Stereo photogrammetry soft tissue imaging

Clinical Center

Investigators

Linked publications & trials

Abstract

The first step performed to achieve the goals stated above was to assess the technical capabilities of a commercially available stereophotogrammetry (SP) system. We performed a study to determine the reliability, stability, validity and precision of a SP system for use in quantifying the complex three-dimensional (3D) structure of the human torso. We performed assessments of the system using images of geometric solids and a human-form mannequin. Analysis of geometric solids revealed excellent intra- and inter-rater reliability of the system for linear, surface area and volume measurements (r > 0.99, P < 0.001). Overall, no significant difference was found between SP and manual measurements (F = 4.23, P > 0.06). The system exhibited excellent stability in images of the mannequin over time (r > 0.99). The limit of precision (error > 5%) of the system to detect objects on the surface of the mannequin was estimated at an object size of 23.5cm2 for surface area and 32mL for volume. After confirming the stability of the SP image acquisition system with human-form mannequin, a human subjects protocol was approved by the NIDCR IRB. A materials transfer agreement with the University of Alberta, Canada was completed in order to share MatLab image analysis software developed by investigators there In this first phase of the clinical study, 102 healthy volunteers were imaged using the SP system, and these images were processed into transverse torso slices between 2.5 and 6 mm using the Matlab image analysis software. The coefficients of variation were calculated for each thickness for desired outcome measures such as pseudocobb angle or maximum difference in area between the right and left half of the torso. Preliminary analyses of data obtained in this phase of the study were presented in March 2013 at the Annual Meeting of the Association of Academic Physiatrists. It was found that all outcome measures had coefficients of variation that were at or below the 30% threshold for stability, with the exception of the pseudoCobb measure. Modifications in the pseudocobb analysis algorithm have recently been completed and have yielded more reliable results. The findings above indicate that this SP system does have the potential to provide quantifiable, multi-planar information about spinal and chest wall deformities. A report of the results of this phase is currently in preparation for publication. The phase one results have allowed us to move on to our second phase of the study and we are continuing to accrue subjects with scoliosis and/or rib deformities in order to assess the ability of the system to discriminate healthy subjects from those with scoliosis, and chest wall deformity. We have now accrued the first 50 subjects with scoliosis and are beginning interim analyses. We have also begun testing the system's sensitivity to detect changes in deformity over time. Subsequent to this accrual and analysis, the information gained will be applied to accrual and development of analyses of the distribution of adiposity in persons with truncal obesity.