Program Information
3D Spatial Compounding Using the Clarity System Improves Ultrasound Image Quality in Gynecological Image-Guided Radiotherapy
S Mason1*, T O'Shea2 , I White3 , H McNair4 , M Blackledge5 , S Laldondrelle6 , J Bamber7 , E Harris8 , (1) ,,,(2) Royal Marsden NHS Foundation Trust, Sutton, Surrey, (3) Royal Marsden NHS Foundation Trust, Sutton, Surrey, (4) Royal Marsden NHS Foundation Trust, Sutton, Surrey, (5) Institute of Cancer Research, Sutton, Surrey, (6) Royal Marsden NHS Foundation Trust, Sutton, Surrey, (7) The Institute of Cancer Research, London, ,(8) Royal Marsden Hospital, Surrey,
Presentations
MO-DE-708-5 (Monday, July 31, 2017) 1:45 PM - 3:45 PM Room: 708
Purpose: A novel method which uses Clarity’s transducer-tracking capability to provide spatially-compounded 3D ultrasound (US) images was developed. Image quality was evaluated as a function of the number of US images used to create a compounded image in a phantom and in vivo.
Methods: 3D-US images registered to the isocenter were acquired from different positions in an US quality assurance phantom (7 images), and transabdominally of the uterus from four volunteers (6 images/imaging session/volunteer; 21 total sessions), and from four cervical cancer patients (4 images/fraction/patient; 15 total fractions). Images were re-sampled onto a Cartesian grid and transferred onto a common frame of reference. Compounded images (imCs) were created by averaging 2,3,4,5,6 or 7 individual images. Phantom: Differences in (1) contrast-to-noise-ratio (CNR) in four gray-scale targets and (2) spatial resolution were determined between the non-compounded image and each imC using ANOVA. In vivo: Three observers independently ranked randomized sets of images (non-compounded and imCs) for image quality of the uterus (rank 1 representing poorest image quality). A Wilcoxon-signed-rank test was used to measure differences in mean rank of 3 observers (MR) between the compounded and non-compounded images.
Results: Phantom: The CNR of imCs was significantly greater, by a range of 38% to 104%, depending on the number of individual images in the ImC and gray-scale target assessed, and spatial resolution was unchanged compared with non-compounded images. In vivo: The non-compounded image had significantly poorer MR than all imCs, with mean [range] MR of 1.3 [1-2.67] and 1.6 [1-2.33], respectively. MR increased significantly with increasing number of 3D-US images used to create a compounded image up to 3 images and 5 images for patient and volunteers, respectively.
Conclusion: Compounding using the Clarity system improves US image quality compared with non-compounded images in both phantom and in vivo.
Contact Email: