Automatic 3-Dimensional Cephalometric Landmarking via Deep Learning.

The increasing use of 3-dimensional (3D) imaging by orthodontists and maxillofacial surgeons to assess complex dentofacial deformities and plan orthognathic surgeries implies a critical need for 3D cephalometric analysis. Although promising methods were suggested to localize 3D landmarks automatically, concerns about robustness and generalizability restrain their clinical use. Consequently, highly trained operators remain needed to perform manual landmarking. In this retrospective diagnostic study, we aimed to train and evaluate a deep learning (DL) pipeline based on SpatialConfiguration-Net for automatic localization of 3D cephalometric landmarks on computed tomography (CT) scans. A retrospective sample of consecutive presurgical CT scans was randomly distributed between a training/validation set (n = 160) and a test set (n = 38). The reference data consisted of 33 landmarks, manually localized once by 1 operator(n = 178) or twice by 3 operators (n = 20, test set only). After inference on the test set, 1 CT scan showed "very low" confidence level predictions; we excluded it from the overall analysis but still assessed and discussed the corresponding results. The model performance was evaluated by comparing the predictions with the reference data; the outcome set included localization accuracy, cephalometric measurements, and comparison to manual landmarking reproducibility. On the hold-out test set, the mean localization error was 1.0 ± 1.3 mm, while success detection rates for 2.0, 2.5, and 3.0 mm were 90.4%, 93.6%, and 95.4%, respectively. Mean errors were -0.3 ± 1.3° and -0.1 ± 0.7 mm for angular and linear measurements, respectively. When compared to manual reproducibility, the measurements were within the Bland-Altman 95% limits of agreement for 91.9% and 71.8% of skeletal and dentoalveolar variables, respectively. To conclude, while our DL method still requires improvement, it provided highly accurate 3D landmark localization on a challenging test set, with a reliability for skeletal evaluation on par with what clinicians obtain.

Interaction between posture and maxillomandibular deformity: a systematic review.

Maxillomandibular deformity (MMD) and body posture appear to be correlated. However, no systematic literature review of the available evidence to support this correlation has been performed to date. The aim of this study was to conduct a systematic literature review on posture and MMD. This systematic literature review was registered in the PROSPERO database. Systematic searches of the MEDLINE, Scopus, Cochrane Library, and Web of Science databases were performed. In total, 13 clinical studies were included. Nine found a significant association between MMD and body posture or body balance: two studies showed a correlation between increased cervical lordosis and skeletal class III MMD, two studies showed an interaction between mandibular deviation and scoliosis, four studies demonstrated a significant association between lumbar column and pelvis anatomy and MMD, and one study found a correlation between displacement of the centre of mass and MMD. However, the level of evidence is low; the methods used to evaluate body posture and MMD were inconsistent. Orthognathic surgery could modify body posture. Although there seems to be an interaction between body posture and facial deformity, the number of studies is too small and the level of evidence too low to strongly support this association.

Accuracy and reliability of automatic three-dimensional cephalometric landmarking.

The aim of this systematic review was to assess the accuracy and reliability of automatic landmarking for cephalometric analysis of three-dimensional craniofacial images. We searched for studies that reported results of automatic landmarking and/or measurements of human head computed tomography or cone beam computed tomography scans in MEDLINE, Embase and Web of Science until March 2019. Two authors independently screened articles for eligibility. Risk of bias and applicability concerns for each included study were assessed using the QUADAS-2 tool. Eleven studies with test dataset sample sizes ranging from 18 to 77 images were included. They used knowledge-, atlas- or learning-based algorithms to landmark two to 33 points of cephalometric interest. Ten studies measured mean localization errors between manually and automatically detected landmarks. Depending on the studies and the landmarks, mean errors ranged from <0.50mm to>5mm. The two best-performing algorithms used a deep learning method and reported mean errors <2mm for every landmark, approximating results of operator variability in manual landmarking. Risk of bias regarding patient selection and implementation of the reference standard were found, therefore the studies might have yielded overoptimistic results. The robustness of these algorithms needs to be more thoroughly tested in challenging clinical settings. PROSPERO registration number: CRD42019119637.

Effect of shoulder model complexity in upper-body kinematics analysis of the golf swing.

The golf swing is a complex full body movement during which the spine and shoulders are highly involved. In order to determine shoulder kinematics during this movement, multibody kinematics optimization (MKO) can be recommended to limit the effect of the soft tissue artifact and to avoid joint dislocations or bone penetration in reconstructed kinematics. Classically, in golf biomechanics research, the shoulder is represented by a 3 degrees-of-freedom model representing the glenohumeral joint. More complex and physiological models are already provided in the scientific literature. Particularly, the model used in this study was a full body model and also described motions of clavicles and scapulae. This study aimed at quantifying the effect of utilizing a more complex and physiological shoulder model when studying the golf swing. Results obtained on 20 golfers showed that a more complex and physiologically-accurate model can more efficiently track experimental markers, which resulted in differences in joint kinematics. Hence, the model with 3 degrees-of-freedom between the humerus and the thorax may be inadequate when combined with MKO and a more physiological model would be beneficial. Finally, results would also be improved through a subject-specific approach for the determination of the segment lengths.

Limiting interpedicular screw displacement increases shear forces in screws: A finite element study.

BACKGROUND CONTEXT: Screw loosening has been reported for non-fusion devices. Forces on pedicle screws could be related to kinematic parameters as the interpedicular displacement (ID), which consists of the displacement between superior and inferior screw heads from full extension to full flexion. PURPOSE: To investigate the relationship between ID and screw loosening for different designs of posterior implants using a finite element model. METHODS: An L3-sacrum previously validated spine FE model was used. Three-rod designs were considered in L4-L5 segment: a rigid screw-rod implant, a flexible one and a specific design with a sliding rod providing limited restrain in ID. In order to simulate intermediate configurations, the friction coefficient between the sliding rods and connectors were varied. The sacrum was rigidly fixed. Rotations (flexion-extension, lateral bending and axial rotation) were applied to L3, for each modeled configuration: intact, injured, injured with different implants. Model consistency was checked with existing experimental in vitro data on intact and instrumented segments. Screw loads were computed as well as ID. RESULTS: In flexion-extension, the ID was less than 2mm for rigid (R) and flexible (F) constructs and 5.5mm for intact spine and the sliding implant (S3). Screw's shear forces were 272N, 153N, 43N respectively for R, F and S3 constructs. CONCLUSIONS: Implants that allow ID presented lower screws loads. A compromise between the ability of the implant to withstand compressive forces, which requires longitudinal stiffness, and its ability to allow ID could be important for future implant designs in order to prevent screw loosening.

Zygomatic bone shape in intentional cranial deformations: a model for the study of the interactions between skull growth and facial morphology.

Intentional cranial deformations (ICD) were obtained by exerting external mechanical constraints on the skull vault during the first years of life to permanently modify head shape. The repercussions of ICD on the face are not well described in the midfacial region. Here we assessed the shape of the zygomatic bone in different types of ICDs. We considered 14 non-deformed skulls, 19 skulls with antero-posterior deformation, nine skulls with circumferential deformation and seven skulls with Toulouse deformation. The shape of the zygomatic bone was assessed using a statistical shape model after mesh registration. Euclidian distances between mean models and Mahalanobis distances after canonical variate analysis were computed. Classification accuracy was computed using a cross-validation approach. Different ICDs cause specific zygomatic shape modifications corresponding to different degrees of retrusion but the shape of the zygomatic bone alone is not a sufficient parameter for classifying populations into ICD groups defined by deformation types. We illustrate the fact that external mechanical constraints on the skull vault influence midfacial growth. ICDs are a model for the study of the influence of epigenetic factors on craniofacial growth and can help to understand the facial effects of congenital skull malformations such as single or multi-suture synostoses, or of external orthopedic devices such as helmets used to correct deformational plagiocephaly.

Muscle parameters estimation based on biplanar radiography.

The evaluation of muscle and joint forces in vivo is still a challenge. Musculo-Skeletal (musculo-skeletal) models are used to compute forces based on movement analysis. Most of them are built from a scaled-generic model based on cadaver measurements, which provides a low level of personalization, or from Magnetic Resonance Images, which provide a personalized model in lying position. This study proposed an original two steps method to access a subject-specific musculo-skeletal model in 30 min, which is based solely on biplanar X-Rays. First, the subject-specific 3D geometry of bones and skin envelopes were reconstructed from biplanar X-Rays radiography. Then, 2200 corresponding control points were identified between a reference model and the subject-specific X-Rays model. Finally, the shape of 21 lower limb muscles was estimated using a non-linear transformation between the control points in order to fit the muscle shape of the reference model to the X-Rays model. Twelfth musculo-skeletal models were reconstructed and compared to their reference. The muscle volume was not accurately estimated with a standard deviation (SD) ranging from 10 to 68%. However, this method provided an accurate estimation the muscle line of action with a SD of the length difference lower than 2% and a positioning error lower than 20 mm. The moment arm was also well estimated with SD lower than 15% for most muscle, which was significantly better than scaled-generic model for most muscle. This method open the way to a quick modeling method for gait analysis based on biplanar radiography.

A new method for the evaluation of the end-to-end distance of the knee ligaments and popliteal complex during passive knee flexion.

BACKGROUND: Accurate knowledge about the length variation of the knee ligaments (ACL, PCL, MCL and LCL) and the popliteal complex during knee flexion/extension is essential for modelling and clinical applications. The aim of the present study is to provide this information by using an original technique able to faithfully reproduce the continuous passive knee flexion-extension kinematics and to reliably identify each ligament/tendon attachment site. METHODS: Twelve lower limbs (femur, tibia, fibula, patella) were tested and set in motion (0-120°) using an ad hoc rig. Tibio-femoral kinematics was obtained using an optoelectronic system. A 3D digital model of each bone was obtained using low-dosage stereoradiography. Knee specimens were dissected and the insertion of each ligament and popliteal complex were marked with radio opaque paint. ACL, PCL and MCL were separated into two bundles. Bone epiphyses were CT-scanned to obtain a digital model of each ligament insertion. Bones and attachment site models were registered and the end-to-end distance variation of each ligament/tendon was computed over knee flexion. RESULTS: A tibial internal rotation of 18°±4° with respect to the femur was observed. The different bundles of the ACL, MCL and LCL shortened, whereas all bundles of the PCL lengthened. The popliteal complex was found to shorten until 30° of knee flexion and then to lengthen. CONCLUSION: The end-to-end distance variation of the knee ligaments and popliteal complex can be estimated during knee flexion using a robust and reliable method based on marking the ligaments/tendon insertions with radiopaque paint. LEVEL OF EVIDENCE: Level IV.

Influence of the overall stiffness of a load-bearing porous titanium implant on bone ingrowth in critical-size mandibular bone defects in sheep.

The aim of this work was to assess the influence of reduction of the apparent mechanical properties of fully load-bearing porous titanium implants used in mandibular bone defects. Segmental 18mm long bone defects were created bilaterally in the lower jaws of adult ewes. One group of 6 ewes (group A) was treated with load-bearing 'rigid' (high stiffness) porous implants on the right side, and with control on the left side. A second group of 6 ewes (group B) was treated with 'flexible' porous and control implants exhibiting apparent mechanical properties ten times lower than the rigid implants. The mechanical behavior of the reconstructed hemi-mandibles was assessed by cantilever testing and bone ingrowth into the segmental defects was assessed by BV/TV measurement within the implant using micro-CT 12 weeks after implantation. A significantly higher rigidity was identified for porous implants compared with control implants at the anterior interface in group B. BV/TV of porous implants was significantly higher than that of control implants in group A. BV/TV differences were significant between porous and control implants in group B and were homogeneous along the main axis. Significantly higher BV/TV was identified in most sub-volumes of group B porous implants compared with group A. This work highlights the critical importance of the tuning of scaffolds to promote bone ingrowth with reference to the local strains occurring within the porous scaffold, which in this application was achieved using fully load-bearing low-stiffness porous titanium implants.

Diffusion model to describe osteogenesis within a porous titanium scaffold.

In this study, we develop a two-dimensional finite element model, which is derived from an animal experiment and allows simulating osteogenesis within a porous titanium scaffold implanted in ewe's hemi-mandible during 12 weeks. The cell activity is described through diffusion equations and regulated by the stress state of the structure. We compare our model to (i) histological observations and (ii) experimental data obtained from a mechanical test done on sacrificed animal. We show that our mechano-biological approach provides consistent numerical results and constitutes a useful tool to predict osteogenesis pattern.

Reliable femoral frame construction based on MRI dedicated to muscles position follow-up.

In vivo follow-up of muscle shape variation represents a challenge when evaluating muscle development due to disease or treatment. Recent developments in muscles reconstruction techniques indicate MRI as a clinical tool for the follow-up of the thigh muscles. The comparison of 3D muscles shape from two different sequences is not easy because there is no common frame. This study proposes an innovative method for the reconstruction of a reliable femoral frame based on the femoral head and both condyles centers. In order to robustify the definition of condylar spheres, an original method was developed to combine the estimation of diameters of both condyles from the lateral antero-posterior distance and the estimation of the spheres center from an optimization process. The influence of spacing between MR slices and of origin positions was studied. For all axes, the proposed method presented an angular error lower than 1° with spacing between slice of 10 mm and the optimal position of the origin was identified at 56 % of the distance between the femoral head center and the barycenter of both condyles. The high reliability of this method provides a robust frame for clinical follow-up based on MRI .

The effect of posterior non-fusion instrumentation on segmental shear loading of the lumbar spine.

BACKGROUND: Lumbar stenosis and facet osteoarthritis represent indications for decompression and instrumentation. It is unclear if degenerative spondylolisthesis grade I with a remaining disc height could be an indication for non-fusion instrumentation. The purpose of this study was to determine the influence of a mobile pedicle screw based device on lumbar segmental shear loading, thus simulating the condition of spondylolisthesis. MATERIALS AND METHODS: Six human cadaver specimens were tested in 3 configurations: intact L4-L5 segment, then facetectomy plus undercutting laminectomy, then instrumentation with lesion. A static axial compression of 400 N was applied to the lumbar segment and anterior displacements of L4 on L5 were measured for posterior-anterior shear forces from 0 to 200 N. The slope of the loading curve was assessed to determine shear stiffness. RESULTS: Homogenous load-displacement curves were obtained for all specimens. The average intact anterior displacement was 1.2 mm. After lesion, the displacement increased by 0.6mm compared to intact (P=0.032). The instrumentation decreased the displacement by 0.5 mm compared to lesion (P=0.046). The stiffness's were: 162 N/mm for intact, 106 N/mm for lesion, 148 N/mm for instrumentation. The difference was not significant between instrumented and intact segments (P=0.591). CONCLUSIONS: Facetectomy plus undercutting laminectomy decreases segmental shear stiffness and increases anterior translational L4-L5 displacement. Shear stiffness of the instrumented segment is higher with the device and anterior displacements under shear loading are similar to the intact spine. This condition could theoretically be interesting for the simulation of non-fusion instrumentation in degenerative spondylolisthesis.