P05.02.a preliminary findings from the mri quality assurance programme for the prospective multi-site australian fig ([fet-pet in glioblastoma) trog 18.06 study

Abstract BACKGROUND The O-(2-[18F]-fluoroethyl)-L-tyrosine positron emission tomography (FET-PET) in Glioblastoma (FIG) study is a prospective study evaluating the management impact of FET-PET imaging in up to 210 newly diagnosed adult glioblastoma patients across 10 Australian sites. Patients undergo contrast MRI and FET-PET at 3 timepoints: pre-chemoradiotherapy (CRT), one-month post-CRT and at suspected progression. MRI QA programme components, approach and integrated workflows are described here. MATERIAL AND METHODS The FIG MRI protocol comprised 3D-T1, 3D-FLAIR, Axial 2D DWI (including derived ADC map), DCE (including T1 mapping), DSC, Axial T2 and 3D T1-post-contrast (T1C) sequences, with SWI and 3D DIR sequences optional. QA components included data acquisition quality and completeness (standard and advanced), motion artefact, low contrast to noise or signal to noise, series description and susceptibility artifacts. Suitability of T1/T1C using modified RANO (mRANO) criteria was assessed. An optimised workflow involved site upload from local PACS systems to centralised database via TROG Central Quality Management System (CQMS) platform and collation into a central imaging database utilising MIM software. The MRI QA workflow encompassed automated anonymisation of DICOM data, data completeness and reconstruction evaluation, then review by two expert neuroimaging analysts. Sites received feedback with request for resubmission where required. RESULTS Between December 2021 and February 2023, MRI data in 74 patients across 9 sites with median of 6 patients (range 2-17) per site was submitted. A total of 141 MRI datasets across all timepoints were collected (per site range:4-35, median:13), with 43 imaging time-points selected for QA (per site range:3-12, median: 4). Importantly, 41/43 (95%) of initial datasets were deemed suitable for mRANO assessment, but only 13/43 (30%) were suitable for advanced MRI analysis. Very few datasets had motion or susceptibility artifacts, low contrast to noise or signal to noise, or incorrect series description. Technical issues identified included incomplete data (DCE - missing T1 maps, DWI - missing ADC maps), incorrect sequence reconstruction (DCE split series and Axial/Sagittal/Coronal reformats for 3D images) and lack of non-mandatory SWI sequences. Feedback to sites resulted in improvements in DCE sequence acquisition from split series in 11/45 (24%) to preferred single series in 18/24 (75%) and similar increases in T1 maps completeness from 7/19 (37%) to 7/7 (100%). CONCLUSION Despite challenges in multisite workflow and substantial multi-modality site and central data management, a robust MRI QA program has confirmed 95% compliance for mRANO assessment. Site specific feedback resulted in increased compliance with advanced MRI sequences to enable detailed future analysis.