IoMT-Based Automated Detection and Classification of Leukemia Using Deep Learning.

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Oncology

Background:

• Computer-aided diagnosis (CAD) and machine learning algorithms are increasingly used for disease identification, including leukemia.
• Leukemia, a white blood cell disorder, has four subtypes requiring accurate early diagnosis for effective treatment.
• Existing diagnostic methods for leukemia subtypes need improvement in effectiveness, learning, and performance.

Purpose of the Study:

• To propose an Internet of Medical Things (IoMT)-based framework for rapid, safe, and accurate leukemia subtype identification.
• To leverage cloud computing and network-connected clinical gadgets for real-time diagnostic coordination.
• To address challenges in leukemia diagnosis, particularly for critical patients during pandemics.

Main Methods:

• Implementation of an IoMT framework integrating clinical devices via cloud computing.
• Utilizing Dense Convolutional Neural Network (DenseNet-121) and Residual Convolutional Neural Network (ResNet-34) for leukemia subtype identification.
• Validation using two public leukemia datasets: ALL-IDB and ASH image bank.

Main Results:

• The proposed IoMT framework facilitated quick and safe identification of leukemia subtypes.
• DenseNet-121 and ResNet-34 models demonstrated superior performance compared to other machine learning algorithms.
• The system enables real-time coordination among patients and healthcare professionals for testing, diagnosis, and treatment.

Conclusions:

• The developed IoMT framework significantly enhances the accuracy and efficiency of leukemia diagnosis.
• The proposed deep learning models (DenseNet-121, ResNet-34) show promise for clinical application in leukemia subtype identification.
• This framework offers a valuable tool for improving patient care and managing critical conditions, even during health crises.