Rater Agreement of Temporomandibular Joint MRIs at 0.55T and 1.5T

Objectives: MRI is the gold standard for imaging of the soft tissue components of the temporomandibular joint (TMJ). Typically, neuroradiologist interpret the images obtained using a clinically available scanner. We assessed intra- and inter-rater agreement in evaluation of TMJ structures obtained using a more accessible new 0.55T scanner and a clinical standard 1.5T MRI scanner interpreted by a neuroradiologist and dentists.
Methods: Six observers (1 neuroradiologist, 1 oral radiologist, 4 orofacial pain specialists) evaluated 40 image-stacks, presented in a blinded randomized fashion. Images were obtained on a 0.55T scanner equipped with a custom-built surface coil for dental applications and 1.5T scanner with a standard coil setup. Standard parasagittal and paracoronal proton-density, with and without fat saturation, as well as T2-weighted images from 5 healthy participants were acquired on both scanners. Various hard and soft tissue anatomical components of the TMJs necessary to derive diagnoses, as well as image quality parameters (noise, resolution, contrast) were assessed using a 3-point scale (0=unacceptable, 1=acceptable, 2=excellent). No training was provided to raters. Intra- and inter-rater agreement were assessed using weighted kappa, while Landis and Koch criteria were used to determine levels of agreement.
Results: Pooled data for both hard and soft tissue anatomy revealed intra-rater kappa values ranging from 0.61 to 0.67 for 0.55T images (95%CI: 0.41 to 0.85) and 0.63 to 0.71 for 1.5T images (95%CI: 0.38 to 0.89). For inter-rater kappa values, point estimates ranged from 0.02 to 0.29 for 0.55T and 0.02 to 0.23 for 1.5T when assessing all possible combinations between raters.
Conclusions: Intra-rater agreement was substantial among all raters for both 0.55T and 1.5T scanners. Inter-rater agreement was slight to fair among all raters for both scanners’ images. This small study suggests images from both scanners are interpreted by raters similarly, despite each rater using the scale differently.