Deep Convolutional Neural Network-Based Brain Magnetic Resonance Imaging Applied in Glioma Diagnosis and Tumor Region Identification

The aim of this study was to explore the application value of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) based on a convolutional neural network (CNN) algorithm in glioma diagnosis and tumor segmentation. 66 patients with gliomas who were diagnosed and treated in the hospital were selected as the research objects. The patients were rolled into the high-grade glioma group (HGG, 46 cases) and the low-grade glioma group (LGG, 20 cases) according to the World Health Organization glioma grading standard. All patients received a conventional plain scan and a DCE-MRI. Parameters such as volume transfer constant (K(trans)), rate constant (K(ep)), extracellular volume (V(e)), and mean plasma volume (V(p)) were calculated, and the parameters of patients of each grade were analyzed. The efficacy of each parameter in diagnosing glioma was analyzed through a receiver operating characteristic curve. All images were segmented by the CNN algorithm. The CNN algorithm showed good performance in DCE-MRI image segmentation. The mean, standard deviation, kurtosis, and skewness of K(trans) and V(e), the standard deviation and skewness of K(ep), and the mean and standard deviation of V(p) were statistically considerable in differentiating HGG and LGG (P < 0.05). ROC analysis showed that the standard deviation of K(trans) (0.885) had the highest diagnostic accuracy in distinguishing HGG and LGG. The values of K(trans), V(e), and V(p) were positively correlated with Ki-67 (r = 0.346, P = 0.014; r = 0.335, P = 0.017; r = 0.323, P = 0.022). In summary, the CNN-based DCE-MRI technology had high application value in glioma diagnosis and tumor segmentation.