Performance Comparison of Ovarian Tumour Classification Using Deep Neural Network

Ovarian tumour originates in ovaries posing a serious threat to women. As a result, it leads to abnormal cells which have the ability to spread to the other regions of the body. Ovarian tumour is a sort of risky improvement that impacts ovaries in females, and is hard to perceive at early the phase because of which it stays as one of the guideline wellsprings of illness end. The ultrasound imaging is the low cost and best to detect tumour in ovaries. To improve tumour detection accuracy in ultrasound ovarian images, enhanced preprocessing, image segmentation, feature extraction and feature selection, and classification method are proposed. The enhanced preprocessing, tumour segmentation, feature extraction and feature selection, and tumour classification methods have been analyzed to segment and classify the tumour effectively in ultrasound ovarian tumour images. The Anisotropic Diffusion Filter (ADF) has been applied for the preprocessing of the ultrasound ovarian image to improve the image quality for further processing such as segmentation, feature extraction and feature selection, and classification. The Improved Whale Search Optimization (IWSO) segmentation method has been applied to segment tumour in the ultrasound image efficiently. The Deep Neural Network (DNN) method has given more accuracy in classifying the ultrasound ovarian tumour image into benign or malignant. The DNN has high sensitivity, high specificity, high accuracy, high PPV, high NPV, low FPR, low FNR, low FDR, low time complexity, low error rate, and low false classification ratio during the tumour classification compared with the existing methods.