Accurate 3D reconstruction of bony surfaces using ultrasonic synthetic aperture techniques for robotic knee arthroplasty

Kerr, William and Rowe, Philip and Pierce, Stephen Gareth (2017) Accurate 3D reconstruction of bony surfaces using ultrasonic synthetic aperture techniques for robotic knee arthroplasty. Computerized Medical Imaging and Graphics, 58. pp. 23-32. ISSN 0895-6111 (

[thumbnail of Kerr-etal-CMIG-2017-Accurate-3D-reconstruction-of-bony-surfaces-using-ultrasonic]
Text. Filename: Kerr_etal_CMIG_2017_Accurate_3D_reconstruction_of_bony_surfaces_using_ultrasonic.pdf
Final Published Version
License: Creative Commons Attribution 4.0 logo

Download (3MB)| Preview


Robotically guided knee arthroplasty systems generally require an individualized, preoperative 3D model of the knee joint. This is typically measured using Computed Tomography (CT) which provides the required accuracy for preoperative surgical intervention planning. Ultrasound imaging presents an attractive alternative to CT, allowing for reductions in cost and the elimination of doses of ionizing radiation, whilst maintaining the accuracy of the 3D model reconstruction of the joint. Traditional phased array ultrasound imaging methods, however, are susceptible to poor resolution and signal to noise ratios (SNR). Alleviating these weaknesses by offering superior focusing power, synthetic aperture methods have been investigated extensively within ultrasonic non-destructive testing. Despite this, they have yet to be fully exploited in medical imaging. In this paper, the ability of a robotic deployed ultrasound imaging system based on synthetic aperture methods to accurately reconstruct bony surfaces is investigated. Employing the Total Focussing Method (TFM) and the Synthetic Aperture Focussing Technique (SAFT), two samples were imaged which were representative of the bones of the knee joint: a human-shaped, composite distal femur and a bovine distal femur. Data were captured using a 5MHz, 128 element 1D phased array, which was manipulated around the samples using a robotic positioning system. Three dimensional surface reconstructions were then produced and compared with reference models measured using a precision laser scanner. Mean errors of 0.82 mm and 0.88 mm were obtained for the composite and bovine samples, respectively, thus demonstrating the feasibility of the approach to deliver the sub-millimetre accuracy required for the application.