Deep Learning-Based Nuclei Segmentation and Melanoma Detection in Skin Histopathological Image Using Test Image...

Area of Science:

• Digital Pathology
• Computational Biology
• Medical Imaging Analysis

Background:

• Histopathological images are vital for skin cancer diagnosis but are very large, making manual analysis time-consuming.
• Automated nuclei segmentation in histopathology is challenging due to variable cell boundaries.
• Artificial Intelligence (AI) offers potential for computer-aided diagnosis (CAD) in skin cancer detection.

Purpose of the Study:

• To develop a deep neural network (DNN)-based technique for automated nuclei segmentation and melanoma detection in histopathological images.
• To expedite the diagnostic process by automating the identification and analysis of abnormal cell nuclei.

Main Methods:

• A DNN-based approach was employed for nuclei segmentation.
• Image augmentation using geometric operations enhanced robustness.
• A morphological technique was applied to the segmented nuclei for melanoma detection.

Main Results:

• The proposed technique achieved a Dice score of 91.61% for nuclei segmentation.
• Melanoma detection accuracy reached a Dice score of 87.9%.
• The method demonstrated effectiveness in analyzing nuclei distribution across tissue sections.

Conclusions:

• The DNN-based technique provides an effective solution for nuclei segmentation and melanoma detection in histopathological images.
• This AI-driven approach significantly improves the efficiency and accuracy of skin cancer diagnosis.
• Automating these processes can accelerate comprehensive diagnostic assessments.