C4.5: programs for machine learning
C4.5: programs for machine learning
| Hi-index | 0.00 |
We propose to develop a portable, handheld, noninvasive solution for accurate screening and real-time monitoring of traumatic brain injury (TBI) in ambulatory/emergency response scenarios. A layered sensing concept that unifies modalities such as a) ultrasound (US) (B-mode, Doppler flow), b) tonometry and c) pulse oximeter to predict TBI, its severity and mode of recommendations for emergency medical service (EMS) personnel is currently investigated. Specifically, we aim to determine novel 3D morphometric parameters of optic nerve sheath (ONS) that can predict elevated intracranial pressure (EICP) from US data. These parameters when combined with intraocular pressure (IOP), blood oxygen saturation (SaO2) and Doppler flow readings of the carotid artery can improve the overall classification accuracy. In addition, we have also developed a preliminary decision-support system (DSS) to provide an automated analysis of subject's brain health status and thereby, recommend further screening, etc. In the demo, we would show the chain of processing starting from capture of our desired signals from a volunteer, pre-processing (reformatting, de-noising) of US data, post-processing of features extracted from the 3D US model and finally, the classification output of the DSS.