Deep Learning-Enabled Clinically Applicable CT Planbox for Stroke with High Accuracy and Repeatability

Background: Computed tomography (CT) plays an essential role in classifying stroke, quantifying penumbra size and supporting stroke-relevant radiomics studies. However, it is difficult to acquire standard, accurate and repeatable images during follow-up. Therefore, we invented an intelligent CT technique with an extremely low cost to evaluate stroke patients during the entire follow-up period. Materials and Method: We deployed a region proposal network (RPN) and V-Net to endow traditional CT with artificial intelligence. Specifically, facial detection was accomplished by identifying adjacent jaw positions through training and testing an RPN on 76,382 human faces using a preinstalled 2-dimensional camera; two regions of interest (ROIs) were segmented by V-Net on another training set with 295 subjects, and the moving distance of the scanning couch was calculated based on a pregenerated calibration table. Multiple cohorts including 1,124 patients were used for performance validation under three clinical scenarios. Results: Cranial Automatic Planbox Imaging Towards AmeLiorating neuroscience (CAPITAL)-CT was performed. The RPN model had an error distance of 4.46±0.02 pixels with a success rate of 98.7% in the training set and a success rate of 100% with 2.23±0.10 pixels in the testing set. We found that in all boundaries, V-Net-derived segmentation maintained a clinically tolerable distance error, within 3 mm on average, and all lines presented with a tolerable angle error, within 3° on average. A calibration table with 2,017,920 matches was generated to support the linkage between the camera and scanner. Real-time, accurate, and repeatable automatic scanning was accomplished with and a lower dose of radiation exposure (all P<0.001). Conclusions: CAPITAL-CT generated standard and reproducible images that could simplify the work of radiologists, which would be of great help in the follow-up of stroke patients and in multifield research in neuroscience. Funding Statement: This study was partially supported by the National Key R&D Program of China. Declaration of Interests: The authors declare no competing interests. Ethics Approval Statement: This study was approved by the ethics commissions of the six hospitals, with a waiver of informed consent.