TY - GEN
T1 - Multilevel segmentation for automatic detection of malignant masses in digital mammograms based on threshold comparison
AU - Rodriguez-Esparza, Erick
AU - Zanella-Calzada, Laura A.
AU - Oliva, DIego
AU - Hinojosa, Salvador
AU - Perez-Cisneros, Marco
PY - 2019/11
Y1 - 2019/11
N2 - Digital image processing implies a fundamental process in the field of medicine, since it supports the improvement of images facilitating their interpretation for the specialists. A technique that has had great relevance is image segmentation, applied to mammography images, in order to improve the diagnosis of breast cancer disease. In this work it is proposed the implementation of a new multilevel segmentation approach based on the minimum cross-entropy threshold - Harris Hawks Optimization (MCET-HHO) metaheuristic algorithm, where five different levels are used in order to compare the behavior of the thresholds applied in mammograms with presence of cancer by allowing the identification of malignant tumors. According to the results, by including four levels in the segmentation of the image, it is possible to identify with a significantly high precision the region of interest (ROI) where the tumor is located, obtaining a similarity of 0.902 with the ROI identified by the specialist. Therefore, it is concluded that the implementation of the MCET-HHO algorithm for multilevel segmentation of mammograms allows to determine the ROI that contains malignant masses, presenting a preliminary support tool for the diagnosis of breast cancer.
AB - Digital image processing implies a fundamental process in the field of medicine, since it supports the improvement of images facilitating their interpretation for the specialists. A technique that has had great relevance is image segmentation, applied to mammography images, in order to improve the diagnosis of breast cancer disease. In this work it is proposed the implementation of a new multilevel segmentation approach based on the minimum cross-entropy threshold - Harris Hawks Optimization (MCET-HHO) metaheuristic algorithm, where five different levels are used in order to compare the behavior of the thresholds applied in mammograms with presence of cancer by allowing the identification of malignant tumors. According to the results, by including four levels in the segmentation of the image, it is possible to identify with a significantly high precision the region of interest (ROI) where the tumor is located, obtaining a similarity of 0.902 with the ROI identified by the specialist. Therefore, it is concluded that the implementation of the MCET-HHO algorithm for multilevel segmentation of mammograms allows to determine the ROI that contains malignant masses, presenting a preliminary support tool for the diagnosis of breast cancer.
KW - Digital Image Processing
KW - Mammograms
KW - MCET-HHO algorithm
KW - Metaheuristic Algorithm
KW - Multilevel Segmentation
UR - http://www.scopus.com/inward/record.url?scp=85083118284&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85083118284&partnerID=8YFLogxK
U2 - 10.1109/LA-CCI47412.2019.9037030
DO - 10.1109/LA-CCI47412.2019.9037030
M3 - Conference contribution
AN - SCOPUS:85083118284
T3 - 2019 IEEE Latin American Conference on Computational Intelligence, LA-CCI 2019
BT - 2019 IEEE Latin American Conference on Computational Intelligence, LA-CCI 2019
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 6th IEEE Latin American Conference on Computational Intelligence, LA-CCI 2019
Y2 - 11 November 2019 through 15 November 2019
ER -