AI-based analysis of the shunt treatment in pre- and post-surgery computed tomography brain scans of iNPH patients

BackgroundThis study examines whether quantifiable changes can be detected in ventricular volume in Idiopathic Normal Pressure Hydrocephalus (iNPH) patients that undergo ventriculo-peritoneal shunt procedures. There is no known metric that characterizes the change in ventricular volume for iNPH patients after shunt placement. MethodsTwo de-identified and independent datasets are studied: O_LI45 brain CT scans (24 diagnosed with iNPH and 21 normal elderly individuals) are used to evaluate the effectiveness of our proposed ventricular volume metric as a diagnostic tool for iNPH. The performance of our deep learning model-based metric is compared to the traditional Evans Index using ROC analysis. C_LIO_LI16 subjects with a total of 50 longitudinal CT scans taken before and after shunt surgery across different imaging centers are studied to quantify the impact of shunt treatment. Clinical symptoms of gait, balance, cognition, and bladder continence are studied with respect to the proposed metric. C_LI ResultsOur proposed metric achieves high accuracy (0.95), precision (0.96), and recall (0.96) in distinguishing between normal and iNPH subjects, surpassing the performance of the Evans Index. This metric allows us to track changes in ventricular volume before and after shunt surgery for 16 subjects. Notably, the 15 subjects with iNPH demonstrate a decrease in ventricular volume post-surgery and a concurrent clinical improvement in their iNPH symptomatology. ConclusionOur novel metric accurately quantifies changes in ventricular volume before and after shunt surgery for iNPH patients, serving as an effective radiographic marker for a functioning shunt in a patient with iNPH. O_LIWhat is already known on this topic - The diagnosis of iNPH involves both clinical and radiographic stigmata. Radiologists rely largely on visual examination of CT scans and provide qualitative evaluations about ventricular volume. C_LIO_LIWhat this study adds - Our study provides quantitative information about the patency and function of the shunt. C_LIO_LIHow this study might affect research, practice, or policy - The validated deep learning-based metric enhances iNPH diagnosis accuracy by tracking radiographic biomarkers. This facilitates decision-making regarding the efficacy of shunt surgery and the effect on brain compliance. We provide a web interface to apply the metric, its reliable performance across multiple institutional scanner types could be adapted to the real-time clinical evaluation of iNPH and improve treatment workflows. C_LI