Semi-Automatic Delineation of the Mitral Valve from Clinical Four-Dimensional Ultrasound Imaging

author:Robert J. Schneider
adviser:Robert D. Howe
year:2011
degree: Ph.D.
institution: Harvard University

Successful clinical treatment of mitral valve disease is dependent on understanding the complexities of patient-specific valve behavior. Mechanical models of the mitral valve, designed to predict valve closure and generated using patient-specific measurements, have proven to be effective tools for studying valve behavior. To be clinically feasible, these models need to be generated from ultrasound images, which are commonly acquired during the diagnosis and treatment of patients. However, methods of generating accurate models from ultrasound data using minimal user-input and interaction are not available.

This work presents methods by which a patient-specific mitral valve model (annulus and leaflets) can be generated from three-dimensional ultrasound. The method first identifies the mitral annulus in a user-specified frame just after valve closure using only a roughly placed valve center point. We find the annulus as the location where the thin leaflet tissue intersects the thicker surrounding tissue. Comparing the results to manual delineations made by experts, we found an average RMS error of 1.81+/-0.78mm. A valve state (open versus closed) predictor and a constrained optical flow algorithm are then used to segment the annulus in the remaining frames in the ultrasound sequence. We compared the automated tracking results to manual delineations made by experts and found an average RMS difference of 1.67+/-0.63mm. The location of the leaflets in a user-specified frame just before valve closure is then automatically found by using the annulus as a means of limiting segmentation efforts. We compared the segmented leaflets to manual tracings and found an average difference of 0.76+/-0.65mm. We then use the leaflet geometry and four-dimensional (4D) annulus in an active surface approach to track the leaflets during valve closure, which for the first time allows for the segmentation of an accurate and detailed coaptation region.

This work additionally describes methods to reconstruct from standard clinical acquisitions high temporal resolution ultrasound sequences required to track the progression of valve closure and generate an accurate coaptation region. The reconstruction method includes a real-time image-based rigid registration method for ultrasound volumes which we use to stabilize images in the presence of respiration and other small movements of the probe.



Harvard BioRobotics Laboratory Home