The temporomandibular joint (TMJ) is a synovial joint composed of an articular disc, mostly fibrous, interposed between two bone surfaces (mandibular condyle and glenoid fossa). TMJ kinematics is complex [1] because it involves numerous muscular structures interacting with the bone and cartilaginous tissues and has six degrees of freedom. The progressive deterioration of the articular surfaces of the TMJ or the muscle imbalance leads to the appearance of temporomandibular dysfunction (TMD) [2]. These disorders are common and have reported a prevalence of 25% in the adult population [3]. TMD could lead to joint pain, clicking or grinding sounds when performing mandibular movements. In the long term and in the absence of appropriate treatment, these disorders can impact the quality of life of patients by limiting the amplitude of mandibular movements, altering masticatory muscle function, prematurely wearing down the articular disc, or even generating osteoarthritis. So, dynamical visualization of the temporomandibular joint is important [4] to improve the knowledge about the specific movements of its structures [5].

Multiple approaches exist to study TMJ kinematics [6]. They include commonly used mechanical axiographs [7], electromagnetic articulography [8], and various dedicated devices [9,10]. These devices are affordable and could be directly accessible in the clinician's office. However, the principal drawback of these techniques is that extra dermic and dental reference points are used to record the global motion of the mandibular bone but do not give any information regarding the real TMJ soft and hard tissue variations [11], and the acquired data is insufficiently reliable [12]. Furthermore, these devices are cumbersome and uncomfortable. Other devices such as infrared [13] and optical tracking systems [14] are considerably more comfortable but still rely on external tracking markers. Currently, only axiographs are widely used by clinicians; electromagnetic articulography is commercialized but used mostly for research purposes, and other devices are in the experimental prototype stage.

In parallel with these devices, TMJ's anatomical structures can be observed due to radiological acquisitions such as Computed Tomography scan (CT-Scan) [15] and Magnetic Resonance Imaging (MRI) [16]. MRI has the advantage to be a non-invasive and non-ionizing technology. However, conventional acquisition protocols imply static positions and thus do not allow assessment of the entire TMJ's movement [17]. Standard static MR images, generally made in two key positions (closed and open jaw), have been shown to frequently not detect symptomatic cases (e.g. temporomandibular disk displacement [18]). Moreover, keeping the mouth in the open position for several minutes is uncomfortable for patients, but is required to avoid motion artifacts. An attempt at TMJ's motion MR imaging has also been made by pseudo-dynamic MRI that enables imaging of intermediate positions [18]. However, it was shown that acquired positions were different from those of natural jaw opening [19].

Real-time MRI is an emerging technology that is largely used for speech studies [20] and numerous medical applications [[21], [22], [23]]. Real-time MRI allows imaging of all intermediate dynamic jaw positions during natural movements. It could replace existing methods for TMJ's investigation by combining advantages of the existing methods: accessibility of motion information (similarly to kinematics-measuring devices), and precise position and general anatomical structure (similarly to conventional MRI). The quality and the efficacity of real-time MRI in oblique sagittal planes were already assessed on cohorts of healthy subjects [24] and patients [25]. It was shown that good-quality quasi-simultaneous multislice imaging was also possible [26]. These pioneers' works offered a real-time MRI of TMJ feasibility evidence and a demonstration of the advantages of the real-time technique over conventional static imaging. They also made an exhaustive qualitative description of condylar motion and introduced the possibility of biomechanical analysis of the TMJ's motion [27]. However, they mostly reported on the characterization of the condyles' motion based on visual observation or manual measurements.

The purpose of the current work was to present an acquisition protocol and a processing pipeline enabling automatizable quantitative evaluation of condylar motion from real-time MRI in the axial plane (which allows strictly simultaneous imaging of both condyles). Oppositely to existing works, we are focusing not on qualitative motion description, but on designing quantitative parameters aiming to provide a clinical physician with quick and comprehensible information describing mandibula's motion symmetricity. Such complement to acquired real-time MRI films would help with patient follow-up and estimation of rehabilitation efficacity based on objective criteria and without the employment of additional motion-tracking devices. This work also reports on the quality evaluation of each step of the proposed pipeline: actual slice orientation for the cohort of volunteers, segmentation, mid-sagittal axis determination, and robustness of the selected parameters. And, finally, we demonstrate the motion parameters variability and their correlation with symmetricity scores based on visual observation.