Non-invasive imaging for predicting labial salivary gland biopsy outcomes in patients with suspected primary Sjögren syndrome.

To identify the value of salivary gland ultrasound (SGUS) combined with magnetic resonance imaging (MRI) and magnetic resonance sialography (MRS) in predicting the results of labial salivary gland biopsy (LSGB) in patients with suspected primary Sjögren syndrome (pSS), and construct a nomogram model to predict LSGB results. A total of 181 patients who were admitted with suspected pSS from December 2018 to April 2023 were examined and divided into a training set (n = 120) and a validation set (n = 61). Baseline data of the two groups were examined, and the value of SGUS, MRI, and MRS in predicting LSGB was analyzed. Multivariate logistic analysis was used to screen for risk factors, and nomogram prediction models were constructed using these results. In the training set, the SGUS, MRI, and MRS scores of patients in the LSGB + group were higher than those in the LSGB - group (all P < 0.001). The positive prediction value (PPV) was 91% for an SGUS score of 3, and 82% for MRI and MRS scores of 2 or more. We developed a nomogram prediction model based on SGUS, MRI, and MRS data, and it had a concordance index (C-index) of 0.94. The Hosmer-Lemeshow test (χ2 = 3.17, P = 0.92) also indicated the nomogram prediction model had good accuracy and calibration for prediction of LSGB results. A nomogram model based on SGUS, MRI, and MRS results can help rheumatologists decide whether LSGB should be performed in patients with suspected pSS.

Overcoming Data Deficiency for Multi-Person Pose Estimation.

Building multi-person pose estimation (MPPE) models that can handle complex foreground and uncommon scenes is an important challenge in computer vision. Aside from designing novel models, strengthening training data is a promising direction but remains largely unexploited for the MPPE task. In this article, we systematically identify the key deficiencies of existing pose datasets that prevent the power of well-designed models from being fully exploited and propose the corresponding solutions. Specifically, we find that the traditional data augmentation techniques are inadequate in addressing the two key deficiencies, imbalanced instance complexity (IC) (evaluated by our new metric IC) and insufficient realistic scenes. To overcome these deficiencies, we propose a model-agnostic full-view data generation (Full-DG) method to enrich the training data from the perspectives of both poses and scenes. By hallucinating images with more balanced pose complexity and richer real-world scenes, Full-DG can help improve pose estimators' robustness and generalizability. In addition, we introduce a plug-and-play adaptive category-aware loss (AC-loss) to alleviate the severe pixel-level imbalance between keypoints and backgrounds (i.e., around 1:600). Full-DG together with AC-loss can be readily applied to both the bottom-up and top-down models to improve their accuracy. Notably, plugging into the representative estimators HigherHRNet and HRNet, our method achieves substantial performance gains of 1.0%-2.9% AP on the COCO benchmark, and 1.0%-5.1% AP on the CrowdPose benchmark.

The causal effects of age at menarche, age at first live birth, and estradiol levels on systemic lupus erythematosus: A two-sample Mendelian randomization analysis.

OBJECTIVES: To determine whether age at menarche (AAM), age at first live birth (AFB), and estradiol levels are causally correlated with the development of systemic lupus erythematosus (SLE). METHODS: A two-sample Mendelian randomization (MR) analysis was performed after data was collected from a dataset of genome-wide association studies (GWASs) related to SLE (as outcome), and from open access databases to find statistics related to AAM, AFB, and estradiol levels (as exposure). RESULT: In our study, a negative causal correlation between AAM and SLE was confirmed by MR analysis (MR egger: beta = 0.116, SE = 0.948, p = 0.909; weighted median: beta = -0.416, SE = 0.192, p = 0.030; and IVW: beta = -0.395, SE = 0.165, p = 0.016). However, there were no genetic causal effects of AFB and the estradiol levels on SLE, based on the results of MR analysis as follows: AFB (MR egger: beta = - 2.815, SE = 1.469, p = 0.065; Weighted median: beta = 0.334, SE = 0.378, p = 0.377; and IVW: beta = 0.188, SE = 0.282, p = 0.505) and the estradiol levels (MR egger: beta = 0.139, SE = 0.294, p = 0.651; weighted median: beta = 0.063, SE = 0.108, p = 0.559; IVW: beta = 0.126, SE = 0.097, p = 0.192). CONCLUSIONS: Our findings revealed that AAM may be associated with increased risk of the development of SLE, while there were no such causal effects from AFB and estradiol levels.

KE-RCNN: Unifying Knowledge-Based Reasoning Into Part-Level Attribute Parsing.

Part-level attribute parsing is a fundamental but challenging task, which requires the region-level visual understanding to provide explainable details of body parts. Most existing approaches address this problem by adding a regional convolutional neural network (RCNN) with an attribute prediction head to a two-stage detector, in which attributes of body parts are identified from localwise part boxes. However, localwise part boxes with limit visual clues (i.e., part appearance only) lead to unsatisfying parsing results, since attributes of body parts are highly dependent on comprehensive relations among them. In this article, we propose a knowledge-embedded RCNN (KE-RCNN) to identify attributes by leveraging rich knowledge, including implicit knowledge (e.g., the attribute "above-the-hip" for a shirt requires visual/geometry relations of shirt-hip) and explicit knowledge (e.g., the part of "shorts" cannot have the attribute of "hoodie" or "lining"). Specifically, the KE-RCNN consists of two novel components, that is: 1) implicit knowledge-based encoder (IK-En) and 2) explicit knowledge-based decoder (EK-De). The former is designed to enhance part-level representation by encoding part-part relational contexts into part boxes, and the latter one is proposed to decode attributes with a guidance of prior knowledge about part-attribute relations. In this way, the KE-RCNN is plug-and-play, which can be integrated into any two-stage detectors, for example, Attribute-RCNN, Cascade-RCNN, HRNet-based RCNN, and SwinTransformer-based RCNN. Extensive experiments conducted on two challenging benchmarks, for example, Fashionpedia and Kinetics-TPS, demonstrate the effectiveness and generalizability of the KE-RCNN. In particular, it achieves higher improvements over all existing methods, reaching around 3% of AP allIoU+F1 on Fashionpedia and around 4% of Accp on Kinetics-TPS. Code and models are publicly available at: https://github.com/sota-joson/KE-RCNN.

Value of magnetic resonance imaging and sialography of the parotid gland for diagnosis of primary Sjögren syndrome.

AIM: To evaluate the utility of magnetic resonance imaging (MRI) and magnetic resonance sialography (MRS) for diagnosis of primary Sjögren syndrome (pSS) singly or integrated with 2016 American College of Rheumatology (ACR)/European League Against Rheumatic Diseases (EULAR) classification criteria. METHODS: The diagnostic efficiencies of MRI, MRS, and labial salivary gland biopsy (LSGB) were evaluated. The prediction model was established by multivariate analysis. Finally, performance of the ACR/EULAR criteria was evaluated after addition of MRI + MRS or replacement of original items by MRI + MRS. RESULTS: The combined use of LSGB + MRI + MRS provided the greatest diagnostic value. MRI and MRS grade had positive correlations with disease duration and pathological grade of the labial gland (both P < 0.001). MRI and MRS grade had positive correlations with xerostomia severity and negative correlations with unstimulated salivary flow rate (both P < 0.001). The consistency of MRI grade and MRS grade in the diagnosis of parotid gland lesions was poor (κ = 0.253, P < 0.001). The diagnostic efficiency of our prediction model (AUC 0.906) was similar to that of criteria from the ACR/EULAR (AUC 0.930). Adding MRI + MRS to the ACR/EULAR criteria improved the sensitivity (92.3% vs 90.8%), whereas the specificity remained the same (88.9% vs 89.1%). Replacing LSGB by MRI + MRS in the ACR/EULAR criteria decreased both sensitivity and specificity (88.1% vs 90.8% and 86.4% vs 89.1%, respectively). CONCLUSION: The combined application of MRI and MRS has ideal clinical application value in the diagnosis of early-stage pSS. Validity of the ACR/EULAR criteria remains high after incorporation of MRI + MRS.