Stitching of 3D ultrasound datasets for the determination of large thyroid volumes - phantom study part II: mechanically-swept probes.

Abstract

To investigate the feasibility and accuracy of 3D-US extended field of view volumetric analyses acquired with mechanically-swept ultrasound probes with different measurement methods for large volume thyroid phantoms. Fifteen thyroid phantoms with different shapes (regular, nodular, thickened isthmus) and volumes (50-400 mL) were created. Two different mechanically-swept US probes were used for the separate scanning of the left and right lobes: convex and linear probe. After specific modifications, the 3D-US datasets were stitched together to an extended field of view using predefined landmarks. Volumetric analyses were performed by conventional ellipsoid model and manual tracing methods. The correspondence of measured and reference volumes was calculated using Pearson's correlation coefficients and limits of agreement according to Bland and Altman. The C-probe proved feasible for the acquisition and processing of the three-dimensional ultrasound extended field of view images; very high levels of agreement (correlation coefficients for volume analyses: 0.9843-0.9992) were observed for all shapes and volumes investigated. The manual tracing method showed superior results in comparison to the ellipsoid model, but was more time consuming. The linear probe was only applicable for the 50 mL phantoms due to its limited field of view. The investigated mechanically-swept convex probe was suitable for the three-dimensional ultrasound extended field of view stitching of large volume thyroid phantoms. Accurate volume analyses could be carried out. The mechanically-swept linear probe is limited to a maximum of 50 m.