Magnetic resonance imaging features to evaluate the neonatal hypoglycemia brain injury and investigation of related risk factors under the fuzzy C‐means clustering intelligent algorithm

• Published in: Expert systems Vol. 41; no. 7
• Main Authors: Jin, Dongmei, Zhang, Zhongxu, Ma, Yanru, Dai, Zhushan, Zhao, Lili, Ma, Tongyao, Chen, Guoping
• Format: Journal Article
• Language: English
• Published: Oxford Blackwell Publishing Ltd 01.07.2024
• Subjects: Algorithms; Clustering; Factor analysis; FCM intelligent algorithm; Glucose; Head injuries; high risk factors; Hypoglycemia; hypoglycemia brain injury; Image quality; Image segmentation; Magnetic resonance imaging; Medical imaging; Risk factors; Traumatic brain injury
• ISSN: 0266-4720; 1468-0394
• DOI: 10.1111/exsy.13491

This research was aimed to analyse the application value of magnetic resonance imaging based on Fuzzy C‐means (FCM) algorithm in neonatal hypoglycemia brain injury (HBI), and explore the risk factors related to the occurrence of brain injury in children, to provide guidance for clinical diagnosis and treatment. 114 children with hypoglycemia were divided into brain injury group (58 cases) and no brain injury group (56 cases) according to whether they had brain injury or not. The MRI image signal performance, general data, average minimum blood glucose value, duration of hypoglycemia, first feeding time, age of onset of hypoglycemia, and algorithm segmentation performance of the two groups of patients were observed and compared. Furthermore, the Logistic factor analysis was carried out to summarize the MRI characteristics and related risk factors of neonatal hypoglycemic brain injury. The results showed that the average minimum blood glucose (1.09 ± 0.53 mmoL/L) in the brain injury group was lower than that in the non‐brain injury group (1.75 ± 0.49 mmoL/L), and the duration of hypoglycemia (43.1 ± 21.07 h) was higher than that in the non‐brain injury group (13.79 ± 6.81 h), p < 0.05. The first feeding time and age of hypoglycemia in the brain injury group were higher than those in the non‐brain injury group, showing a difference with p < 0.05. In the brain injury group, all 58 cases showed high DWI (diffusion weighted imaging) signal at the damaged site at the early stage of MRI (magnetic resonance imaging), and 23 cases (39.66%) were involved in parieto‐occipital lobe. Image segmentation coefficient of Vpc increased significantly under FCM clustering algorithm (p < 0.05). Late first feeding time, low blood sugar level, and long duration were high risk factors for hypoglycemic brain injury. In conclusion, MRI images based on FCM clustering algorithm had higher image quality. Late first feeding time, low blood sugar level, and long duration of hypoglycemia were high risk factors for hypoglycemic brain injury.