A Regression Method Based on Noninvasive Clinical Data to Predict the Mechanical Behavior of Ascending Aorta Aneurysmal Tissue.

Abstract

Goal: Ascending aorta aneurysms represent a severe life-threatening condition associated with asymptomatic risk of rupture. Prediction of aneurysm evolution and rupture is one of the hottest investigation topics in cardiovascular science, and the decision on when and whether to surgically operate is still an open question. We propose an approach for estimating the patient-specific ultimate mechanical properties and stress-stretch characteristics based on noninvasive data. Methods: As for the characteristics, we consider a nonlinear constitutive model of the aortic wall and assume patient-specific model coefficients. Through a regression model, we build the response surfaces of ultimate stress, ultimate stretch, and model coefficients in function of patient data that are commonly available in the clinical practice. We apply the approach to a dataset of 59 patients. Results: The approach is fair and accurate response surfaces can be obtained for both ultimate properties and model coefficients. Conclusion: Prediction errors are acceptable, even though a larger patient dataset will be required to stabilize the surfaces, making it possible to apply the approach in the clinical practice. Significance: A fair prediction of the patient aortic mechanical behavior, based on clinical information noninvasively acquired, would improve the decision process and lead to more effective treatments.Goal: Ascending aorta aneurysms represent a severe life-threatening condition associated with asymptomatic risk of rupture. Prediction of aneurysm evolution and rupture is one of the hottest investigation topics in cardiovascular science, and the decision on when and whether to surgically operate is still an open question. We propose an approach for estimating the patient-specific ultimate mechanical properties and stress-stretch characteristics based on noninvasive data. Methods: As for the characteristics, we consider a nonlinear constitutive model of the aortic wall and assume patient-specific model coefficients. Through a regression model, we build the response surfaces of ultimate stress, ultimate stretch, and model coefficients in function of patient data that are commonly available in the clinical practice. We apply the approach to a dataset of 59 patients. Results: The approach is fair and accurate response surfaces can be obtained for both ultimate properties and model coefficients. Conclusion: Prediction errors are acceptable, even though a larger patient dataset will be required to stabilize the surfaces, making it possible to apply the approach in the clinical practice. Significance: A fair prediction of the patient aortic mechanical behavior, based on clinical information noninvasively acquired, would improve the decision process and lead to more effective treatments.