Effects of sampling frequency on the accuracy of phenomenological and pharmacokinetic parameters derived from dynamic contrast enhanced MRI data. (October 2017)

• Record Type: Journal Article
• Authors: Knight, Silvin P.; Browne, Jacinta E.; Meaney, James F.; Fagan, Andrew J.
• Is Part Of: Physica medica. Volume 42(2017)
• Page Start: 353
• Publication Date: 2017-10
• Subjects: Medical physics -- Periodicals; Biophysics -- Periodicals; Biophysics -- Periodicals; Imagerie médicale -- Périodiques; Radiothérapie -- Périodiques; Rayons X -- Sécurité -- Mesures -- Périodiques; Physique -- Périodiques; Médecine -- Périodiques; 610.153
• DOI: 10.1016/j.ejmp.2017.05.006
• Languages: English
• ISSNs: 1120-1797

Abstract

Abstract : The sampling frequencies (fs) at which dynamic contrast enhanced (DCE) MRI data are acquired is known to affects the accuracy of derived parameters. Phenomenological parameters empirically characterise the shape and structure of the signal-time curves, whereas pharmacokinetic approaches estimate physiologically relevant parameters from the concentration-time curves, such as the volume-transfer constant (Ktrans) and the extravascular-extracellular fractional volume (ve). Both of these approaches have shown promise in the detection of prostate cancer, however to date a method has been lacking to quantitatively compare MR-measured to known 'ground truth' parameter values. In the present work, a newly developed anthropomorphic DCE-MRI prostate phantom was utilised to this end. Precisely known 'healthy' and 'tumorous' curves were measured using a 3T scanner (Philips, Netherlands) and 32-channel detector coil (3D-SPGR, TR/TE = 4.3/1.4 ms, α = 10°, FOV = 224 × 224 × 80 mm 3, voxels = 1 × 1 × 4 mm 3 ) at fs = 0.5, 0.26, 0.19, 0.12, 0.061, and 0.041 Hz. Phenomenological analysis revealed that wash-in was measured with the lowest errors ( < 15%) at fs ⩾ 0.12 Hz, whereas wash-out gave comparatively higher errors (up to 27%) across this same range of fs. Lowest errors in time-to-peak ( < 11%) were measured using fs = 2 s and 8.1 s. Pharmacokinetic analysis using the Tofts model gave errors in Ktrans and ve values < 14% and < 10%, for acquisitions with fs ⩾ 0.12 Hz and fs ⩾ 0.061 respectively, with errors increasing dramatically (up to 172%) with fs < 0.061 Hz. Quantitative phantom-based approaches, such as the one used herein, offers the prospect of a standardisation in the way DCE-MRI is performed in the prostate, and possibly a wider acceptance of the technique for routine clinical use. Acknowledgement: This work is funded by Irish Cancer Society Research Scholarship CRS13KNI (supported by Movember).