AUTHOR=Wang Xiuxin , Yang Yuling , Wu Ting , Zhu Hao , Yu Jisheng , Tian Jian , Li Hongzhong 

TITLE=Energy minimization segmentation model based on MRI images

JOURNAL=Frontiers in Neuroscience

VOLUME=Volume 17 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2023.1175451

DOI=10.3389/fnins.2023.1175451

ISSN=1662-453X

ABSTRACT=Medical image segmentation is an important tool for doctors to accurately analyze the volume of brain tissue and lesions, which is important for the correct diagnosis of brain diseases. However, manual image segmentation methods are time-consuming, subjective and lack of reproducibility, so there is a need to develop automatic and reliable methods for image segmentation. Magnetic Resonance Imaging (MRI), a non-invasive and non-invasive imaging technique, is commonly used to detect, characterize and quantify tissues and lesions in the brain. However, the presence of partial volume effect, gray scale in homogeneity, and lesions in MRI images poses a great challenge for automatic medical image segmentation methods. To this end, this paper is dedicated to address the impact of partial volume effect and multiple sclerosis lesions on the segmentation accuracy of magnetic resonance images.Based on the fuzzy clustering space model and energy model, the objective function of the improved model and the post-processing method of lesion filling are studied in depth. First, an energy-minimized segmentation algorithm based on anatomical mapping is proposed. Through experimental verification, the AR-FCM algorithm can better overcome the problem of low segmentation accuracy of the RFCM algorithm for tissue boundary voxels and improve the segmentation accuracy of the RFCM algorithm. Thus, a multi-channel input energy-minimization segmentation method based on lesion filling and anatomical mapping is further proposed, and finally, the feasibility of the lesion filling strategy using post-processing is verified by comparison experiments, and the segmentation accuracy is improved.