Utilizing Deep Feature Fusion for Automatic Leukemia Classification: An Internet of Medical Things-Enabled Deep...

Area of Science:

• Medical Diagnostics
• Artificial Intelligence in Healthcare
• Computational Biology

Background:

• Acute lymphoblastic leukemia (ALL) diagnosis is challenging, requiring time-consuming and expensive specialist tests.
• Early ALL detection is crucial for timely and effective treatment initiation.
• Advancements in Artificial Intelligence (AI) and Internet of Things (IoT) offer new diagnostic possibilities.

Purpose of the Study:

• To introduce a novel AI-based Internet of Medical Things (IoMT) framework for automated leukemia detection from peripheral blood smear (PBS) images.
• To develop and evaluate a deep learning-based fusion model for accurate ALL classification.
• To enhance the speed and efficiency of leukemia diagnosis.

Main Methods:

• A fusion deep learning model was developed, utilizing two input channels: original and segmented PBS images.
• VGG16 and DenseNet-121 were employed for feature extraction from original and segmented images, respectively.
• The model was trained on 6512 images from 89 individuals and evaluated for classification performance.

Main Results:

• The proposed fusion model achieved high diagnostic accuracy (99.89%), precision (99.80%), and recall (99.72%).
• The model demonstrated superior performance compared to several state-of-the-art Convolutional Neural Network (CNN) models.
• A web application (Beta Version) was developed for simulating the leukemia detection methodology.

Conclusions:

• The developed AI-based IoMT framework and fusion model show significant potential for accurate and efficient leukemia detection.
• This approach can potentially save lives and reduce diagnostic efforts.
• The findings have implications for advancing computer-aided leukemia detection in biomedical research.