Measurement plane of the cross-sectional area of the masseter muscle in patients with skeletal Class III malocclusion: An artificial intelligence model.

INTRODUCTION: This study aimed to determine a measurement plane that could represent the maximum cross-sectional area (MCSA) of masseter muscle using an artificial intelligence model for patients with skeletal Class III malocclusion. METHODS: The study included 197 patients, divided into subgroups according to sex, mandibular symmetry, and mandibular plane angle. The volume, MCSA, and the cross-sectional area (CSA) at different levels were calculated automatically. The vertical distance between MCSA and mandibular foramen, along with the ratio of the masseter CSA at different levels to the MCSA (R), were also calculated. RESULTS: The MCSA and volume showed a strong correlation in the total sample and each subgroup (P <0.001). The correlation between the CSA at each level and MCSA was statistically significant (P <0.001). The peak of the r and the correlation coefficient between the CSA at different levels and MCSA were mostly present 5-10 mm above the mandibular foramen for the total sample and the subgroups. The mean of RA5 to RA10 was >0.93, whereas the corresponding correlation coefficient was >0.96, both for the entire sample and for the subgroups. CONCLUSIONS: MCSA could be used as an indicator for masseter muscle size. For patients with skeletal Class III malocclusion, the CSA 5-10 mm above the mandibular foramen, parallel to the Frankfort plane, could be used to estimate the masseter muscle MCSA.

Changes in masseter muscle morphology after surgical-orthodontic treatment in patients with skeletal Class III malocclusion with mandibular asymmetry: The automatic masseter muscle segmentation model.

INTRODUCTION: This study evaluated the masseter muscle changes after surgical-orthodontic treatment in patients with a skeletal Class III malocclusion using automatic segmentation. METHODS: Images of 120 patients with skeletal Class III malocclusion were obtained and reconstructed at T0 (pretreatment), T1 (presurgery), and T2 (6-12-month postsurgery). The patients were divided into symmetrical and asymmetrical groups. The volume, major axis length, maximum cross-sectional area, horizontal cross-sectional area 5 mm above the mandibular foramen (CSAF), and orientation were calculated automatically. RESULTS: In the asymmetrical group, the volume and major axis length on the deviated side were lower than on the nondeviated side at T0, T1, and T2 (P <0.05). There were no significant differences in maximum cross-sectional area and CSAF bilaterally. The orientation was coronally more vertical and sagittally more forward on the deviated side (both P <0.001). In the symmetrical group, there were no significant bilateral differences at T0, T1, and T2. The volume, major axis length, and CSAF decreased, and the coronal orientation was more vertical on the nondeviated side at T2 than at T0 in both groups (P <0.05). The coronal plane orientation was more inclined on the deviated side at T2 than at T0 in the asymmetrical group (P <0.05). CONCLUSIONS: The smaller volume on the deviated side at T2 indicates the need for myofunctional training after surgery. The masseter muscle volume and the cross-sectional area did not recover to the preorthodontic levels. Studies with longer follow-up durations are needed to confirm these findings.

HC-Net: A hybrid convolutional network for non-human primate brain extraction.

Brain extraction (skull stripping) is an essential step in the magnetic resonance imaging (MRI) analysis of brain sciences. However, most of the current brain extraction methods that achieve satisfactory results for human brains are often challenged by non-human primate brains. Due to the small sample characteristics and the nature of thick-slice scanning of macaque MRI data, traditional deep convolutional neural networks (DCNNs) are unable to obtain excellent results. To overcome this challenge, this study proposed a symmetrical end-to-end trainable hybrid convolutional neural network (HC-Net). It makes full use of the spatial information between adjacent slices of the MRI image sequence and combines three consecutive slices from three axes for 3D convolutions, which reduces the calculation consumption and promotes accuracy. The HC-Net consists of encoding and decoding structures of 3D convolutions and 2D convolutions in series. The effective use of 2D convolutions and 3D convolutions relieves the underfitting of 2D convolutions to spatial features and the overfitting of 3D convolutions to small samples. After evaluating macaque brain data from different sites, the results showed that HC-Net performed better in inference time (approximately 13 s per volume) and accuracy (mean Dice coefficient reached 95.46%). The HC-Net model also had good generalization ability and stability in different modes of brain extraction tasks.

Robust Collaborative Learning of Patch-Level and Image-Level Annotations for Diabetic Retinopathy Grading From Fundus Image.

Diabetic retinopathy (DR) grading from fundus images has attracted increasing interest in both academic and industrial communities. Most convolutional neural network-based algorithms treat DR grading as a classification task via image-level annotations. However, these algorithms have not fully explored the valuable information in the DR-related lesions. In this article, we present a robust framework, which collaboratively utilizes patch-level and image-level annotations, for DR severity grading. By an end-to-end optimization, this framework can bidirectionally exchange the fine-grained lesion and image-level grade information. As a result, it exploits more discriminative features for DR grading. The proposed framework shows better performance than the recent state-of-the-art algorithms and three clinical ophthalmologists with over nine years of experience. By testing on datasets of different distributions (such as label and camera), we prove that our algorithm is robust when facing image quality and distribution variations that commonly exist in real-world practice. We inspect the proposed framework through extensive ablation studies to indicate the effectiveness and necessity of each motivation. The code and some valuable annotations are now publicly available.

Automatic Masseter Muscle Accurate Segmentation from CBCT Using Deep Learning-Based Model.

Segmentation of the masseter muscle (MM) on cone-beam computed tomography (CBCT) is challenging due to the lack of sufficient soft-tissue contrast. Moreover, manual segmentation is laborious and time-consuming. The purpose of this study was to propose a deep learning-based automatic approach to accurately segment the MM from CBCT under the refinement of high-quality paired computed tomography (CT). Fifty independent CBCT and 42 clinically hard-to-obtain paired CBCT and CT were manually annotated by two observers. A 3D U-shape network was carefully designed to segment the MM effectively. Manual annotations on CT were set as the ground truth. Additionally, an extra five CT and five CBCT auto-segmentation results were revised by one oral and maxillofacial anatomy expert to evaluate their clinical suitability. CBCT auto-segmentation results were comparable to the CT counterparts and significantly improved the similarity with the ground truth compared with manual annotations on CBCT. The automatic approach was more than 332 times shorter than that of a human operation. Only 0.52% of the manual revision fraction was required. This automatic model could simultaneously and accurately segment the MM structures on CBCT and CT, which can improve clinical efficiency and efficacy, and provide critical information for personalized treatment and long-term follow-up.