Craniocerebral Magnetic Resonance Imaging Features of Benign Paroxysmal Positional Vertigo under Artificial Intelligence Algorithm and the Correlation with Cerebrovascular Disease

Abstract

This research aimed to discuss the characteristics of benign paroxysmal positional vertigo (BPPV) and the correlation with cerebrovascular disease. An artificial intelligence algorithm under a parallel dual-domain concatenated convolutional neural network (PDDC-CNN) was proposed to process the images of magnetic resonance imaging (MRI). MRI, magnetic resonance angiography (MRA), and susceptibility weighted imaging (SWI) were performed on all 60 research objects with a 3.0 MRI scanner. The number of cases with cerebral microbleeds (CMBs), SWI image display of small veins, the number of lacunar infarctions, vertebral artery dominance, and vertebrobasilar morphology were observed in the two groups. The number of lacunar infarctions was 2.400 ± 3.358 and 0.672 ± 0.251, respectively, in the BBPV group with 30 cases and the control group with the other 30 cases. The positive rates of CMBs on SWI images were 48% and 27% in the BBPV group and the control group, respectively, and the average CMBs were counted as 1.670 ± 2.326 and 0.487 ± 0.865. CMBs were shown as round or oval lesions of conventional sequence deletion in the images with a diameter of less than 1.5 cm. SWI images of the BBPV group showed a significant increase in intracerebral small veins compared to those of the control group. The curvature of the vertebrobasilar artery in the BBPV group was significantly higher than that in the control group, and the curvature of the basilar artery was slightly higher than that in the control group. In conclusion, the MRI features of BPPV patients were related to their own microvascular lesions closely, and it was speculated that the cerebrovascular factors might play a dominant role in the early onset of BPPV.