Program Information
Impact of Pulse Sequence Selection On T1-Weighted MRI Radiomic Textural Features
F Yang, J Ford, N Dogan, Department of Radiation Oncology, University of Miami School of Medicine, Miami, FL
Presentations
SU-I-GPD-J-61 (Sunday, July 30, 2017) 3:00 PM - 6:00 PM Room: Exhibit Hall
Purpose: Radiomic textural features extracted from diverse MRI modalities have been investigated regarding their predictive and prognostic values in a variety of cancers. However, the impact of pulse sequence selection on MRI-derived radiomic features has not yet been fully assessed. With the aid of a 3D realistic digital MRI phantom, the objective of this study was to examine the role of pulse sequence selection in the validity and reliability of T1-weighted MRI radiomic metrics.
Methods: MR images of the employed digital phantom were realized with SimuBloch, a simulation package made for fast generation of image sequences based on Bloch equations. Pulse sequences being investigated consisted of spin echo (SE), gradient echo (GRE), inversion recovery spin echo (IR-SE), inversion recovery gradient echo (IR-GRE), and spoiled gradient echo (SP-GRE). 29 radiomic features related respectively to gray-level histogram (GLHI), co-occurrence matrices (GLCOM), zone size matrices (GLZSM), and neighborhood difference matrices (GLNDM) were evaluated for each resultant image within two VOIs with VOI_I featuring less spatial heterogeneity and VOI_II having strong heterogeneous patterns on the ground truth image.
Results: For VOI_I radiomic features extracted from images simulated with the pulse sequence of SE, GRE, IR-SE, IR-GRE, and SP-GRE varied with a range of 0.5-498%, 2.3-409%, 0.4-889%, 1.3-134%, and 2.1-387% relative to their ground truth data, respectively, while varying with a range of 0.9-396%, 1.6-272%, 2.3-336%, 0.1-240% and 1.3-348% for VOI_II. When comparing across the five assessed pulse sequences, coefficient of variations of radiomic features varied ranging from 2-115% for VOI_I while ranging from 0.5-76% for VOI_II.
Conclusion: Variability of textural appearance on T1-weighted MRI with respect to the choice of pulse sequence is significant and feature-dependent. It calls for caution in employing MRI radiomic features especially through pooling data from multiple institutions towards correlating treatment response and clinical outcomes.
Contact Email: