Data Mining Based Noninvasive Intracranial Pressure Assessment

描述（由申请人提供）：现有颅内压（ICP）评估技术是侵入性的，并且对于许多神经外科疾病，需要穿透颅骨以放置传感器。在许多临床情况下，ICP信息可能具有重要的诊断和/或预后意义，与侵入性ICP程序相关的风险可能会影响ICP评估。本项目的总体目标是开发和验证一种准确的、无创的ICP评估方法，该方法能够根据颅内动力学、脑血流动力学、信号处理和软件工程的知识估计ICP。 本项目的目标是开发和验证一种基于数据挖掘的无创ICP评估方法。该方法是基于一个创新的数据挖掘框架，该框架统一了在构建ICP模拟模型的过程中所涉及的各个模块。它从信号数据库中获取信息，该数据库包含来自患者人群的同时动脉血压（ABP）和脑血流速度（CBFV）的记录。在本研究中，将使用同步侵入式ICP测量进行验证。为了实现这一目标，将解决以下具体目标：1）实现自适应信号分割算法，该算法将有助于信号数据库的构建; 2）实现非线性映射函数，用于将血流动力学特征向量与ICP估计值的相异性度量相关联; 3）比较并验证DM NICP与两个不同患者人群数据库的性能：创伤性脑损伤（TBI）和正常压力脑积水（NPH）。 如果得到验证，这种无创ICP原型可以以具有成本效益的方式集成到需要无创ICP评估的现有临床协议的工作流程中。无创ICP模块的集成可能会改变患有各种神经外科疾病的患者的当前治疗和管理方案，由于当前ICP测量技术的侵入性，可能尚未对这些患者进行充分的ICP评估。

DESCRIPTION (provided by applicant): Existing intracranial pressure (ICP) assessment techniques are invasive, and for many neurosurgical conditions, require penetration of the skull for placement of the sensor. The risks associated with invasive ICP procedures may obviate ICP assessment in many clinical situations in which ICP information could be of vital diagnostic and/or prognostic importance. The overall goal of this project is to develop and validate an accurate, noninvasive ICP assessment method that is able to estimate ICP based on knowledge from intracranial dynamics, cerebral hemodynamics, signal processing and software engineering. The objectives of this project are to develop and validate a data mining based, noninvasive ICP assessment method. The method is based on an innovative data mining framework that unifies various modules involved in the process of constructing an ICP simulation model. It acquires information from a signal database that contains records of simultaneous arterial blood pressure (ABP) and cerebral blood flow velocity (CBFV) from a population of patients. In this study, simultaneous invasive ICP measurement will be used for validation purposes. To accomplish this objective, the following specific aims will be addressed: 1) to implement an adaptive signal segmenation algorithm that will assist the construction of signal database; 2) to implement a nonlinear mapping function for relating a hemodynamic feature vector to a dissimilarity measure of an ICP estimate; 3) to compare and validate the performance of DM NICP with two distinct patient population databases: traumatic brain injury (TBI) and normal pressure hydrocephalus (NPH). If validated, this noninvasive ICP prototype can be integrated, in a cost-effective way, into the workflow of an existing clinical protocol where noninvasive ICP assessment is needed. The integration of a noninvasive ICP module will potentially alter the current treatment and management protocol for patients with various neurosurgical disorders, for whom adequate ICP assessment has probably not been done due to the invasive nature of current ICP measurement techniques.