Activity

Using femurs and tibiae from ten male C57BL/6 mice, aged six to eight weeks, obtained from the Laboratory Animal Center of Guizhou Medical University, BMDM cells were isolated and cultured. Subjects were divided into a control group, an LPS group treated with 1 gram per milliliter for 24 hours, and three further groups pre-treated with varying concentrations of PLX3397 (100, 500, and 1000 nanomoles per liter for 4 hours respectively), subsequently administered with 1 gram per milliliter of LPS for 24 hours. The markers characteristic of macrophages were ascertained via flow cytometry. Cell viability was determined using the Cell Counting Kit-8 assay, and cellular senescence was identified through senescence-associated β-galactosidase (SA-β-gal) staining. Protein expression of cycle-dependent kinase inhibitors p16, p21, and CSF-1R was measured using Western blotting, and the intracellular presence of p16 and p21 proteins was confirmed through immunofluorescence. Quantitative polymerase chain reaction (qPCR) with real-time fluorescence detection was used to measure mRNA levels of senescence-associated secretory phenotype (SASP) genes, including interleukin (IL)-1, chemokine-1/10 (CXCL-1/10), matrix metalloproteinase-8 (MMP-8), and transforming growth factor- (TGF-). Treatment with PLX3397, at both medium and high concentrations, led to a substantial decrease in the rate of SA,gal positive staining (3933493% and 3633306% respectively) compared to the LPS group (5200300%). The reductions were statistically significant (P=0.020 and P=0.005). CSF-1R protein levels, following pretreatment with low, medium, and high concentrations of PLX3397 (074018, 061007, 054006), were demonstrably lower than in the LPS group (116008), exhibiting statistically significant differences (P=0013, P=0002, P<0001). Groups exposed to low, medium, and high concentrations of PLX3397 (104006, 090005, 118008) presented a consistent trend (290025) in CSF-1R mRNA expression levels, with a statistically significant difference (P < 0.0001). Compared to the LPS group (6,688,585), the average p16 fluorescence intensity in the PLX3397 pretreatment groups was significantly lower (4,976,365, 4,821,172, and 4,799,126), as determined by p-values less than 0.0001 for each comparison. Pretreatment with PLX3397 at medium and high concentrations led to a significant reduction in the average fluorescence intensity of p21 (3,443,362 and 3,013,086, respectively) when compared to the LPS group (4,682,533), which displayed a statistically significant decrease (P=0.0043 and P=0.0007 respectively). The expression of p16 protein was significantly lower in the PLX3397 pretreatment groups (056004, 055004, 035019) compared to the LPS group (098010), producing statistically significant results (p=0.0003, p=0.0002, p<0.0001). The same was true for p21 protein expression (069020, 042008, 026014) in these groups, demonstrating significant reduction compared to the LPS group (098010) with p-values (p=0.0032, p=0.0002, p<0.0001). Pre-treatment with PLX3397 resulted in a significant decrease in the expression of IL-6, IL-1, CXCL-1, CXCL-10, and MMP-8 compared to the LPS control group (P < 0.0001), with a simultaneous significant increase in TGF-β expression (P < 0.0001), as determined by RT-qPCR. LPS stimulation can induce cellular senescence, elevate SASP release, and exacerbate local inflammation by activating CSF-1R on bone marrow-derived macrophages’ cell surfaces. LPS-mediated CSF-1R activation might be diminished by the CSF-1R inhibitor PLX3397, which could contribute to preventing senescence in LPS-induced bone marrow-derived macrophages.

A study was undertaken to investigate the correlation between the type of attachment of the lateral pterygoid muscle and the placement of the temporomandibular joint (TMJ) disc, specifically in individuals suffering from temporomandibular disorders, utilizing a wireless amplified magnetic resonance imaging detector (WAND) coupled with a standard head and neck coil. From October 2019 to January 2022, eighty-five patients exhibiting TMD, diagnosed by oral and maxillofacial surgeons at Guizhou Provincial People’s Hospital, were included in the study. In total, 160 TMJs were taken into account for the analysis. The group comprised 16 males and 69 females, with their ages collectively totaling (327142) years. Bilateral TMJ scans, along with open, closed oblique sagittal and coronal WAND head and neck coils, were used to image every patient. To explore the link between LPM attachment types and TMJ disc position in patients with temporomandibular disorders (TMD), high-resolution imaging of the TMJ and LPM was employed, along with an assessment of the potential clinical value of these attachment types in TMD patients. neuronal signaling inhibitors 2 tests and Pearson correlation analysis were used to determine the correlation’s existence between LPM attachment type and TMJ disc location. In a sample of 160 instances, 51 exhibited LPM attachment type 1, representing 319% of the total. Separately, 77 cases displayed type 2, accounting for 481% of the total, and 32 cases demonstrated type 3, which comprised 200% of the total. A considerable correlation was observed between the characteristics of the LPM attachment and the position of the articular disc (F=2820, P=0.0002, r=0.776). No statistically significant relationship was found between the type of LPM attachment and the reversible displacement of the articular disc, based on the analysis (2=0.24, P=0.887, r=0.825). There’s a discernible relationship between the type of LPM attachment and the disc’s placement in TMD patients. Reliable imaging to assess LPM attachment type and disc location in TMD patients is possible with a high-resolution TMJ scan, using WNAD coupled with a conventional head and neck joint coil.

To automatically categorize orthodontic image data, this research aims to develop a multi-classification orthodontic image recognition system, leveraging the SqueezeNet deep learning model. The Department of Orthodontics, Capital Medical University School of Stomatology, acquired a total of 35,000 clinical orthodontic images; data collection periods included October-November 2020 and June-July 2021. 490 orthodontic patients’ images, with a male-to-female ratio of 4951 and an age range from 4 to 45 years, formed the image dataset. After the application of selection criteria to cleanse the data, the ultimate image set consisted of 17,453 facial pictures (frontal, smiling, 90-degree right, 90-degree left, 45-degree right, and 45-degree left), 8,026 intraoral photographs (frontal occlusion, right occlusion, left occlusion, upper occlusal view – original and mirrored, lower occlusal view – original and mirrored, and occlusal relationship coverage), 4,115 X-ray images (left lateral skull view, right lateral skull view, frontal skull view, cone-beam CT, and wrist bone view), and 684 further non-orthodontic images. Image labeling tools were used by a team of orthodontic doctoral students, associate professors, and professors to categorize orthodontic images. The 20 categories included 6 face image categories, 8 intraoral image categories, 5 X-ray image categories, and miscellaneous image categories. Within the Python programming language, the random function was used to randomly segment the data for each label into training, validation, and testing sets, observed in an 81:1:1 ratio. To train the model, the enhanced SqueezeNet deep learning architecture was used; further, 13,000 natural images from the freely accessible ImageNet database provided supplementary non-orthodontic data to optimize the algorithm’s anomaly processing routines. A system for recognizing orthodontic images, employing deep learning models for multi-classification, was created. Prediction results from the test set, measured using precision, recall, F1 score, and confusion matrix, informed the assessment of orthodontic image classification accuracy. The reliability of the model’s image classification judgment rationale was established using the gradient-weighted class activation mapping (Grad-CAM) method, resulting in the creation of heat maps. Following data cleansing and labeling, a total of 30,278 orthodontic images were incorporated into the dataset. The classification accuracy of the test set revealed that the majority of labels had precision, recall, and F1 scores of 100%. Just 5 out of 3,047 images were incorrectly classified, giving a system accuracy of 99.84% (meaning 3,042 images were classified correctly). Every anomaly data point was processed with 100% precision; this translates to 10,500 correctly processed data points out of a total of 10,500. A consistency was observed between human judgment and the image classification basis of the SqueezeNet deep learning model, as illustrated by the heat map. Using an upgraded SqueezeNet deep learning model, this study built a multi-classification orthodontic image recognition system to automatically sort and classify 20 types of orthodontic images. Orthodontic image classification demonstrated high accuracy within the system.

Our study proposes an automatic method for locating anatomical reference points within the three-dimensional (3D) maxillary complex data, and it aims for a preliminary assessment of its accuracy and reproducibility. From the patient database at Peking University School and Hospital of Stomatology’s Department of Oral and Maxillofacial Surgery, spiral CT data of 31 patients with comparatively normal craniofacial morphology was selected between June 2021 and December 2022. A sample consisting of fifteen males and sixteen females, exhibiting an age of 33383 years, was included. Mimics software facilitated the creation of a 3D reconstruction of the maxillary complex, which was then further processed with mesh refinement using Geomagic software. A total of 24 anatomical landmarks were precisely located by the combined efforts of two attending physicians and one associate chief physician who manually landmarked each of the 31 maxillary complex datasets. The average of the three expert landmarking results served as the expert-defined landmarks. A case exhibiting the average 3D morphological characteristics common to healthy individuals’ craniofacial bones was chosen as the template, and the 30 remaining cases were used as target data. The MeshMonk program (a non-rigid registration algorithm), open-source, facilitated an initial alignment of the template and target datasets, relying on four landmarks: nasion, left zygomatic arch prominence, right zygomatic arch prominence, and anterior nasal spine.