Applying cuckoo search based algorithm and hybrid based neural classifier for breast cancer detection using ultrasound images

• Published in: Evolutionary intelligence Vol. 15; no. 2; pp. 989 - 1006
• Main Authors: Michahial, Stafford, Thomas, Bindu A.
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.06.2022; Springer Nature B.V
• Subjects: Algorithms; Applications of Mathematics; Artificial Intelligence; Bioinformatics; Breast cancer; Classifiers; Control; Engineering; Feature extraction; Image segmentation; Mathematical and Computational Engineering; Mechatronics; Medical imaging; Performance evaluation; Robotics; Search algorithms; Sensitivity; Special Issue; Statistical Physics and Dynamical Systems; Tumors; Ultrasonic imaging; ROI extraction; Clustering; Ultrasound imaging; Multiple kernel-based fuzzy C-means (MKFCM); Classification
• ISSN: 1864-5909; 1864-5917
• DOI: 10.1007/s12065-019-00268-9

Ultrasound examination is one of the most convenient and appropriate processes used for the diagnosis of tumors that make use of ultrasound images. Ultrasound imaging is a noninvasive modality utilized commonly for the detection of breast cancer, which is a common and dangerous cancer found in women. This paper proposes an approach for the detection of breast cancer using ultrasound images using MKF-cuckoo search (MKF-CS) algorithm and hybrid based neural (H-BN) classifier. In pre-processing, the input images to be diagnosed are pre-processed by ROI extraction using a novel algorithm, four way search. The pre-processed image is allowed to perform segmentation using MKF-CS algorithm. The key features, such as mean, variance, standard deviation, and so on, are extracted in feature extraction and are fed to the proposed H-BN classifier. Based on the training data, H-BN classifier classifies the data into benign or malignant tumor classes, for the detection of breast cancer. To evaluate the performance of the proposed MKFCS-HBN approach, three metrics, such as accuracy, sensitivity, and specificity, are utilized. The experimental results show that MKFCS-HBN could attain the maximum performance with an accuracy of 0.8889, the sensitivity of 1, and specificity of 0.85 and thus, prove its effectiveness.