Region-Based Segmentation and Classification for Ovarian Cancer Detection Using Convolution Neural Network.

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Ovarian cancer is a leading cause of death in women globally.
• Accurate classification is crucial for effective treatment and improved patient outcomes.
• Existing methods using Artificial Neural Networks (ANNs) show promise but can be improved.

Purpose of the Study:

• To present a novel annotated ovarian image classification method for early and precise diagnosis.
• To enhance classification accuracy compared to manual annotation.
• To demonstrate the superiority of machine learning classification for ovarian cancer detection.

Main Methods:

• Utilizing a novel rapid region-based Convolutional neural network (FaRe-ConvNN) for region of interest (ROI) segmentation and annotation.
• Categorizing input images into epithelial, germ, and stroma cells.
• Employing an ensemble technique with Support Vector Classifier (SVC) and Gaussian Naive Bayes (NB) classifiers for feature classification.

Main Results:

• FaRe-ConvNN achieved over 95% precision.
• SVC demonstrated 95.96% precision, 94.31% recall, and 97.39% specificity.
• Gaussian NB achieved 97.7% precision, 97.7% recall, and 98.69% specificity, with further enhancement by FR-CNN.

Conclusions:

• The proposed FaRe-ConvNN method significantly improves accuracy in ovarian cancer identification.
• Machine learning-based classification offers higher diagnostic accuracy than manual methods.
• This approach aids in precise disease diagnosis, potentially lowering mortality rates.