(Wednesday, 8/1/2018) 7:30 AM - 9:30 AM
Room: Room 202
Purpose: Several clinical trials that make decisions based upon quantitative imaging (QI) metrics are on-going. It is critical to assess whether QI metrics acquired in each patient who is enrolled on the trial are reliable to support clinical decision. This study developed and evaluated a methodology and metrics to assess accuracy and repeatability of blood volume (BV) maps derived from DCE-MRI in a phase-II randomized clinical trial for poor prognosis head and neck cancers (HNC).
Methods: Fifty-one HNC patients enrolled on the clinical trial underwent DCE-MRI on 3T MR scanner at preRT and 2wkRT. 3D T1-weighted DCE images were acquired using a gradient-echo sequence in sagittal orientation to cover primary and nodal cancers, carotid artery, and cerebellum. The latter was out of treatment region thus used as a reference region to quality assessment (QA) of BV maps derived from DCE images. After registering BV maps to post-Gd T1-weighted images at preRT, BV values in volumes of interest of cerebellum were obtained at the pre-RT and 2wkRT. Descriptive statistics, within- and between-subject means of squared errors (WMS and BMS, respectively), and repeatability coefficient (RC) of BV from two examinations were calculated. BV changes in individual patients were compared to the RC to determine reliability of BV measurements.
Results: A mean BV value (2.22±0.14 sd) at preRT was not significantly different from the one (2.23±0.18) at 2wkRT(p=0.7). However, three patients were identified to have BV differences between two measurements far beyond the RC (16.85%) with 95% confidence. Repeatability of arterial input function (AIF) peaks cannot solely explain the repeatability of BV values.
Conclusion: Our method can identify individual patients who have unreliable BV values during a clinical trial. Accuracy and precision of QI metrics can be measured in individual patients compared to group statistics and repeatability coefficients to support a clinical trial.
Funding Support, Disclosures, and Conflict of Interest: This research work was supported by NIH grants: U01CA183848 and R01CA184153
Keywords
MRI, Quantitative Imaging, Quality Assurance
Taxonomy
IM- MRI : Quantitative imaging/analysis
Contact Email
