Search Results (23)

This paper proposes emotion recognition methods for consecutive facial images and implements the inference of a neural network model on a field-programmable gate array (FPGA) for real-time sensing of human motion. The proposed emotion recognition methods are based on a neural network architecture called Convolutional Long Short-Term Memory Fully Connected Deep Neural Network (CLDNN), which combines convolutional neural networks (CNNs) for spatial feature extraction, long short-term memory (LSTM) for temporal modeling, and fully connected neural networks (FCNNs) for final classification. This architecture can analyze the local feature sequences obtained through convolution of data, making it suitable for processing time-series data such as consecutive facial images. The method achieves an average recognition rate of 99.51% on the RAVDESS database, 87.80% on the BAUM-1s database and 96.82% on the eNTERFACE’05 database, using 10-fold cross-validation on a personal computer (PC). The comparisons in this paper show that our methods outperform existing related works in recognition accuracy. The same model is implemented on an FPGA chip, where it achieves identical accuracy to that on a PC, confirming both its effectiveness and hardware compatibility. Full article

Objectives: Facial soft tissue imaging is crucial in orthodontic treatment planning, and the structured light scanning technology found in the latest iPhone models constitutes a promising method. Currently, studies which evaluate the accuracy of smartphone-based three-dimensional (3D) facial scanners are scarce. This study compares smartphone scans with cone-beam computed tomography (CBCT) images. Materials and Methods: Three-dimensional images of 23 screened patients were captured with the camera of an iPhone 13 Pro Max and processed with the Scandy Pro application; CBCT scans were also taken as a standard of care. After establishing unique image pairs of the same patient, linear and angular measurements were compared between the images to assess the scanner’s two-dimensional trueness. Following the co-registration of the virtual models, a heat map was generated, and root mean square (RMS) deviations were calculated for quantitative assessment of 3D trueness. Precision was determined by comparing consecutive 3D facial scans of five participants, while intraobserver reliability was assessed by repeating measurements on five subjects after a two-week interval. Results: This study found no significant difference in soft tissue measurements between smartphone and CBCT images (p > 0.05). The mean absolute difference was 1.43 mm for the linear and 3.16° for the angular measurements. The mean RMS value was 1.47 mm. Intraobserver reliability and scanner precision were assessed, and the Intraclass Correlation Coefficients were found to be excellent. Conclusions: Smartphone facial scanners offer an accurate and reliable alternative to stereophotogrammetry systems, though clinicians should exercise caution when examining the lateral sections of those images due to inherent inaccuracies. Full article

Background/Objectives: Quantitative assessments of facial muscle function and cognitive responses can enhance the clinic evaluations in neuromuscular disorders such as Bell’s palsy and psychiatric conditions including anxiety and depression. This study explored the application of Digital Image Speckle Correlation (DISC) in detecting enervation of facial musculature and assessing reaction times in response to visual stimuli. Methods: A consistent video recording setup was used to capture facial movements of human subjects in response to visual stimuli from a calibrated database. The DISC method utilizes the displacement of naturally occurring skin pores to map the specific locus of underlying muscular movement. The technique was applied to two distinct case studies: Patient 1 had unilateral Bell’s palsy and was monitored for 1 month of recovery. Patient 2 had a comorbidity of refractory depression and anxiety disorders with ketamine treatment and was assessed over 3 consecutive weekly visits. For patient 1, facial asymmetry was calculated by comparing left-to-right displacement signals. For patient 2, visual reaction time was measured, and facial motion intensity and response rate were compared with self-reported depression and anxiety scales. Results: DISC effectively mapped biomechanical properties of facial motions, providing detailed spatial and temporal resolution of muscle activity. In a control cohort of 10 subjects, when executing a facial expression, the degree of left/right facial asymmetry was determined to be 13.2 (8)%. And showed a robust response in an average of 275 (81) milliseconds to five out of the five images shown. For patient 1, obtained an initial asymmetry of nearly 100%, which decreased steadily to 20% in one month, demonstrating a progressive recovery. Patient 2 exhibited a prolonged reaction time of 518 (93) milliseconds and reduced response rates compared with controls of 275 (81) milliseconds and a decrease in the overall rate of response relative to the control group. The data obtained before treatment in three visits correlated strongly with selected depression and anxiety scores. Conclusions: These findings highlight the utility of DISC in enhancing clinical monitoring, complementing traditional examinations and self-reported measures. Full article

Drowsiness has been recognized as a leading factor in road accidents worldwide. Despite considerable research in this area, this paper aims to improve the precision of drowsiness detection specifically for long-haul travel by employing the Dlib-based facial feature detection algorithm. This study proposes two algorithms: a static and adaptive frame threshold. Both approaches utilize eye closure ratio (ECR) and mouth aperture ratio (MAR) parameters to determine the driver’s level of drowsiness. The static threshold method issues a warning when the ECR and/or MAR values reach specific thresholds. In this method, the ECR threshold is established at 0.15 and the MAR threshold at 0.4. The static threshold method demonstrated an accuracy of 89.4% and a sensitivity of 96.5% using 1000 images. The adaptive frame threshold algorithm uses a counter to monitor the number of consecutive frames that meet the drowsiness criteria before triggering a warning. Additionally, the number of consecutive frames required is adjusted dynamically over time to enhance detection accuracy and more accurately indicate a state of drowsiness. The adaptive frame threshold algorithm was tested using four 30 min videos, from a publicly available dataset achieving a maximum accuracy of 98.2% and a sensitivity of 64.3% with 500 images. Full article

Background: The techniques of choice used in the treatment of extensive vascular lesions of the face are methods based on high-energy light sources, such as lasers and IPL (intense pulsed light). The techniques commonly employed to detect blood vessel abnormalities in skin primarily rely on semi-quantitative or qualitative scales. Methods: The study was conducted on a group of 38 volunteers; a series of three treatments was performed using an IPL source (Lumecca, Inmode, Israel). The vessels’ response to the high-energy light was verified using the following quantitative methods: mexametric measurements, hyperspectral imaging, and directional reflectance measurements. Results: In the mexameter measurement, statistically non-significant differences were obtained in the hemoglobin content in the skin condition prior to and following multiple intense pulsed light sessions. Studies performed using a hyperspectral camera showed that at a wavelength of 580 nm, the increase in skin reflectance after the treatment was statistically significant. Total directional reflectance measurements showed that at wavelengths of 400–540 nm and 480–600 nm, following the IPL treatment, the skin reflectance increased statistically significantly. Implementing three consecutive intense pulsed light procedures appears adequate to obtain a clinically satisfactory reduction in vascular changes in the facial skin. Mexametric measurements do not enable the assessment of the reaction of blood vessels to IPL. Conclusions: Hyperspectral imaging is an effective method for the quantitative assessment of skin vascular lesions. The best results in the assessment of vascular lesions using hyperspectral imaging are obtained at wavelengths of 420 nm and 580 nm. The hemispheric directional reflectance method allows for a quick, accurate, and repeatable assessment of vascular skin changes. Full article

Background and Objectives: Microsurgical resection with intraoperative neuromonitoring is the gold standard for acoustic neurinomas (ANs) which are classified as T3 or T4 tumors according to the Hannover Classification. Microscope-based augmented reality (AR) can be beneficial in cerebellopontine angle and lateral skull base surgery, since these are small areas packed with anatomical structures and the use of this technology enables automatic 3D building of a model without the need for a surgeon to mentally perform this task of transferring 2D images seen on the microscope into imaginary 3D images, which then reduces the possibility of error and provides better orientation in the operative field. Materials and Methods: All patients who underwent surgery for resection of ANs in our department were included in this study. Clinical outcomes in terms of postoperative neurological deficits and complications were evaluated, as well as neuroradiological outcomes for tumor remnants and recurrence. Results: A total of 43 consecutive patients (25 female, median age 60.5 ± 16 years) who underwent resection of ANs via retrosigmoid osteoclastic craniotomy with the use of intraoperative neuromonitoring (22 right-sided, 14 giant tumors, 10 cystic, 7 with hydrocephalus) by a single surgeon were included in this study, with a median follow up of 41.2 ± 32.2 months. A total of 18 patients underwent subtotal resection, 1 patient partial resection and 24 patients gross total resection. A total of 27 patients underwent resection in sitting position and the rest in semi-sitting position. Out of 37 patients who had no facial nerve deficit prior to surgery, 19 patients were intact following surgery, 7 patients had House Brackmann (HB) Grade II paresis, 3 patients HB III, 7 patients HB IV and 1 patient HB V. Wound healing deficit with cerebrospinal fluid (CSF) leak occurred in 8 patients (18.6%). Operative time was 317.3 ± 99 min. One patient which had recurrence and one further patient with partial resection underwent radiotherapy following surgery. A total of 16 patients (37.2%) underwent resection using fiducial-based navigation and microscope-based AR, all in sitting position. Segmented objects of interest in AR were the sigmoid and transverse sinus, tumor outline, cranial nerves (CN) VII, VIII and V, petrous vein, cochlea and semicircular canals and brain stem. Operative time and clinical outcome did not differ between the AR and the non-AR group. However, use of AR improved orientation in the operative field for craniotomy planning and microsurgical resection by identification of important neurovascular structures. Conclusions: The single-center experience of resection of ANs showed a high rate of gross total (GTR) and subtotal resection (STR) with low recurrence. Use of AR improves intraoperative orientation and facilitates craniotomy planning and AN resection through early improved identification of important anatomical relations to structures of the inner auditory canal, venous sinuses, petrous vein, brain stem and the course of cranial nerves. Full article

Image inpainting and Deepfake techniques have the potential to drastically alter the meaning of visual content, posing a serious threat to the integrity of both images and videos. Addressing this challenge requires the development of effective methods to verify the authenticity of investigated visual data. This research introduces UFCC (Unified Forensic Scheme by Content Consistency), a novel forensic approach based on deep learning. UFCC can identify tampered areas in images and detect Deepfake videos by examining content consistency, assuming that manipulations can create dissimilarity between tampered and intact portions of visual data. The term “Unified” signifies that the same methodology is applicable to both still images and videos. Recognizing the challenge of collecting a diverse dataset for supervised learning due to various tampering methods, we overcome this limitation by incorporating information from original or unaltered content in the training process rather than relying solely on tampered data. A neural network for feature extraction is trained to classify imagery patches, and a Siamese network measures the similarity between pairs of patches. For still images, tampered areas are identified as patches that deviate from the majority of the investigated image. In the case of Deepfake video detection, the proposed scheme involves locating facial regions and determining authenticity by comparing facial region similarity across consecutive frames. Extensive testing is conducted on publicly available image forensic datasets and Deepfake datasets with various manipulation operations. The experimental results highlight the superior accuracy and stability of the UFCC scheme compared to existing methods. Full article

This study delves into the acoustic environment within dental clinics, particularly focusing on the impact of extraoral suction devices employed for infection control amid the COVID-19 pandemic. The research encompasses a comprehensive investigation, including a questionnaire survey of dental professionals, sound level measurements at suction device openings, acoustic imaging, and a detailed analysis of sound levels and their spectral characteristics during dental procedures. Additionally, ambient sound levels within clinical settings were monitored over two consecutive days. The findings reveal notable observations. Dental professionals expressed concerns regarding increased sound levels and associated distress caused by extraoral suction device operation. Objective measurements identified varying A-weighted sound pressure levels ranging from 86.0 dB to 96.7 dB at suction device openings, highlighting elevated sound pressure levels and a wide frequency range, especially in the vicinity of both the dentist and the patient’s facial area during dental aerosol procedures. On the other hand, for the entire clinical room, the equivalent continuous A-weighted sound pressure level during the consultation hours was not considered problematic. In light of these findings, it becomes evident that there is a pressing necessity to refine the acoustic characteristics of extraoral suction devices to foster a more accommodating acoustic environment for both patients and dental healthcare professionals within dental clinics. Full article

Introduction: Obstructive sleep apnea (OSA) is a disorder in which ventilation becomes disrupted due to a complete or partial upper airway obstruction Altered craniofacial morphology is one of the most important anatomical factors associated with obstructive sleep apnea (OSA). Studies have assessed craniofacial features in the non-syndromic pediatric population. The aim of this study was to analyze the orthodontic and facial characteristic of craniofacial syndromic children referred for polysomnography (PSG) and to assess the correlation with the apnea–hypopnea index (AHI). Methods: In the current cross-sectional study, consecutive syndromic patients referred for PSG were invited to participate. A systematic clinical examination including extra- and intra-oral orthodontic examination was performed by calibrated orthodontists. Standardized frontal and profile photographs with reference points were taken and analyzed using ImageJ^® software to study the craniofacial morphology. PSG data were analyzed for correlation with craniofacial features. STROBE guidelines were strictly adopted during the research presentation. Results: The sample included 52 syndromic patients (50% females, mean age 9.38 ± 3.36 years) diagnosed with 17 different syndromes, of which 24 patients had craniofacial photography analysis carried out. Most of the sample (40%) had severe OSA, while only 5.8% had no OSA. Down’s syndrome (DS) was the most common syndrome (40%) followed by Goldenhar syndrome (5%), Pierre Robin Sequence (5%), and other syndromes. The severity of AHI was significantly correlated with decreased midfacial height. increased thyromental angle and cervicomental angle, decreased mandibular angle, and decreased upper facial height. All patients with DS were diagnosed with OSA (57% severe OSA), and their ODI was significantly correlated with increased intercanthal distance. Obesity was not correlated to the severity of AHI for syndromic patients. Conclusions: Decreased midfacial height and obtuse thyromental angle were correlated with increased AHI for syndromic patients. Increased intercanthal distance of DS patients could be a major predictor of OSA severity. Obesity does not seem to play a major role in the severity of OSA for syndromic patients. Further studies with larger samples are necessary to confirm these findings. Full article

Objective: The study investigated the preoperative anatomical variables that affect the outcome of surgical correction in patients with Class III facial asymmetry. Methods: The study recruited 37 consecutive patients with facial asymmetry who had 2-jaw orthognathic surgery. They were divided into two groups based on the surgical outcome: symmetrical (S group) or asymmetrical (A group), according to the asymmetry index. The CBCT images were obtained before surgery (T0) and after debond (T1). The 3D dentofacial measurements were compared between groups S and A by the Mann–Whitney test. Spearman correlation analysis was performed to examine the relationship of all dentoskeletal variables in T0 with the facial symmetry outcome. Results: Significant between-group differences were observed in preoperative time, including maxillary anterior occlusal canting, maxillary posterior occlusal canting, the gonion–Frankfort horizontal plane (FHP) distance, the mandibular ramus axis–FHP distances, and sagittal and transverse of condyle position. Conclusions: For patients with severe skeletal Class III asymmetry, preoperative anatomical variables, particularly preoperative “roll” and “yaw” discrepancies and anatomical limitations of the mandible, should be considered for favorable asymmetry correction. Based on the anatomic variables that affect the outcomes of facial asymmetry correction, prognosis and treatment limitation could be predicted before treatment. Full article

To explore the influence of campus public space characteristics on students’ emotions, we investigated the association mechanism between public space characteristics and students’ emotions concerning the distribution of students’ emotions in public spaces. The present study used photographs of facial expressions taken over two consecutive weeks as a source of data regarding the students’ affective reactions. The collected facial expression images were analyzed using facial expression recognition. Values were assigned to the expression data, combined with geographic coordinates to create an emotion map of the campus public space using GIS software. Then, spatial feature data via emotion marker points were collected. We used smart wearable devices to combine the ECG data with spatial characteristics and took SDNN and RMSSD as ECG indicators to assess mood changes. We analyzed the correlation between these spatial features and heart rate variability and developed regression models for the ECG data. The findings show that sky visibility, space D/H, green visibility, skyline change index, and boundary permeability can engage students’ positive emotions in a meaningful way. On the other hand, paving visibility and road linearity tends to induce negative emotions in students’ minds. Full article

The inspection of patients’ soft tissues and the effects of various dental procedures on their facial physiognomy are quite challenging. To minimise discomfort and simplify the process of manual measuring, we performed facial scanning and computer measurement of experimentally determined demarcation lines. Images were acquired using a low-cost 3D scanner. Two consecutive scans were obtained from 39 participants, to test the scanner repeatability. An additional ten persons were scanned before and after forward movement of the mandible (predicted treatment outcome). Sensor technology that combines red, green, and blue (RGB) data with depth information (RGBD) integration was used for merging frames into a 3D object. For proper comparison, the resulting images were registered together, which was performed with ICP (Iterative Closest Point)-based techniques. Measurements on 3D images were performed using the exact distance algorithm. One operator measured the same demarcation lines directly on participants; repeatability was tested (intra-class correlations). The results showed that the 3D face scans were reproducible with high accuracy (mean difference between repeated scans <1%); the actual measurements were repeatable to some extent (excellent only for the tragus-pogonion demarcation line); computational measurements were accurate, repeatable, and comparable to the actual measurements. Three dimensional (3D) facial scans can be used as a faster, more comfortable for patients, and more accurate technique to detect and quantify changes in facial soft tissue resulting from various dental procedures. Full article

Unlike masking human faces from images, facial masking in real-time, frame by frame from a video stream, presents technical challenges related to various factors such as camera-to-human distance, head direction, and mosaic schemes. In many existing studies, expensive equipment and huge computational resources are strongly required, and it is not easy to effectively realize real-time facial masking with a simpler approach. This study aims to develop a secure streaming system to support remote medicine education and to quantitatively evaluate consecutive and effective facial masking using image-based bone sensing. Our key idea is to use the facial feature of bone sensing instead of general face recognition techniques to perform facial masking from the video stream. We use a general-purpose computer and a USB fixed-point camera to implement the eye line mosaic and face mosaic. We quantitatively evaluate the results of facial masking at different distances and human head orientations using bone sensing technology and a depth camera. we compare the results of a similar approach for face recognition with those of bone sensing. As the main results, consecutive face masking using bone sensing is unaffected by distance and head orientation, and the variation width of the mosaic area is stable within around 30% of the target area. However, about three-fourths of the results using conventional face recognition were unable to mask their faces consecutively. Full article

The reproducibility of facial expressions has been previously explored, however, there is no detailed information regarding the reproducibility of lip morphology forming a social smile. In this study, we recruited 93 young adults, aged 21–35 years old, who agreed to participate in two consecutive study visits four weeks apart. On each visit, they were asked to perform a social smile, which was captured on a 3D facial image acquired using the 3dMD camera system. Assessments of self-perceived smile attractiveness were also performed using a VAS scale. Lip morphology, including smile shape, was described using 62 landmarks and semi-landmarks. A Procrustes superimposition of each set of smiling configurations (first and second visit) was performed and the Euclidean distance between each landmark set was calculated. A linear regression model was used to test the association between smile consistency and self-perceived smile attractiveness. The results show that the average landmark distance between sessions did not exceed 1.5 mm, indicating high repeatability, and that females presented approximately 15% higher smile consistecy than males (p < 0.05). There was no statistically significant association between smile consistency and self-perceived smile attractiveness (η² = 0.015; p = 0.252), when controlling for the effect of sex and age. Full article

Whether a one-jaw or two-jaw design is used in orthognathic surgery for patients with cleft remains varied and controversial. This study aimed to compare the two approaches using 3-dimensional imaging surgical simulation. This study was conducted on 41 consecutive patients with complete unilateral cleft lip and palate treated in the craniofacial center. All patients had original two-jaw virtual planning and orthognathic surgery. Simulation of one-jaw LeFort I surgery was performed using the same final dental occlusion on 3-dimensional images. Cephalometric data and asymmetry index were collected and compared among the designs. Average advancement of the maxilla was 7.46 mm in one-jaw and 4.80 mm in two-jaw design. SNA, SNB, and A-N perpendicular were significantly greater and close to normative data in the one-jaw design. ANB angles were similar in both designs. The anterior and posterior occlusal plane cants, the deviation of midline landmarks, and the asymmetry index were more improved in the two-jaw approach. It is concluded that the two-jaw cleft orthognathic surgery could significantly improve facial midline and symmetry compared with the one-jaw approach. However, the two-jaw surgery with mandibular setback produced less protrusive facial contour although a harmonious relationship between the maxilla and mandible was achieved. Full article