Investigation of the optimal freehand three-dimensional ultrasound configuration to image scoliosis: An in-vitro study

Abstract

Scoliosis is a three-dimensional (3D) spinal deformity. 3D ultrasound has been used to image scoliotic spines. This in-vitro study was to investigate the optimal 3D ultrasound parameters by comparing the reconstructed images and the phantom. A medical ultrasound system, a convex and a linear ultrasound probes, both with built-in positioning sensors, were used to scan cadaveric vertebra T7 immersed in a water tank. The operating frequencies were set at 2.5 MHz, 3.3MHz, 4.0 MHz, 6.6 MHz, and 10.0 MHz. The voxel-based method was deployed for the 3D reconstruction. The minimum distances (0.1, 0.2, and 0.3 mm) between two adjacent B-scan images and the reconstruction resolutions (0.2, 0.6, and 1.0 mm) were the inputs for the reconstruction algorithm. A total of 45 configurations were investigated. Four distance parameters were measured three times on both the images and the cadaveric vertebra by one rater in one week apart to minimize memory bias. The mean absolute difference was the distance measurement difference between the images and phantom. The paired Student’s t-test was used to determine the probability between the two populations. The determination of the optimal configuration was based on both accuracy and intuitive image quality. The results showed that the convex probe with the configuration (0.2-mm minimum distance, 0.6-mm reconstruction resolution, and 4.0 MHz) provided the best reconstructed image.