An Efficient Segmentation and Classification System in Medical Images Using Intuitionist Possibilistic Fuzzy C-Mean...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Biomedical Engineering

Background:

• Viruses like herpesvirus, polyomavirus, papillomavirus, and retrovirus are linked to breast cancer.
• Accurate segmentation and classification of medical images are crucial for early breast cancer detection.
• Existing fuzzy c-mean algorithms face challenges with noisy or coincident clusters in medical image analysis.

Purpose of the Study:

• To propose an efficient segmentation and classification system for breast cancer detection in Mammography Image Analysis Society (MIAS) images.
• To develop a novel clustering methodology, intuitionistic possibilistic fuzzy c-mean (IPFCM), by hybridizing Intuitionist Fuzzy C-Mean (IFCM) and Possibilistic Fuzzy C-Mean (PFCM).
• To evaluate the efficacy of the proposed IPFCM method against other fuzzy clustering and classification algorithms.

Main Methods:

• Developed an intuitionistic possibilistic fuzzy c-mean (IPFCM) clustering technique by combining IFCM and PFCM algorithms.
• Applied the IPFCM method for segmenting and classifying abnormal regions in mammogram images from the MIAS dataset.
• Compared the performance of IPFCM with Support Vector Machine (SVM), Decision Tree (DT), Rough Set Data Analysis (RSDA), and Fuzzy-SVM classification algorithms.

Main Results:

• The IPFCM method achieved high average segmentation accuracy on MIAS images with varying noise levels (91.25% at 5% noise, 87.50% at 7% noise, 85.30% at 9% noise).
• In Fuzzy-SVM classification, IPFCM yielded the highest average accuracy rate of 98.85%, outperforming Otsu, Fuzzy c-mean, IFCM, and PFCM.
• The proposed approach demonstrated significant effectiveness in both clustering and classification of breast cancer images.

Conclusions:

• The developed IPFCM clustering technique is highly effective for segmenting and classifying breast cancer in mammograms.
• IPFCM offers improved robustness against noise and outliers compared to traditional fuzzy clustering methods.
• This advanced image analysis system shows great promise for enhancing breast cancer detection and diagnosis.