RETRACTED: Skin cancer segmentation and classification by implementing a hybrid FrCN-(U-NeT) technique with machine...

Area of Science:

• Dermatology and Medical Imaging
• Artificial Intelligence in Healthcare
• Computational Pathology

Background:

• Skin cancer presents a significant global health challenge with rising mortality rates.
• Early detection and accurate diagnosis are crucial for effective treatment and improved patient prognosis.
• Current diagnostic methods can be limited, necessitating advancements in detection technology.

Purpose of the Study:

• To develop and evaluate an improved hybrid image segmentation method for enhanced skin cancer detection.
• To compare the proposed method's performance against existing advanced techniques for classifying skin lesions.
• To provide a novel tool for early disease diagnosis, potentially guiding clinical practices and public health strategies.

Main Methods:

• Utilized the ISBI-2017 dataset for training and validation.
• Implemented a hybrid image segmentation approach combining Fully Convolutional Networks (FrCN) with U-Net architecture.
• Employed the Region-based Convolutional Neural Network (R-CNN) algorithm for the classification phase.

Main Results:

• The proposed FrCN-(U-Net) hybrid model demonstrated superior performance in both training and testing accuracy compared to existing methods.
• The hybrid segmentation and classification approach effectively enhanced early disease diagnosis capabilities.
• Achieved significant improvements in detecting skin cancer types, including benign lesions and melanoma.

Conclusions:

• The developed hybrid image segmentation technique offers a promising alternative for skin cancer detection.
• Enhanced early diagnosis through this method can lead to improved treatment outcomes and patient survival rates.
• This approach has the potential to serve as a valuable guide for clinical decision-making and public health initiatives aimed at reducing skin cancer deaths.