Fig. 4

We fit three different idealized geometries to the segmented PVS, and we fit a circle to the vessel. A, C, and D Three idealized PVS geometries: spline, polynomial, and ellipse fits. The spline fit is defined by vessel radius \(R_{1}\), offset \(O_{f}\), and points \(P_1\)–\(P_6\), whose locations are specified in Table 2. The polynomial fit is defined by vessel radius \(R_{1}\), width W from the vessel to the end of the PVS, height \(H_{C}\) from the top of the vessel top of the PVS, height \(H_{1}\) of the PVS next to the vessel, height \(H_{2}\) of the PVS midway between the vessel and the end, and height \(H_{3}\) of the PVS at the point furthest from the vessel. The ellipse fit is defined by vessel radius \(R_{1}\), major axis \(R_{2}\), minor axis \(R_{3}\), horizontal offset \(O_{\rm h}\), and vertical offset \(O_{\rm v}\). We fit each of these idealized shapes to the PVS and vessel segmentation to determine which best characterized the PVS. B A cross section of a vessel and an outline of its segmentation. We fit a circle to points located on the edge of the segmentation. Because image quality degraded with depth, the segmentation on the bottom of the vessel has greater uncertainty. Edge points that were inside an eroded convex hull of the segmentation were excluded from the fit