Adoption of Magnetic Resonance Image Features under Segmentation Algorithm in Effect Evaluation of Ginkgo Diterpenoid Lactone Glucamine Injection in Treatment of Cerebral Infarction

Abstract

Magnetic resonance imaging (MRI) image segmentation based on a segmentation algorithm was performed to assess neurological function in patients with acute cerebral infarction, to investigate the efficacy evaluation of Ginkgo diterpene lactones meglumine injection (GDLI) in the treatment of cerebral infarction and the efficiency of MRI image segmentation algorithm. First, the results of the fast semisupervised segmentation algorithm (algorithm group) and traditional processing (control group) were compared and analyzed. The recall rate, accuracy, recognition accuracy, and segmentation time of the two groups were compared. The control group was given conventional treatment, while the algorithm group was given GDLI based on conventional treatment. Finally, the difference in serum vascular endothelial growth factor (VEGF), hypoxia-inducible factor-la (HIF-la), angiotensin (Ang)-1, Ang-2, and interleukin (IL)-6 protein concentration was analyzed after treatment. The algorithm evaluation results showed that the accuracy and recall rate of MRI images recognized by the algorithm group fluctuate at 90%. In the control group, the accuracy and recall rate of MRI image results fluctuated at 80%, and the data were statistically different ( $p<0.05$ ). The clinical index test results showed that the serum VEGF content of the test group was higher than that of the control group, and the data was statistically different ( $p<0.05$ ). In addition, the cerebral blood flow (CBF) and cerebral blood volume (CBV) of the lesion side of the algorithm group were greatly higher than those of the control group on the 30th day, and the differences were significant ( $p<0.05$ ). There was little difference between the method presented in this study and the manual delineation by a physician. Compared with traditional manual segmentation, this method greatly reduced the time required for the segmentation of lesions. The diagnostic specificity, sensitivity, and accuracy of the images segmented by the fast semisupervised algorithm were higher than those of the conventional method, and the diagnostic accuracy of acute cerebral infarction was high. In addition, it was sensitive and accurate to detect acute cerebral infarction, which provided a reliable reference for early diagnosis and condition judgment of patients.