1. Quality‐based pharmacokinetic model selection on DCE‐MRI for characterizing orbital lesions
- Author
-
Mathieu Zmuda, Olivier Bergès, Jonathan El Methni, Jean-Claude Sadik, Louise Marais, Augustin Lecler, Julien Savatovsky, Daniel Balvay, O. Galatoire, Kevin Zuber, Laure Fournier, C.A. Cuenod, and Edgar Farah
- Subjects
Adult ,Male ,Multivariate analysis ,Quality Assurance, Health Care ,Population ,Contrast Media ,Blood volume ,Logistic regression ,Permeability ,030218 nuclear medicine & medical imaging ,03 medical and health sciences ,0302 clinical medicine ,Image Processing, Computer-Assisted ,Medical imaging ,Humans ,Medicine ,Radiology, Nuclear Medicine and imaging ,Prospective Studies ,Stage (cooking) ,education ,education.field_of_study ,Receiver operating characteristic ,business.industry ,Univariate ,Reproducibility of Results ,Middle Aged ,Magnetic Resonance Imaging ,Perfusion ,ROC Curve ,Multivariate Analysis ,Orbital Neoplasms ,Female ,Nuclear medicine ,business ,Algorithms - Abstract
BACKGROUND Although several studies have evaluated dynamic contrast-enhanced (DCE) MRI in the orbit, showing its utility when detecting and diagnosing orbital lesions, none have evaluated the pharmacokinetic models. PURPOSE To provide a quality-based pharmacokinetic model selection for characterizing orbital lesions using DCE-MRI at 3.0T. STUDY TYPE Prospective. POPULATION From December 2015 to April 2017, 151 patients with an orbital lesion underwent MRI prior to surgery, including a high temporal resolution DCE sequence, divided into one training and one test dataset with 100 and 51 patients, respectively. FIELD STRENGTH/SEQUENCE 3T/DCE. ASSESSMENT Six different pharmacokinetic models were tested. STATISTICAL TESTS Univariate and multivariate analyses were performed using Wilcoxon-2-sample tests and a logistic regression to compare parameters between malignant and benign tumors for each pharmacokinetic model for the whole cohort. Receiver operating characteristic (ROC) curve analyses were performed on the training dataset to determine area under curve (AUC) and optimal cutoff values for each pharmacokinetic model, then validated on the test dataset to calculate sensitivity, specificity, and accuracy. RESULTS Regardless of the model, tissue blood flow and tissue blood volume values were significantly higher in malignant vs. benign lesions: 103.8-195.1 vs. 65-113.8, P [
- Published
- 2019
- Full Text
- View/download PDF