymphoblastic leukaemia detection methods categorises image-processing, conventional machine learning and deep learning techniques, emphasising the rise of transfer learning and convolutional network architectures in improving sensitivity and specificity. Another study demonstrates an Internet of Medical Things-based system that links clinical imaging devices with cloud computing resources, employing DenseNet-121 and ResNet-34 backbones to classify leukaemia subtypes in real time, thereby enabling rapid remote diagnosis and facilitating continuity of care.

Deep Learning Applications in Leukemia Detection publication trend

The graph below shows the total number of articles in deep learning applications in leukemia detection across all publications each year (not limited to Nature Index journals).

Technical terms

Convolutional neural network (CNN): A class of deep learning models that applies convolutional filters to extract spatial hierarchies of features from images.

Transfer learning: A technique that adapts a model pre-trained on a large dataset to a new, related task, reducing training time and data requirements.

Explainable AI: Methods that provide insight into the decision-making process of complex models, enhancing transparency and trust.

Multiple instance learning: A framework where labels are assigned to sets of instances (e.g. cell images), allowing weakly supervised classification.

Swarm optimisation: Bio-inspired algorithms that simulate collective behaviour to select optimal features or parameter values.

Internet of Medical Things (IoMT): A network of interconnected medical devices and cloud-based systems that supports continuous data exchange for remote diagnosis.

References

1. Explainable AI identifies diagnostic cells of genetic AML subtypes. PLOS Digital Health (2023).
2. Efficient Classification of White Blood Cell Leukemia with Improved Swarm Optimization of Deep Features. Scientific Reports (2020).
3. An Intelligent Decision Support System for Leukaemia Diagnosis using Microscopic Blood Images. Scientific Reports (2015).
4. Deep Transfer Learning in Diagnosing Leukemia in Blood Cells. Computers (2020).
5. IoMT‐Based Automated Detection and Classification of Leukemia Using Deep Learning. Journal of Healthcare Engineering (2020).
6. A Systematic Review on Recent Advancements in Deep and Machine Learning Based Detection and Classification of Acute Lymphoblastic Leukemia. IEEE Access (2022).