Artificial Intelligence Algorithm-Based Positron Emission Tomography (PET) and Magnetic Resonance Imaging (MRI) in the Treatment of Glioma Biopsy

Abstract

This study was aimed at exploring the application value of positron emission tomography (PET) + magnetic resonance imaging (MRI) technology based on convolutional neural network (CNN) in the biopsy and treatment of intracranial glioma. 35 patients with preoperatively suspicious gliomas were selected as the research objects. Their imaging images were processed using CNN. They were performed with the preoperative head MRI, fluorodeoxyglucose (FDG) PET, and ethylcholine (FECH) PET scans to construct the cancer tissue contours. In addition, the performance of CNN was evaluated, and the postoperative pathology of patients was analyzed. The results suggested that the CNN-based PET + MRI technology showed a recognition accuracy of 97% for images. Semiquantitative analysis was adopted to analyze the standard uptake value (SUV). It was found that the SUVFDG and SUVFECH of grade II/III glioma were 9.77 ± 4.87 and 1.82 ± 0.50, respectively, and the SUVFDG and SUVFECH of grade IV glioma were 13.91 ± 1.83 and 3.65 ± 0.34, respectively. According to FDG PET, the mean value of SUV on the lesion side of grade IV glioma was greater than that of grade II-III glioma, and the difference was significant P<0.05, and similar results were obtained on FECH PET. It showed that CNN-based PET + MRI fusion technology can effectively improve the recognition effect of glioma, can more accurately determine the scope of glioma lesions, and can predict the degree of malignant glioma to a certain extent.