Feasibility of Dynamic Contrast Enhanced MRI in Oral Cavity Cancer: A Comparison between Reference Region Model, General Kinetic Model and Pathological Grading

Purpose: The growing importance of dynamic contrast-enhanced MRI (DCE-MRI) in the oncological application prompted us to examine its clinical feasibility in patients with suboptimal scanning condition, in our case: the patients of malignant oral cavity tumors susceptible to motion artifact and image distortion due to coughing, swallowing, fixed dental implant and etc. We aimed to examine (1) how successful the DCE-MRI was in image distortion-prone oral cavity and (2) if the consistency exists in two main T1-weighted DCE-MRI perfusion parameters (K and Ve) generated from two different pharmacokinetic models: general kinetic model (GKM) [1] and reference region model (RRM) [2,3], the former one was regarded as the standard DCE-MRI protocol and the latter one was advantaged by no AIF need, but hampered by relatively suboptimal SNR, and (3) to assess if the DCE-MRI parameters are relevant to the pathologic tumor grading. Materials and Methods: Within one year period, 22 patients aged 34-80 years (mean age, 50 ± 10 years) who had oral cavity squamous cell carcinomas were enrolled for preoperative MRI evaluation. Surgical pathologic findings were the reference standard for the tumor grading, while the K and Ve were calculated from DCE-MRI data using two pharmacokinetic models (GKM and RRM). For the RRM analysis, the concentration time curve of a reference region was obtained from the tumor ROI and then fitted to Kovar’s method to estimate K and Ve . The posterior neck muscle away from the tumor used as reference region for normalization was the same for two models. Results: (1) Of the 22 cases, the data from 18 cases were analyzable with the DCE quantification (Fig 1). Four cases with severe imaging distortion