Automated Classification of Lymphoma Subtypes From Histopathological Images Using a U-Net Deep Learning Model:...

Area of Science:

• Medical Imaging
• Computational Pathology
• Artificial Intelligence in Medicine

Background:

• Accurate lymphoma classification is crucial for treatment planning.
• Traditional methods are subjective and inefficient, necessitating automated solutions.
• Deep learning offers a promising approach for objective and efficient diagnostics.

Purpose of the Study:

• To investigate the U-Net deep learning model for classifying and grading lymphoma subtypes.
• To enhance diagnostic precision and efficiency in lymphoma pathology.
• To develop an automated solution for challenging diagnostic tasks.

Main Methods:

• Utilized the U-Net model with attention mechanisms and residual networks for segmentation and classification.
• Processed 620 histopathological images from The Cancer Genome Atlas and Cancer Imaging Archive.
• Employed data augmentation and five-fold cross-validation for model robustness and generalization.

Main Results:

• The U-Net model achieved high segmentation accuracy, improving input quality for classification.
• Achieved 92% accuracy, 91.04% sensitivity, and 89.04% specificity in classifying 3 lymphoma subtypes.
• Demonstrated strong clinical applicability as an assistive diagnostic tool with an AUC of 0.95.

Conclusions:

• Deep learning with U-Net architecture significantly improves lymphoma subtype classification and grading.
• The model provides efficient and precise support for clinical decision-making.
• Future work includes multicenter validation and integration into digital pathology workflows.