7/31/ D spatial compounding with the Clarity system: Improving ultrasound image quality in gynaecological image-guided radiotherapy

Transcription

1 in partnership with 3D spatial compounding with the Clarity system: Improving ultrasound image quality in gynaecological image-guided radiotherapy Sarah Mason, Ingrid White, Helen McNair, Susan Lalondrelle, Matthew Blackledge, Tuathan O Shea, Jeff Bamber, and Emma Harris AAPM July 2017 Making the discoveries that defeat cancer The uterus can be visualized with ultrasound 2 Bladder Uterus Superimposition of ultrasound and corresponding CBCT Ultrasound (US) characteristics: Non-ionizing Non-invasive (transabdominal) Excellent soft-tissue contrast Promising alternative to CBCT for daily adaptation of radiotherapy based on softtissue information. Challenges in acquiring high quality transabdominal ultrasound images of the uterus Small bladder volume and high BMI signal attenuation Small field of view Research Portions of Question: the uterus Does may 3D be spatial obstructed compounding by bone or with gasthe Clarity System Segmentation improve the quality algorithms of transabdominal inaccurate ultrasound when image images artifacts of the present. uterus? (S. Mason et al., Med. Phys. (2017) and The the Good, Bad, UGLY! Uterus? Bladder Gas bubble 3 shadowing Manual contour Algorithm segmentation 1

2 3D spatial compounding with Elekta Clarity System Definition: Average (mean) of at least 2 spatially registered ultrasound image volumes to form a composite image Potential benefits: Increase contrast to nose ratio (CNR), extend field of view, and reduce impact of image artifacts on composite image Clarity probe-tracking hardware Example 3D image compound 4 Enables mapping of US images to room coordinates = Study Design 5 Cohort Phantom (n = 1) Healthy Volunteers (n = 4) Patients (n = 4) # individual US images # of time points per subject Total compound sets analyzed Method for assessing image quality CNR, spatial resolution (FWHM of LSF) (empty and full bladders) 21 Image ranking by 3 independent observers 1(worst) 6 (best) (during treatment) 15 Image ranking by 3 independent observers 1(worst) 4 (best) > Original (single) image 2 6 im compounds 1 compound set 7 im compound 6 Phantom Study: Contrast to Noise Ratio (CNR) and spatial resolution results 2

3 Resolution (mm) [Mean FWHM of line spread function] CNR Phantom Study Results: CNR 7-image compound 7 > image compound image (original) Number of images in compound Phantom Study Results: Spatial Resolution 8 1 (original image) Number of images in compound 9 In vivo studies: Image quality measured by observer rankings 3

4 Patient 3 Patient 4 Image Rankings (n = 15) Image Rankings (n = 21) Patient 4 External ostium c = significantly different than original image = significantly different than neighbor Original Image 3-image compound 6-image compound Volunteer 2, Day 2, empty bladder Observer Rankings: Cervical cancer patients 11 Lege c CTV (ut cervix, a of vagin Bladder Uterine External = significantly different than original image = significantly different than neighbor a Original image 2-imC 3-imC 4-imC 12 b Re c 4

5 Summary 13 Image compounding provides significant improvement in CNR (range of improvement from 35% - 104%) No significant differences in spatial resolution Image quality of uterus had significantly higher ranking in image compounds compared with the original image Future Work: Investigate whether performance of segmentation algorithms is improved on compounded images compared with original images in partnership with Acknowledgements: Helen McNair Naomi Hopkins Juniper Radiographer team David Collins Nandita desouza Kate Downey The patients and volunteers that participated in the study! Thank you for your attention! 5

6 6