Meeting Banner
Abstract #4617

An Automated Processing Pipeline to Assess and Improve Data Quality for Multicentre Pediatric Dynamic Susceptibility Contrast Imaging

Stephen Powell1,2, Stephanie Withey2,3,4, Yu Sun2,5, James Grist2, Lesley MacPherson6, Laurence Abernathy7, Barry Pizer8, Richard Grundy9, Simon Bailey10, Dipayan Mitra11, Dorothee Auer12, Shivaram Avula7, Theodoros N. Arvanitis2,3,13, and Andrew Peet2,3
1Physical Sciences for Health CDT, University of Birmingham, Birmingham, United Kingdom, 2Institute of Cancer and Genomic Sciences, University of Birmingham, Birmingham, United Kingdom, 3Department of Oncology, Birmingham Children's Hospital, Birmingham, United Kingdom, 4RRPPS, University Hospitals Birmingham NHS Foundation Trust, Birmingham, United Kingdom, 5School of Biological Sciences and Medical Engineering, Southeast University, Nanjing, China, 6Radiology, Birmingham Children's Hospital, Birmingham, United Kingdom, 7Radiology, Alder Hey Children's NHS Foundation Trust, Liverpool, United Kingdom, 8Oncology, Alder Hey Children's NHS Foundation Trust, Liverpool, United Kingdom, 9The Children’s Brain Tumour Research Centre, University of Nottingham, Nottingham, United Kingdom, 10Sir James Spence Institute of Child Health, Royal Victoria Infirmary, Newcastle, United Kingdom, 11Neuroradiology, The Newcastle upon Tyne Hospitals NHS Foundation Trust, Newcastle, United Kingdom, 12Sir Peter Mansfield Imaging Centre, University of Nottingham, Nottingham, United Kingdom, 13Institute of Digital Healthcare, University of Warwick, Coventry, United Kingdom

Obtaining robust perfusion measures from pediatric Dynamic Susceptibility Contrast (DSC-) MRI, such as cerebral blood volume (CBV), is challenging due to variability in acquisition protocols between centres and a heterogeneous patient population. Quality control (QC) is currently carried out by expert qualitative review. An automated QC pipeline was developed which used denoising to salvage data, and assessed data quality using logistic regression classification, with signal-to-noise ratio (SNR) and root mean square error (RMSE) in a gamma variate fit to the first pass as predictors. SNR was the key factor in data quality and denoising is important in assuring appropriate analysis.

This abstract and the presentation materials are available to 2020 meeting attendees and eLibrary customers only; a login is required.

Join Here