Encrypted login | home

Program Information

Characterization of the Structure Sensor as a Surface Imaging Device for Medical Physics Purposes

no image available
P Bardos

P Bardos1*, L Padilla1 , (1) Virginia Commonwealth University, Richmond, VA


WE-RAM3-GePD-J(A)-3 (Wednesday, August 2, 2017) 10:30 AM - 11:00 AM Room: Joint Imaging-Therapy ePoster Lounge - A

Purpose: To assess the accuracy of the surface models captured by the Structure Sensor in comparison to external CT contours. To examine the effects of different measurement distances and resolution settings on the quality of the surfaces acquired with the Structure sensor.

Methods: A Structure Sensor (Occipital, Inc.), mounted on an iPad and with the “Scanner” application (Occipital, Inc.), was used to create surface models of the Rando phantom head. Surfaces were recorded with high and low resolution settings at distances of 40 cm and 60 cm from the phantom. The external contour of the phantom from a CT scan (3mm slices, 120 kV, 30 mA, outside patient air threshold=0.6g/cc) was used as the “ground truth”. The surfaces were analyzed in MATLAB with an in-house code. The Iterative Closest Point algorithm was used to register the surfaces and the separation between the registered surfaces was obtained using the nearest neighbor algorithm.

Results: Both the resolution setting and the scanning distance from the object had an impact on the surface quality. The mean ± standard deviation values were 2.789 ± 1.004mm (low resolution at 40cm), 1.182 ± 0.225mm (low resolution at 60cm), 1.776 ± 0.403mm (high resolution at 40cm), and 1.660 ± 0.563mm (high resolution at 60cm). Captures taken at larger distances showed better agreement with the external surface of the CT scan.

Conclusion: This work shows the feasibility of the Structure Sensor as a tool to create patient surface models. Our tests indicate that the high or low resolution settings at larger distances provide surfaces that match the external CT contours within 2mm. This accuracy shows its potential use for a variety of clinical applications. This work will be used in the future for the creation of a patient model for collision detection during treatment planning.

Funding Support, Disclosures, and Conflict of Interest: This work was funded by a grant from Philips.

Contact Email: