Automatic grading of patients with a unilateral facial paralysis based on the Sunnybrook Facial Grading System - A deep learning study based on a convolutional neural network.

PURPOSE: In order to assess the severity and the progression of a unilateral peripheral facial palsy the Sunnybrook Facial Grading System (SFGS) is a well-established grading system due to its clinical relevance, sensitivity, and robust measuring method. However, training is required in order to achieve a high inter-rater reliability. This study investigated the automated grading of facial palsy patients based on the SFGS using a convolutional neural network. METHODS: A total of 116 patients with a unilateral peripheral facial palsy and 9 healthy subjects were recorded performing the Sunnybrook poses. A separate model was trained for each of the 13 elements of the SFGS and then used to calculate the Sunnybrook subscores and composite score. The performance of the automated grading system was compared to three clinicians experienced in the grading of a facial palsy. RESULTS: The inter-rater reliability of the convolutional neural network was within the range of human observers, with an average intra-class correlation coefficient of 0.87 for the composite Sunnybrook score, 0.45 for the resting symmetry subscore, 0.89 for the symmetry of voluntary movement subscore, and 0.77 for the synkinesis subscore. CONCLUSIONS: This study showed the potential of the automated SFGS to be implemented in a clinical setting. The automated grading system adhered to the original SFGS, which makes the implementation and interpretation of the automated grading more straightforward. The automated system can be implemented in numerous settings such as online consults in an e-Health environment, since the model used 2D images captured from a video recording.

A Novel and Practical Protocol for Three-Dimensional Assessment of Alveolar Cleft Grafting Procedures.

OBJECTIVE: To evaluate the reproducibility and accuracy of a new, easy-to-use volumetric assessment of the alveolar cleft. DESIGN: Twelve cone-beam computed tomography (CBCT) datasets of patients with a unilateral cleft lip, alveolus, and palate were evaluated by two investigators. Residual alveolar cleft calcified volume one year after surgery was analyzed by using standardized landmarks to determine the borders of the cleft defect and semi-automatically segment the alveolar cleft defect. RESULTS: The Dice-coefficient between observers for the segmented preoperative alveolar cleft defect was 0.81. Average percentage of residual alveolar cleft calcified material was 66.7% one year postoperatively. CONCLUSIONS: This study demonstrates a reliable and practical semi-automatic three-dimensional volumetric assessment method for unilateral clefts using CBCT.

Creating Three-Dimensional Templates of Smiling and Pouting Faces for Different Sex- and Age Groups.

Smile appearance has a major psychological impact. Orthognathic surgery, which has harmonizing results on skeletal structures, can negatively influence the smile appearance due to soft tissue effects. To enhance the aesthetic effects of orthognathic surgery on soft tissues, reference models for large parts of the hospital's adherent area are called for. This study aims to create average facial models for different sex and age groups in two facial exercises: maximum closed smile, and pouting. These models were created using coherent point drift and Procrustes algorithms in MATLAB. Principal component analysis was performed, and of 20 surgical landmarks, the in-group variation using standard deviation was calculated. Three distances were analyzed: nasal width, philtral width, and mouth width. To correct for facial size, these distances were analyzed as a ratio of intercanthal width. In total, 328 healthy subjects were included in the study. Subjects were grouped by sex, and in age categories spanning four years each, with an adult group with all ages >16 years. For both smiling and pouting faces, all ratios increased with ageing. These templates and data can benefit facial surgeons, to obtain an expected outcome according to the age of the patient.

Cochlear implant positioning and fixation using 3D-printed patient specific surgical guides; a cadaveric study.

HYPOTHESIS: To develop and validate the optimal design and evaluate accuracy of individualized 3D- printed surgical guides for cochlear implantation. BACKGROUND: Positioning and fixation of the cochlear implant (CI) are commonly performed free hand. Applications of 3-dimensional (3D) technology now allow us to make patient specific, bone supported surgical guides, to aid CI surgeons with precise placement and drilling out the bony well which accommodates the receiver/stimulator device of the CI. METHODS: Cone beam CT (CBCT) scans were acquired from temporal bones in 9 cadaveric heads (18 ears), followed by virtual planning of the CI position. Surgical, bone-supported drilling guides were designed to conduct a minimally invasive procedure and were 3D-printed. Fixation screws were used to keep the guide in place in predetermined bone areas. Specimens were implanted with 3 different CI models. After implantation, CBCT scans of the implanted specimens were performed. Accuracy of CI placement was assessed by comparing the 3D models of the planned and implanted CI's by calculating the translational and rotational deviations. RESULTS: Median translational deviations of placement in the X- and Y-axis were within the predetermined clinically relevant deviation range (< 3 mm per axis); median translational deviation in the Z-axis was 3.41 mm. Median rotational deviations of placement for X-, Y- and Z-rotation were 5.50°, 4.58° and 3.71°, respectively. CONCLUSION: This study resulted in the first 3D-printed, patient- and CI- model specific surgical guide for positioning during cochlear implantation. The next step for the development and evaluation of this surgical guide will be to evaluate the method in clinical practice.

Depth accuracy of the RealSense F200: Low-cost 4D facial imaging.

The RealSense F200 represents a new generation of economically viable 4-dimensional imaging (4D) systems for home use. However, its 3D geometric (depth) accuracy has not been clinically tested. Therefore, this study determined the depth accuracy of the RealSense, in a cohort of patients with a unilateral facial palsy (n = 34), by using the clinically validated 3dMD system as a gold standard. The patients were simultaneously recorded with both systems, capturing six Sunnybrook poses. This study has shown that the RealSense depth accuracy was not affected by a facial palsy (1.48 ± 0.28 mm), compared to a healthy face (1.46 ± 0.26 mm). Furthermore, the Sunnybrook poses did not influence the RealSense depth accuracy (p = 0.76). However, the distance of the patients to the RealSense was shown to affect the accuracy of the system, where the highest depth accuracy of 1.07 mm was measured at a distance of 35 cm. Overall, this study has shown that the RealSense can provide reliable and accurate depth data when recording a range of facial movements. Therefore, when the portability, low-costs, and availability of the RealSense are taken into consideration, the camera is a viable option for 4D close range imaging in telehealth.