Chimp Optimization Algorithm Influenced Type-2 Intuitionistic Fuzzy C-Means Clustering-Based Breast Cancer Detection System

• Published in: Cancers Vol. 15; no. 4; p. 1131
• Main Authors: Balaji, Prasanalakshmi, Muniasamy, Vasanthi, Bilfaqih, Syeda Meraj, Muniasamy, Anandhavalli, Tharanidharan, Sridevi, Mani, Devi, Alsid, Linda Elzubir Gasm
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 10.02.2023; MDPI
• Subjects: Algorithms; Artificial intelligence; Breast cancer; Cancer; Cell division; Classification; Datasets; Deep learning; Diagnosis; Disease; Fuzzy algorithms; Fuzzy logic; Fuzzy systems; Image processing; Malignancy; Mammography; Mathematical optimization; Neural networks; Optimization algorithms; Performance evaluation; Saudi Arabia; clustering methods; specificity; cancer detection; magnetic resonance imaging; sensitivity; optimization
• ISSN: 2072-6694; 2072-6694
• DOI: 10.3390/cancers15041131

In recent years, breast cancer detection is an important area of concentration towards curative image dispensation and exploration. Detection of a disease at an early stage is an important factor in taking it to the next level of treatment. Accuracy plays an important role in the detection of disease. COA-T2FCM (Chimp Optimization Algorithm Based Type-2 Intuitionistic Fuzzy C-Means Clustering) is constructed for detection of such malignancy with the highest accuracy in this paper. The proposed detection process is designed with the combination of type-2 intuitionistic fuzzy c-means clustering in addition to oppositional function. In the type-2 intuitionistic fuzzy c-means clustering, the efficient cluster center can be preferred using the chimp optimization algorithm. Initially, the objective function of the type-2 intuitionistic fuzzy c-means clustering is considered. The chimp optimization algorithm is utilized to optimize the cluster center and fuzzifier in the clustering method. The projected technique is implemented, and in addition, performance metrics such as specificity, sensitivity, accuracy, Jaccard Similarity Index (JSI), and Dice Similarity Coefficient (DSC) are assessed. The projected technique is compared with the conventional technique such as fuzzy c means clustering and k mean clustering methods. The resulting method was also compared with existing methods to ensure the accuracy in the proposed method. The proposed algorithm is tested for its effectiveness on the mammogram images of the three different datasets collected from the Mini–Mammographic Image Analysis Society (Mini–MIAS), the Digital Database for Screening Mammography (DDSM), and Inbreast. The accuracy and Jaccard index score are generally used to measure the similarity between the proposed output and the actual cancer affected regions from the image considered. On an average the proposed method achieved an accuracy of 97.29% and JSI of 95%