Automatic Diagnosis and Measurement of Intracranial Aneurysms Using Deep Learning in MRA Raw Images

Qingning Yang ^1,2 Fengxi Chen ³ Li Li ^1,2 Rong Zeng ⁴ Jiaqing Li ^1,2 Jingxu Xu ⁵ Chencui Huang ⁵ Junbang Feng ^1,2 Chuanming Li ^1,2*

• ¹ Chongqing Key Laboratory of Emergency Medicine, Chongqing Emergency Medical Center, Chongqing, China
• ² Chongqing University Central Hospital, Chongqing, China
• ³ Southwest Hospital, Army Medical University, Chongqing, China
• ⁴ Second Affiliated Hospital, Chongqing Medical University, Chongqing, China
• ⁵ Department of Research Collaboration, R&D Center, Beijing Deepwise & League of PHD Technology Co., Beijing, China

The traditional procedure of intracranial aneurysm (IA) diagnosis and evaluation in MRA is manually operated, which is time-consuming and labor-intensive. In this study, a deep learning model was established to automatically diagnose and measure IA based on original MR images.: 1014 IAs ( 852 patients ) from the hospital 1 were included and randomly divided into training set, test set and internal validation set in a ratio of 7:2:1. Additionally, 315 patients (179cases with IA, 136 cases without IA) from the hospital 2 were used for independent validation. A deep learning model of MR 3DUnet was established for IA diagnosis and size measurement. The True Positive (TP), False Positive (FP), False Negative (FN), Recall, Sensitivity and Specificity indexes were used to evaluate the diagnosis performance of the MR 3DUnet. The two-sample t-test was usedto compare the difference between the size measurement results of the MR 3DUnet and two radiologists. P < 0.05 was considered to be statistically significant.The fully automatic model could process original MRA data with 13.6s and obtain real-time results including IA diagnosis and size measurement. For the IA diagnosis, in the training set, testing set, and internal validation set, the recall rates were 0.80, 0.75, 0.79, and the sensitivities were 0.82, 0.75, and 0.75, respectively. In the independent validation set, the recall rate, sensitivity specificity and AUC were 0.71, 0.74, 0.77 and 0.75, respectively. Subgroup analysis showed a recall rate of 0.74 for IAs diagnosis based on DSA. For the IA size measurement, no significant difference was found between our MR 3DUnet and the manual measurements of DSA or MRA.: In this study, a one-click fully automatic deep learning model was developed for automatic IA diagnosis and size measurement based on 2D original images. It could greatly help doctors improve their work efficiency and save patients' examination time, which was of great significance in clinical work.