ABSTRACT
PROBLEM: The purpose of this work is to provide some validation methods for evaluating the hemodynamic assessment of Cerebral Arteriovenous Malformation (CAVM). This article emphasizes the importance of validating noninvasive measurements for CAVM patients, which are designed using lumped models for complex vessel structure. METHODS: The validation of the hemodynamics assessment is based on invasive clinical measurements and cross-validation techniques with the Philips proprietary validated software's Qflow and 2D Perfursion. RESULTS: The modeling results are validated for 30 CAVM patients for 150 vessel locations. Mean flow, diameter, and pressure were compared between modeling results and with clinical/cross validation measurements, using an independent two-tailed Student t test. Exponential regression analysis was used to assess the relationship between blood flow, vessel diameter, and pressure between them. Univariate analysis is used to assess the relationship between vessel diameter, vessel cross-sectional area, AVM volume, AVM pressure, and AVM flow results were performed with linear or exponential regression. DISCUSSION: Modeling results were compared with clinical measurements from vessel locations of cerebral regions. Also, the model is cross validated with Philips proprietary validated software's Qflow and 2D Perfursion. Our results shows that modeling results and clinical results are nearly matching with a small deviation. CONCLUSION: In this article, we have validated our modeling results with clinical measurements. The new approach for cross-validation is proposed by demonstrating the accuracy of our results with a validated product in a clinical environment.
ABSTRACT
Cerebral arteriovenous malformations (CAVM) are congenital lesions that contain a cluster of multiple arteriovenous shunts (NIDUS). Cardiac arrhythmia in CAVM patients causes irregular changes in blood flow and pressure in the NIDUS area. This paper proposes the framework for creating the lumped model of tortuous vessel structure near NIDUS based on radiological images. This lumped model is to analyze flow variations, with various combinations of the transient electrical signals, which simulate similar conditions of cardiac arrhythmia in CAVM patients. This results in flow variation at different nodes of the lumped model. Here we present two AVM patients with evaluation of 150 vessels locations as node points, with an accuracy of 93%, the sensitivity of 95%, and specificity of 94%. The calculated p-value is smaller than the significance level of 0.05.
