Noninvasive prediction for intracranial pressure related parameters in traumatic brain injury patients using computed tomography radiomic features.

Abstract

Abstract Background: Current intracranial pressure (ICP) related parameters monitoring is invasive and tends to cause complications, which limited their use to predict patients’ intracranial status and prognosis. Objective: To utilize postoperative computed tomography (CT) images radiomic features techniques to predict abnormal ICP related parameters levels consisting of an index of cerebrospinal compensatory reserve(RAP) and a pressure reactivity index (PRx)in traumatic brain injury (TBI) patients noninvasively. Methods: 48 patients were enrolled and randomized to training (n=34) and test (n=14) sets. A total of 107 radiomic features were extracted from each patients’ CT image. Their clinical and imaging data was collected and analyzed to establish prediction models of RAP and PRx respectively. Pearson correlation and univariate regression analysis were used for feature selection, multivariate logistic regression was used to develop the predicting models. The performance of models was assessed with their discrimination, calibration and clinical use. Results: The RAP model showed a good discrimination with the area under receiver operating characteristic curve (AUC) of training and test set were 0.771 and 0.727, and a good calibration; The performance of PRx model was inferior to the RAP model, but still have a significant discrimination with the AUCs of training and test were 0.713 and 0.667. Decision curve analysis indicated the prediction model have the potential clinical utility. Conclusion: The study illustrated that CT radiomic features as a clinical aid may have ability to predict ICP related parameters to reflect the intracranial condition of TBI patients noninvasively, given its potential for clinical treatment guidance and prognosis indication.