Slice of a digital phantom with centerline points overlaid (left), same slice with computed from one segment (middle), and 3D view of the digital phantom (right), with initial centerline points in dark, connected points with our proposed speed function in lighter gray, and connected points using constant speed in white.
Left: Initial centerline points obtained from CTA, connected centerline points using FM and end points and bifurcation points. Middle: slice of the distance field from one segment. Right top: centerlines of the postprocedure model (light), and preprocedure centerline (dark) avoiding the calcified plaque pointed by arrows. Right bottom: corresponding CCA image.
CCA images of a treated patient. Top row: preprocedure, bottom row: postprocedure. Overlay images are from left to right: segmentation, distances from the vessel boundary, and centerline branches colored according to the the lengths from .
Phantom used for the experiment, with tube centerlines and ground truth models overlaid, (a); angiography image at 0°, (b); 3D volume rendering of the CT data, (c); CT slice with tube segmentation overlaid, (d).
Detail of the preprocedure (first row) and postprocedure CCAs (second row), used to construct the vessel models (last two rows) in the five patients studied; in the third row, first three columns: preCTA+preCCA models (inner models) vs postCTA+postCCA models (outer models); in the third row, last two columns: preCTA+preCCA models; in the fourth row: preCTA+preCCA models (inner models) vs preCTA+postCCA models (outer models).
Surface models obtained from the phantom at four different angle acquisitions with our method, colored by the distance to the ground truth.
Top row: Pre-procedure models semiautomatically generated using pre-CTA + pre-CCA (inner models) and manually generated using pre-CTA alone (outer models) in the five patients studied. Bottom row: corresponding semiautomatic pre-CTA + pre-CCA models with colors representing the distances to the pre-CTA manual models.
Top row: Semiautomatic models generated using post-CTA + post-CCA (inner models) and manual models from post-CTA alone (outer models). Middle row: Semiautomatic post-CTA + post-CCA models with colors representing the distances to the post-CTA manual models. Bottom row: Semiautomatic pre-CTA + post-CCA models with colors representing the distances to the post-CTA manual models. These models correspond to patients 1, 2, and 3.
Average absolute differences, in mm, of the measured vessel radius between the 3D models and the CCA images in all patients, with the SD shown as thin lines over the bars. Top row: semiautomatic CTA + CCA models; bottom row: manual models from CTA. First column: preprocedure models (pre-CTA + pre-CCA and manual pre-CTA); second column: postprocedure models (post-CTA + post-CCA and manual post-CTA), third column (top): predictive postprocedure models (pre-CTA + post-CCA). Bottom right: radii values vs distance from the bifurcation center, for one bifurcation branch extracted from a pre-CTA manual model a pre-CTA + pre-CCA model and from CCA.
Murray ratio for each bifurcation treated, computed using semiautomatic CTA + CCA models, using manual CTA models and CCA images, in the preprocedure (left) and postprocedure cases (right).
Summary of patient data. P. T-stent: Provisional T-stenting. P. T-stent wFK: Provisional T-stenting with final kissing.
Dice similarity index and average distances computed between 3D semiautomatic models generated using CT + CCA and 3D ground truth models of the phantom for two different bifurcations at four different angle positions of the angiography.
Dice similarity index and average distances (in mm) computed between 3D semiautomatic models generated using CTA + CCA and manual models generated using CTA alone for preprocedure and postprocedure cases and for the five patients studied.
Avarage radii values of vessels measured in CCA data along the vessel centerlines in mm.
Article metrics loading...
Full text loading...