Fully‐automatic identification of gynaecological abnormality using a new adaptive frequency filter and histogram of oriented gradients (HOG). Issue 3 (13th August 2021)

• Record Type: Journal Article
• Authors: Hussein, Ihsan Jasim; Burhanuddin, Mohd Aboobaider; Mohammed, Mazin Abed; Benameur, Narjes; Maashi, Marwah Suliman; Maashi, Mashael S.
• Other Names: Gupta Deepak guestEditor.; Kose Utku guestEditor.; Castillo Oscar guestEditor.; Al‐Turjman Fadi guestEditor.
• Is Part Of: Expert systems. Volume 39:Issue 3(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-08-13
• Subjects: artificial neural network -- breast cancer -- histogram of oriented gradients -- new adaptive frequency filter -- ovarian tumour -- ultrasound images; Expert systems (Computer science); 006.33
• DOI: 10.1111/exsy.12789
• Languages: English
• ISSNs: 0266-4720

Abstract

Abstract: Ultrasound imaging (US) is one of the most common diagnostic imaging tools for producing images of the human body in clinical practice. This work is devoted to studying ultrasound images collected from gynaecological tests for medical purposes regarding ovarian and breast defects. The study revolves around (i) Enhancing the texture of the image by applying a new effective framework that can help in reducing the speckle noise from the image while preserving the most important information; (ii) Extracting the most prominent features using the histogram of oriented gradients (HOG) and; (iii) Fusing the features that are produced by the edge operators and using them as an input to the ANN classifier to generate three trained classifiers. The fusion technique has been used to get an effective decision by using the whole features. The experimental results of the proposed method for the breast cancer and ovarian tumour using the second experiment achieved 97.96% accuracy, 96.05% sensitivity, and 99.17% specificity by utilizing the breast cancer information set. Overall, 95.87% precision, 97.01% sensitivity, and 93.33% specificity have been achieved for the ovarian tumour data collection. Consequently, the proposed method has been improved to validate the output of modern computerized and automated technologies. This method analyzes the gynaecological ultrasound images to identify suspicious objects or cases with health consequences for women.