Early Prediction of Radiation Treatment Response via Longitudinal Analysis of CBCT Radiomic Features for Locally Advanced Rectal Cancer

PURPOSE/OBJECTIVE(S)

Patients respond to the same radiation treatment course differently due to inter- and intra- patient variability in radiosensitivity. Despite widespread use of AI/ML in radiation oncology, there is a lack of monitoring strategies used during treatment courses to evaluate early predictors of treatment response in a systematic fashion. This work advances a straightforward, yet effective, method for the early detection of treatment response through systematically analyzing daily CBCT radiomic features. The goal is to aid clinicians in assessing the treatment efficacy routinely with a view towards optimizing personalized treatment.

MATERIALS/METHODS

We included a cohort of 30 patients diagnosed with locally advanced rectal cancer who underwent neo-adjuvant fractionated radiation treatment (RT) with a prescription dose of 50.4 Gy (28 fractions), followed by total mesorectal excision surgery after completion of ChemoRT. Daily IGRT imaging was acquired prior to each fraction resulting in a total of 840 CBCTs. Patients were divided into responder (14 patients) and non-responder (16 patients) groups based on post-RT pathological response. Mutual information algorithms were utilized to rigorously register daily CBCT images to the planning CT, and longitudinal radiomic features of the target were extracted from the daily CBCTs during the entire treatment course. All longitudinal features for a given patient were standardized with Z-Score normalization, followed by linear fitting using the least square method, resulting in radiomic feature trends (RFT) represented by slope values. Statistical significance was established via a two-sample U test and P-value with a threshold of 0.05. Logistic regression was performed to eliminate RFT with accuracy rates lower than 0.5. The final trending model was developed using random forest. For each patient at fraction N, our investigation involved independent 27 group experiments, where each experiment considered image group from fraction #1 to N, to confirm the effectiveness and stability of the model.

RESULTS

The proposed RFT demonstrated a high level of precision and consistency for post-RT response based on longitudinal CBCT images for LARC patients. The trending model yielded an accuracy of 0.9556, 95% CI (0.94, 0.972) when each daily image was considered, the prediction consistency was 0.964. Given the first 14 experiments (considering group images of fraction #1-15), the prediction accuracy was 0.9357, 95% CI (0.915, 0.956) and the prediction consistency was 0.952.

CONCLUSION

A strategy for monitoring and early prediction of LARC patients' radioresponse was evaluated via longitudinal CBCT assessment. Our trending models demonstrate a significant difference between the responder vs non-responder groups as early as the 15th fraction. Our strategy achieved superior accuracy and consistency to predict post-RT response of LARC patients.

Ceruloplasmin regulating fibrosis in orbital fibroblasts provides a novel therapeutic target for Graves' orbitopathy

PURPOSE

In diagnosing the pathogenesis of Graves' orbitopathy (GO), there is a growing interest in fibrosis generated by orbital fibroblasts (OFs); nevertheless, the involvement of ceruloplasmin (CP) in OFs remains unknown.

METHODS

Differentially expressed genes (DEGs) were identified through bioinformatic analysis. OFs were isolated from orbital tissue and identified with immunofluorescent staining. The levels of DEGs were validated in GO tissue samples and TGF-β-challenged OFs, and CP was selected for the following laboratory investigations. CP overexpression or knockdown was achieved, and cell viability and fibrosis-associated proteins were investigated to assess the cell phenotype and function. Signaling pathways were subsequently investigated to explore the mechanism of CP function in OFs.

RESULTS

CP and cathepsin C (CTSC) are two overlapped DEGs in GSE58331 and GSE105149. OFs were isolated and identified through fibrotic biomarkers. CP and CTSC were downregulated in GO tissue samples and TGF-β-challenged OFs. CP overexpression or knockdown was achieved in OFs by transducing a CP overexpression vector or small interfering RNA against CP (si1-CP or si2-CP) and verified using a qRT-PCR. CP overexpression inhibited cell viability and reduced the levels of α-SMA, vimentin, fibronectin, and collagen I, whereas CP knockdown exerted opposite effects on OFs. CP overexpression inhibited the phosphorylation of Smad3, Erk1/2, p38, JNK, and AKT; conversely, CP knockdown exerted opposite effects on the phosphorylation of factors mentioned above.

CONCLUSION

CP was downregulated in GO and suppressed the expression of fibrosis-associated proteins in both GO and normal OFs. CP might serve as a promising therapeutic agent in the treatment regimens for GO.

A Novel Multi-Objective Based Feature Selection Method for Response Prediction

PURPOSE/OBJECTIVE(S)

Accurate response prediction is essential towards personalized treatment in radiation therapy. Excessive imaging features, extracted from medical images, pose a great challenge in radiomic analyses. Feature selection is an essential step to remove redundant and irrelevant features for model construction.

MATERIALS/METHODS

We proposed a novel multi-objective based radiomic feature selection method (MRMOPSO), where the number of features, sensitivity, and specificity are jointly considered as optimization objectives for feature selection. The MRMOPSO innovated by three aspects: 1) Fisher score initialize the feature population to speed up the convergence; 2) Min-redundancy particle generation operations to reduce the redundancy between radiomic features, a truncation strategy was also introduced; 3) Particle selection operation guided by elitism strategies to improve local search ability of the algorithm. We evaluated the effectiveness of the proposed MRMOPSO method using a cohort of oropharyngeal cancer patients from The Cancer Imaging Archive (TCIA). 357 patients were used for model training and additional 64 patients were used for independent evaluation. The proposed methods were compared with (a) classical feature selection methods, i.e., Lasso, minimal-redundancy-maximal-relevance criterion (mRMR), F-score, and mutual information (MI), (b) single-objective feature selection methods, i.e., genetic algorithm (GA), particle swarm optimization algorithm (PSO) and (c) multi-objective feature selection methods, i.e., multiple objective particle swarm optimization (MOPSO), nondominated sorting genetic algorithm II (NSGA II).

RESULTS

The other feature selection methods yielded AUCs, sensitivity, specificity of (0.48-0.71), (0.49-0.86), (0.33-0.67), respectively. The MRMOPSO achieved significantly highly AUC of 0.84 with smaller number of selected features on the independent dataset (Table 1). Additionally, the MRMOPSO remarkably improved the sensitivity (0.81), specificity (0.81) and achieved an excellent balance between sensitively and specificity.

CONCLUSION

We demonstrated a novel multi-objective based radiomic feature selection method. The proposed algorithm effectively reduced feature dimension, and achieved superior AUC with simultaneous improved sensitivity and specificity, for radiomic response prediction.

Impact of Prophylactic Pelvic Lymph Node Irradiation in De-Novo Oligometastatic Prostate Cancer

PURPOSE/OBJECTIVE(S)

To evaluate the impact of prophylactic pelvic nodal irradiation in de-novo oligometastatic prostate cancer treated with radiotherapy (RT) for both primary tumor and all metastatic lesions.

MATERIALS/METHODS

This was a single-center prospective cohort study. De novo oligometastatic prostate cancer patients with RT for both primary tumor and all metastatic lesions were included. Kaplan-Meier method, log rank test and cox regression were used to calculate OS and PFS. PFS included PSA failure, local or distant failure assessed by imaging.

RESULTS

This study analyzed 202 patients from 10/2011 to 1/2022 with median follow-up of 48 months. A total of 126 (62.4%) patients were treated with pelvic lymph node RT. The dose was 47.5 Gy with 1.9 Gy per fraction. Among them, 66 (32.7%) patients were treated with whole pelvic RT (WPRT), which the upper limit was at the aortic bifurcation. 60 (29.7%) patients were treated with mini-WPRT, which the upper limit was at the lower margin of obturator foramen. The incidence of diarrhea (P = 0.038) and leukocyte reduction (P = 0.040) in the WPRT subgroup during radiotherapy was significantly higher than that in the mini-WPRT and non-pelvic RT subgroup. For the whole cohort, the median OS and PFS were not reached. The subgroup analysis showed that the elective pelvic nodal irradiation could improve PFS (P = 0.042). However, there was no difference of PFS between standard WPRT and mini-WPRT.

CONCLUSION

The study suggests that for de-novo oligometastatic prostate cancer, elective pelvic nodal irradiation may improve PFS. For patients who cannot tolerate WPRT, mini-WPRT may be an alternative option. However, it needs to be verified in the prospective RCT study.

Unsupervised Domain Adaptation of Auto-Segmentation on Multi-Source MRIs

PURPOSE/OBJECTIVE(S)

Deep learning has achieved great success in medical image segmentation. Most existing deep learning (DL) approaches make no adjustments to the model prior to inference. These models can perform well on the data of the same distribution, but their performance usually degrades when applied to the images from different source, i.e., different scanners. To tackle the problem caused by domain shift, we proposed an unsupervised domain adaptation (UDA) method based on entropy minimization and physical consistency constraints.

MATERIALS/METHODS

The proposed method combines feature-level and instance-level domain adaptation techniques to transfer knowledge from the source to the target domain. Specifically, the feature-level adaptation technique uses a graph-based entropy minimization to reduce the discrepancy between the source and target domains. The instance-level adaptation technique employs a novel consistency loss to regularize the physical consistency of the same object, such as volume, length, and centroid, thus improving the segmentation accuracy of the target domain. A collection of 93 abdominal MR images, comprising 45 cases from a 0.35T MRI scanner (TRUFI) and 48 cases from a 1.5T MRI scanner (T2), was utilized to evaluate the effectiveness of the proposed method. The contours of 6 organs-at-risk were delineated by a senior radiation oncologist, serving as the ground truth. Three models, the source model (SRC) trained on the source domain, the target model (TGT) trained on the target domain, and the UDA model adapted from the source domain to the target domain, were compared on the target domain using the Dice Similarity Coefficient (DSC).

RESULTS

In the experiment of 0.35T-to-1.5T, the proposed UDA method outperformed the source model, achieving an average DSC score of 0.82 ± 0.11, compared to 0.58 ± 0.23 (SRC) and 0.85 ± 0.09 (TGT), respectively. In the inverse experiment 1.5T-to-0.35T, the UDA model achieved an average DSC score of 0.79±0.13, compared to DSCs of 0.52 ± 0.25 and 0.81 ± 0.11 for the SRC and TGT respectively. The UDA method yielded a significant improvement of 46%, compared to the SRC. Particularly, OARs (organ at risk) with higher deformability such as the stomach and duodenum achieved a 58% and 63% improvement in performance, respectively.

CONCLUSION

This work presents a compelling approach of UDA for auto-segmentation on multi-source MRIs. Experimental results demonstrate that the UDA effectively improve the segmentation performance of the source model in the target domain, resulting in a more robust segmentation model.

Feature Selection Based on Unsupervised Clustering Mechanism on Multiple-Sequence MRIs for Predicting Neoadjuvant Chemoradiation Response in Locally Advanced Rectal Cancer

PURPOSE/OBJECTIVE(S)

Accurate response prediction allows for personalized cancer management. We developed an unsupervised clustering mechanism to improve effectiveness and efficiency in feature selection operation for accurate patient stratification.

MATERIALS/METHODS

Forty-three locally advanced rectal cancer (LARC) patients underwent neoadjuvant chemoradiation were included, pre-treatment T2 and ADC MRIs were acquired for each patient. An initial feature space consisting of 200 radiomic features extracted from manually delineated GTVs from two sequences of MR images. Additional 960 high-order radiomic features extracted from a 3D convolutional neural network (CNN). To remove redundant and irrelevant features, we developed an unsupervised clustering-based feature selection operation to determine the combination of features with potential best performance. The normal process of feature selection involves searching new feature combinations and training new classifiers for evaluating their performance via an iterative process based on selected feature set, the overall time cost is tremendous. To balance the computational cost and search efficiency, firstly, we proposed an unsupervised clustering analysis metric- Comprehensive Cluster Analysis Index (CCAI) through the K-means algorithm, where the average distances between the sample points and the cluster centroids and so on, to construct a multiple linear regression model. Secondly, we extracted sample points by varying the number of features and feature ratios between radiomic features and 3D-CNN features in the output of feature selection. Thirdly, we optimized the model using the sampling points to calculate the CCAI. Two typical feature combination search algorithms, the random forest recursive feature elimination (RF-RFE) and the differential evolution (DE), were used to perform feature selection with CCAI.

RESULTS

The accuracy, area-under-curve (AUC) and specificity, based on combined 3D-CNN and radiomic features extracted from combined T2 and ADC images, were 0.852, 0.871, and 0.735, respectively. Our experiments illustrated higher predictive power (AUC = 0.846) based on high-order abstract features extracted from the CNN on ADC and T2 images, compared to the traditional radiomic model (AUC = 0.714). Additionally, the predictive models constructed based on radiomics and CNN features extracted from ADC images were more predictable in terms of treatment responses than the radiomic and CNN imaging features extracted from T2 images. The average computational time of DE and RF-RFE were 50.5s and 128.6s in one single computation, the average computational time were 24.2s and 91.3s with CCAI, respectively.

CONCLUSION

We proposed an unsupervised clustering analysis mechanism to improve the effectiveness of feature selection while decreasing its time cost markedly, which highlight the correlation and complementarity between low- and high-level imaging features, achieving better predictive accuracy.

Per- and polyfluoroalkyl substances in umbilical cord serum and body mass index trajectories from birth to age 10 years: Findings from a longitudinal birth cohort (SMBCS)

BACKGROUND

Prenatal exposure to per- and polyfluoroalkyl substances (PFAS) has been linked to low birth weight but higher childhood weight and obesity. However, little is known regarding the associations between PFAS exposure and dynamic body mass index (BMI) trajectories, particularly from birth through preadolescence.

OBJECTIVE

To evaluate the associations of cord serum PFAS concentrations with BMI trajectories from birth to age 10 years and longitudinal BMI in different periods.

METHODS

Based on 887 mother-child pairs in the longitudinal prospective birth cohort, we measured 12 PFAS congeners in cord serum and calculated BMI with anthropometric indicators at 9 follow-up time points from birth to age 10 years. The BMI trajectories were identified using group-based trajectory model (GBTM). To estimate the associations of cord serum PFAS levels with BMI trajectories and longitudinal changes in BMI, logistic regression models, linear mixed models, Bayesian kernel machine regression, and quantile-based g-computation models (QGC) were used.

RESULTS

The median concentrations of 10 PFAS congeners included in statistical analysis ranged from 0.047 to 3.623 μg/L. Two BMI trajectory classes were identified by GBTM, characterized by high group and low group. In logistic regression models, five PFAS congeners (PFBA, PFHpA, PFHxS, PFHpS, and PFDoDA) were associated with the higher probability of being in high BMI trajectory group (odds ratio, OR: 1.21 to 1.74, p < 0.05). Meanwhile, higher PFAS mixture were related to elevated odds for the high group in both BKMR models and QGC models, with PFHpA and PFHpS being the two most important drivers jointly. In the sex-stratified analysis, the positive associations remained significant exclusively among males. In the longitudinal analysis, PFUnDA and PFDoDA were associated with increased BMI from birth to age 10 years. Furthermore, PFBS and PFHpA were negatively related to BMI throughout infancy and toddlerhood (from birth to age 3 years), whereas PFDoDA confirmed a positive association with mid-childhood (from age 6 to 10 years) BMI.

CONCLUSIONS

Prenatal PFAS exposure was positively associated with BMI trajectories from birth to preadolescence and longitudinal BMI in various periods. Future research could use better trajectory modeling strategies to shape more complete growth trajectories and explore the relationship between BMI trajectories and adulthood health.

[Occurrence and exposure assessment of fumonisins from grains and grain products in Zhejiang Province in 2018-2019]

OBJECTIVE

To monitor fumonisins(FBs) in grains and grain products in Zhejiang and assess the exposure risks of FBs to local residents.

METHODS

Liquid chromatography coupled with tandem mass spectrometry method was used to determine the occurrence of FBs in rice, millet, dried noodles, instant noodles, and maize grains, and food frequency questionnaires were used to collect the food consumption data of Zhejiang population. Then, the simple probability distribution model was used to assess the exposure risk.

RESULTS

The levels of FBs in rice, millet, dried noodles and instant noodles were relatively low. The occurrence of FB_1, FB_2 and FB_3 in these foods was 0-23.7%, 0-16.7% and 0-5.4%, respectively, and the mean levels were not detected(ND)-22.36, ND-20.63 and ND-7.19 μg/kg correspondingly. However, the levels of FBs in maize grains were relatively high. The occurrence of FB_1, FB_2, and FB_3 in maize grains was 100%, 93.6% and 90.3%, respectively, and the mean levels were 638.99, 103.54 and 59.69 μg/kg correspondingly. In 12.9% of the maize grain samples, the levels of FBs were higher than the standard reference. The residents were at low exposure risk overall. The mean estimated daily intake(EDI) of FBs was far lower than the provisional maximum tolerable daily intake of 2 μg/(kg·BW·d). However, 0.30% of the residents were at high risk. Among people of different ages, the mean EDI of children, adults, and elderly were 0.43, 0.28 and 0.29 μg/(kg·BW·d) respectively, and children were in the highest exposure levels of FBs. Among the tested five foodstuffs, rice and maize grains were the main sources of FBs exposure.

CONCLUSION

Except for maize grains, the levels of FBs in grains and grain products were relatively low, and Zhejiang residents were at low FBs exposure risk generally.

Cadmium Exposure in Aquatic Products and Health Risk Classification Assessment in Residents of Zhejiang, China

Cadmium (Cd) pollution of food safety is a prominent food safety concern worldwide. The concentration of Cd in six aquatic food categories collected from 2018 to 2022 was analyzed using inductively coupled plasma mass spectrometry, and the Cd exposure levels were calculated by combining the Cd concentration and food consumption data of 18913 urban and rural residents in Zhejiang Province in 2015-2016. The mean Cd concentration was 0.699 mg/kg and the mean Cd exposure of aquatic foods was 0.00951 mg/kg BW/month for the general population. Marine crustaceans were the largest Cd contributor, corresponding to 82.7%. The regional distribution results showed that the average Cd exposure levels of 11 cities did not exceed the provisional tolerable monthly intake (PTMI). According to the subgroups, the Cd mean exposure level of 2-3-year-old children was significantly higher than that of the other age groups but did not exceed the PTMI. Health risk classification assessment demonstrated that the final risk score was six, and the health risk level of Cd exposure in aquatic products in the Zhejiang population was medium. These results demonstrated that the risk of Cd exposure in certain food types or age groups should be given more concern.

[Contamination and dietary exposure assessment of 2-chloropropanol esters in vegetable oils available on Zhejiang market during 2016-2020]

OBJECTIVE

To investigate the occurrence data for 2-chloropropanol(2-MCPD) ester in edible vegetable oils purchased in Zhejiang Province during 2016-2020, and to estimate dietary exposures of 2-MCPD ester via vegetable oil.

METHODS

A total of 404 samples of edible oils were collected from markets, stores, online shopping in Zhejiang Province, the occurrence of 2-MCPD ester was detected by gas chromatography mass spectrometry from 2016-2020. Data of vegetable oils consumption was obtained from the survey result of urban and rural residents in Zhejiang Province in 2008. The exposure levels of 2-chloropropanol ester was calculated for people aged 4-6, 7-10, 11-17, 18-59 and 60 years old and older from the consumption of vegetable oils.

RESULTS

The detection rate of 2-MCPD esters in 404 samples was 82.7%(334/404) with the mean level of 0.32 mg/kg. Among them, the mean content of camellia oil was the highest with 1.23 mg/kg, followed by rice oil(0.69 mg/kg); sunflower oil, olive oil and soybean oil have relatively low average values, respectively with 0.11, 0.12 and 0.13 mg/kg. There are significant differences in 2-MCPD ester content in different types of edible oils of the same brand(P<0.05), the content of 2-MCPD ester in different brands of peanut oil was significant(P<0.05), but in different brands of corn oil was not statistically significant. Among the edible vegetable oil consumers, the average exposure of 2-MCPD esters in different age groups ranged from 0.21 to 0.69 μg/(kg·BW·d).

CONCLUSION

2-MCPD esters pollution is widespread in vegetable oils, the severity of pollution was affected by the type and brand of the edible vegetable oil, and the intake of 2-MCPD esters was relatively high among people with high vegetable oil consumption and the general population of 4-10 years old.

[Analysis of the characteristics of viral infections in children with diarrhea in Beijing from 2018 to 2022]

Objective: To explore the characteristics of viral infections in children with diarrhea in Beijing from 2018 to 2022. Methods: Real-time PCR and enzyme-linked immunosorbent assay were used to detect viral nucleic acid of Norovirus (NoV), Sappovirus (SaV), Astrovirus (AstV), Enteric Adenovirus (AdV) or antigen of Rotavirus (RV) in 748 stool samples collected from Beijing Capital Institute of Pediatrics from January 2018 to December 2021. Subsequently, the reverse transcription PCR or PCR method was used to amplify the target gene of the positive samples after the initial screening, followed by sequencing, genotyping and evolution analysis, so as to obtain the characteristics of these viruses. Phylogenetic analysis was performed using Mega 6.0. Results: From 2018 to 2021, the overall detection rate of the above five common viruses was 37.6%(281/748)in children under 5 years old in Beijing. NoV, Enteric AdV and RV were still the top three diarrhea-related viruses, followed by AstV and SaV, accounting for 41.6%, 29.2%, 27.8%, 8.9% and 7.5%, respectively. The detection rate of co-infections with two or three diarrhea-related viruses was 4.7% (35/748). From the perspective of annual distribution, the detection rate of Enteric AdV was the highest in 2021, while NoV was predominant in the other 4 years. From the perspective of genetic characteristics, NoV was predominant by GII.4, and after the first detection of GII.4[P16] in 2020, it occupied the first two gene groups together with GII.4[P31]. Although the predominant RV was G9P[8], the rare epidemic strain G8P[8] was first detected in 2021. The predominant genotypes of Enteric AdV and AstV were Ad41 and HAstV-1. SaV was sporadic spread with a low detection rate. Conclusion: Among the diarrhea-related viruses infected children under 5 years of age in Beijing, the predominant strains of NoV and RV have changed and new sub-genotypes have been detected for the first time, while the predominant strains of AstV and Enteric AdV are relatively stable.

Epidemiological characteristics of Vibrio parahaemolyticus outbreaks, Zhejiang, China, 2010-2022

Background

Vibrio parahaemolyticus is one of the most common foodborne pathogens and poses a significant disease burden. The purpose of the study was to elucidate the epidemiological characteristics of Vibrio parahaemolyticus outbreaks in Zhejiang Province, and provide insights for the targeted prevention and control of foodborne diseases.

Methods

Descriptive statistical methods were utilized to analyze the data on Vibrio parahaemolyticus outbreaks reported by all Centers for Disease Control and Prevention (CDCs) through Foodborne Disease Outbreaks Surveillance System (FDOSS) in Zhejiang Province from 2010 to 2022.

Results

From 2010 to 2022, a total of 383 outbreaks caused by Vibrio parahaemolyticus were reported by 90 CDCs in 11 prefectures of Zhejiang Province, resulting in 4,382 illnesses, 326 hospitalizations and 1 death. The main symptoms of the outbreak-related cases were diarrhea (95.18%), abdominal pain (89.23%), nausea (55.64%), vomiting (50.57%), fever (24.21%), etc. The outbreaks occurring between July and September accounted for 77.54% of all outbreaks (297 out of 383). Outbreaks associated with restaurants accounted for the majority (57.96%, 222/383) of all outbreaks, followed by those linked to staff canteens (15.40%, 59/383) and rural banquets (11.23%, 43/383). 31.85% of all outbreaks were associated with the consumption of aquatic products, while ready-to-eat foods such as Chinese cold dishes and cooked meat products accounted for 12.53% of all outbreaks. Serotype O3:K6 (81.94%, 118/144) was the predominant serotype responsible for outbreaks from 2010 to 2020, while serotype O10:K4 (57.89%, 33/57) was the predominant serotype from 2021 to 2022.

Conclusion

In-depth and comprehensive analysis of long-term surveillance data on Vibrio parahaemolyticus outbreaks is essential to gain insight into the epidemiological characteristics, identify long-term patterns and recent trends, and enable governments to prioritize interventions and develop targeted policies to mitigate such outbreaks.

ST3D++: Denoised Self-Training for Unsupervised Domain Adaptation on 3D Object Detection

In this paper, we present a self-training method, named ST3D++, with a holistic pseudo label denoising pipeline for unsupervised domain adaptation on 3D object detection. ST3D++ aims at reducing noise in pseudo label generation as well as alleviating the negative impacts of noisy pseudo labels on model training. First, ST3D++ pre-trains the 3D object detector on the labeled source domain with random object scaling (ROS) which is designed to reduce target domain pseudo label noise arising from object scale bias of the source domain. Then, the detector is progressively improved through alternating between generating pseudo labels and training the object detector with pseudo-labeled target domain data. Here, we equip the pseudo label generation process with a hybrid quality-aware triplet memory to improve the quality and stability of generated pseudo labels. Meanwhile, in the model training stage, we propose a source data assisted training strategy and a curriculum data augmentation policy to effectively rectify noisy gradient directions and avoid model over-fitting to noisy pseudo labeled data. These specific designs enable the detector to be trained on meticulously refined pseudo labeled target data with denoised training signals, and thus effectively facilitate adapting an object detector to a target domain without requiring annotations. Finally, our method is assessed on four 3D benchmark datasets (i.e., Waymo, KITTI, Lyft, and nuScenes) for three common categories (i.e., car, pedestrian and bicycle). ST3D++ achieves state-of-the-art performance on all evaluated settings, outperforming the corresponding baseline by a large margin (e.g., 9.6%  ∼  38.16% on Waymo → KITTI in terms of AP[Formula: see text]), and even surpasses the fully supervised oracle results on the KITTI 3D object detection benchmark with target prior. Code is available at https://github.com/CVMI-Lab/ST3D.

Fully Convolutional Networks for Panoptic Segmentation With Point-Based Supervision

In this paper, we present a conceptually simple, strong, and efficient framework for fully- and weakly-supervised panoptic segmentation, called Panoptic FCN. Our approach aims to represent and predict foreground things and background stuff in a unified fully convolutional pipeline, which can be optimized with point-based fully or weak supervision. In particular, Panoptic FCN encodes each object instance or stuff category with the proposed kernel generator and produces the prediction by convolving the high-resolution feature directly. With this approach, instance-aware and semantically consistent properties for things and stuff can be respectively satisfied in a simple generate-kernel-then-segment workflow. Without extra boxes for localization or instance separation, the proposed approach outperforms the previous box-based and -free models with high efficiency. Furthermore, we propose a new form of point-based annotation for weakly-supervised panoptic segmentation. It only needs several random points for both things and stuff, which dramatically reduces the annotation cost of human. The proposed Panoptic FCN is also proved to have much superior performance in this weakly-supervised setting, which achieves 82% of the fully-supervised performance with only 20 randomly annotated points per instance. Extensive experiments demonstrate the effectiveness and efficiency of Panoptic FCN on COCO, VOC 2012, Cityscapes, and Mapillary Vistas datasets. And it sets up a new leading benchmark for both fully- and weakly-supervised panoptic segmentation.

Diabetes, Edentulism, and Cognitive Decline: A 12-Year Prospective Analysis

Diabetes mellitus (DM) is a recognized risk factor for dementia, and increasing evidence shows that tooth loss is associated with cognitive impairment and dementia. However, the effect of the co-occurrence of DM and edentulism on cognitive decline is understudied. This 12-y cohort study aimed to assess the effect of the co-occurrence of DM and edentulism on cognitive decline and examine whether the effect differs by age group. Data were drawn from the 2006 to 2018 Health and Retirement Study. The study sample included 5,440 older adults aged 65 to 74 y, 3,300 aged 75 to 84 y, and 1,208 aged 85 y or older. Linear mixed-effect regression was employed to model the rates of cognitive decline stratified by age cohorts. Compared with their counterparts with neither DM nor edentulism at baseline, older adults aged 65 to 74 y (β = -1.12; 95% confidence interval [CI], -1.56 to -0.65; P < 0.001) and those aged 75 to 84 y with both conditions (β = -1.35; 95% CI, -2.09 to -0.61; P < 0.001) had a worse cognitive function. For the rate of cognitive decline, compared to those with neither condition from the same age cohort, older adults aged 65 to 74 y with both conditions declined at a higher rate (β = -0.15; 95% CI, -0.20 to -0.10; P < 0.001). Having DM alone led to an accelerated cognitive decline in older adults aged 65 to 74 y (β = -0.09; 95% CI, -0.13 to -0.05; P < 0.001); having edentulism alone led to an accelerated decline in older adults aged 65 to 74 y (β = -0.13; 95% CI, -0.17 to -0.08; P < 0.001) and older adults aged 75 to 84 (β = -0.10; 95% CI, -0.17 to -0.03; P < 0.01). Our study finds the co-occurrence of DM and edentulism led to a worse cognitive function and a faster cognitive decline in older adults aged 65 to 74 y.

Comprehensive Dynamic Influence of Multiple Meteorological Factors on the Detection Rate of Bacterial Foodborne Diseases under Spatio-Temporal Heterogeneity

Foodborne diseases are a critical public health problem worldwide and significantly impact human health, economic losses, and social dynamics. Understanding the dynamic relationship between the detection rate of bacterial foodborne diseases and a variety of meteorological factors is crucial for predicting outbreaks of bacterial foodborne diseases. This study analyzed the spatio-temporal patterns of vibriosis in Zhejiang Province from 2014 to 2018 at regional and weekly scales, investigating the dynamic effects of various meteorological factors. Vibriosis had a significant temporal and spatial pattern of aggregation, and a high incidence period occurred in the summer seasons from June to August. The detection rate of Vibrio parahaemolyticus in foodborne diseases was relatively high in the eastern coastal areas and northwestern Zhejiang Plain. Meteorological factors had lagging effects on the detection rate of V. parahaemolyticus (3 weeks for temperature, 8 weeks for relative humidity, 8 weeks for precipitation, and 2 weeks for sunlight hours), and the lag period varied in different spatial agglomeration regions. Therefore, disease control departments should launch vibriosis prevention and response programs that are two to eight weeks in advance of the current climate characteristics at different spatio-temporal clustering regions.

The role and molecular mechanism of gut microbiota in Graves' orbitopathy

PURPOSE

Graves' orbitopathy (GO) is an autoimmune orbital disorder. Gut microbiota dysfunction plays a vital role in autoimmune diseases, including Graves' disease (GD) and GO. In the present study, we aimed to investigate the change of gut microbiota in GD/GO using mouse model.

METHODS

The murine model of GD/GO was established by the challenge of adenovirus expressing thyroid-stimulating hormone (TSH) receptor (TSHR) (Ad-TSHR). The histological changes of orbital and thyroid tissues were analyzed by hematoxylin and eosin (H&E), Masson staining, and immunohistochemistry (IHC) staining. The fecal samples were collected for 16S rRNA gene sequencing and bioinformatics analysis.

RESULTS

The GD/GO model was established successfully, as manifested as the broadened eyelid, exophthalmia and conjunctive redness, severe inflammatory infiltration among thyroid glands and between extraocular muscle space, hypertrophic extraocular muscles, elevated thyroxine (T4) and decreased TSH, and positive CD34, CD40, collagen I, and α-SMA staining. A total of 222 operational taxonomic units (OUTs) were overlapped between mice in the Ad-NC and Ad-TSHR groups. The microbial composition of the samples in the two groups was mainly Bacteroidia and Clostridia, and the Ad-NC group had a significantly lower content of Bacteroidia and higher content of Clostridia. KEGG orthology analysis results revealed differences in dehydrogenase, aspartic acid, bile acid, chalcone synthase, acetyltransferase, glutamylcyclotransferase, glycogenin, and 1-phosphatidylinositol-4-phosphate 5-kinase between two groups; enzyme commission (EC) analysis results revealed differences in several dehydrogenase, oxidase, thioxy/reductase between two groups; MetaCyc pathways analysis results revealed differences in isoleucine degradation, oxidation of C1 compounds, tricarboxylic acid (TCA) cycle IV, taurine degradation, and biosynthesis of paromamine, heme, colonic acid building blocks, butanediol, lysine/threonine/methionine, and histidine/purine/pyrimidine between two groups.

CONCLUSION

This study induced a mouse model of GD/GO by Ad-TSHR challenge, and gut microbiota characteristics were identified in the GD/GO mice. The Bacteroidia and Clostridia abundance was changed in the GD/GO mice. These findings may lay a solid experimental foundation for developing personalized treatment regimens for GD patients according to the individual gut microbiota. Given the potential impact of regional differences on intestinal microbiota, this study in China may provide a reference for the global overview of the gut-thyroid axis hypothesis.

Epidemiology of foodborne diseases caused by Salmonella in Zhejiang Province, China, between 2010 and 2021

Objective

Salmonella infection is a common cause of bacterial foodborne diseases (FBDs) globally. In this study, we aimed to explore the epidemiological and etiological characteristics of Salmonella infection from 2012-2021 in Zhejiang Province, China.

Methods

Descriptive statistical methods were used to analyze the data reported by the Centers for Disease Control and Prevention at all levels in Zhejiang Province through the China National Foodborne Diseases Surveillance Network from 2012-2021.

Results

A total of 11,269 Salmonella cases were reported, with an average positive rate of 3.65%, including 1,614 hospitalizations. A significant seasonal trend was observed for Salmonella cases, with the highest rate over the summer period, peaking from May to October, accounting for 77.96%. The results indicated a higher positive rate among respondents aged 0-4 years, especially for the scattered children (P < 0.05). The highest number of Salmonella infections were caused due to contaminated fruit and fruit products. Households (54.69%) had the most common exposure settings. Serotypes analysis revealed that Salmonella typhimurium (36.07%), Salmonella enteritidis (15.17%), and Salmonella london (6.05%) were the dominant strains among the 173 serotypes. Diarrhea, abdominal pain, fever, nausea, and vomiting were the main symptoms of these serotypes.

Conclusions

FBDs caused by Salmonella are important issues for public health in Zhejiang Province, and there is a need to focus on the epidemiological and etiological characteristics to control Salmonella infections.

Prenatal perfluoroalkyl substances exposure and neurodevelopment in toddlers: Findings from SMBCS

BACKGROUND

Prenatal perfluoroalkyl substances (PFAS) exposure has been reported to affect offspring neurodevelopment, while epidemiological evidences were limited and inconsistent.

OBJECTIVES

We aimed to evaluate the associations between cord serum PFAS concentrations and neurodevelopment in toddlers from 1 to 3 years of age.

METHODS

A total of 716 children from Sheyang Mini Birth Cohort Study (SMBCS) were included in this study. 12 PFAS concentrations were quantified in cord serum. Neurodevelopment was assessed using the Developmental Screen Test for Children Aged 0-6 Years at 1 year and the Gesell Developmental Schedules (GDS) at 2 and 3 years, respectively. Development quotient (DQ) z-score was standardized from DQ to eliminate the difference caused by two methods. We used generalized linear model (GLM) and Bayesian kernel machine regression (BKMR) to explore the associations of single or mixture PFAS exposure with neurodevelopment measurements at each time point. Associations between PFAS exposure and longitudinal changes in DQ z-score were investigated through generalized estimating equation (GEE) and trajectory analysis.

RESULTS

In general, prenatal PFAS concentrations showed negative associations with neurodevelopment measurements at specific age. When accounting for longitudinal changes from 1 to 3 years of age, PFOA was negatively associated with DQ z-score (β = -0.212, 95% CI: -0.422, -0.003), the association was only found significant in boys after stratified by gender (β = -0.327, 95% CI: -0.616, -0.038). Meanwhile, increased PFBS (OR = 2.159, 95% CI: 1.177, 3.959) and PFHpA (OR = 1.700, 95% CI: 1.016, 2.846) exposure was associated with elevated odds for the low-score trajectory group. The results of mixture of PFAS further confirmed above findings.

CONCLUSIONS

Our findings suggested that prenatal PFAS exposure may be associated with adverse neurodevelopment effects in the first 3 years of life. Further studies are warranted to confirm our findings.

The Liver Tumor Segmentation Benchmark (LiTS)

In this work, we report the set-up and results of the Liver Tumor Segmentation Benchmark (LiTS), which was organized in conjunction with the IEEE International Symposium on Biomedical Imaging (ISBI) 2017 and the International Conferences on Medical Image Computing and Computer-Assisted Intervention (MICCAI) 2017 and 2018. The image dataset is diverse and contains primary and secondary tumors with varied sizes and appearances with various lesion-to-background levels (hyper-/hypo-dense), created in collaboration with seven hospitals and research institutions. Seventy-five submitted liver and liver tumor segmentation algorithms were trained on a set of 131 computed tomography (CT) volumes and were tested on 70 unseen test images acquired from different patients. We found that not a single algorithm performed best for both liver and liver tumors in the three events. The best liver segmentation algorithm achieved a Dice score of 0.963, whereas, for tumor segmentation, the best algorithms achieved Dices scores of 0.674 (ISBI 2017), 0.702 (MICCAI 2017), and 0.739 (MICCAI 2018). Retrospectively, we performed additional analysis on liver tumor detection and revealed that not all top-performing segmentation algorithms worked well for tumor detection. The best liver tumor detection method achieved a lesion-wise recall of 0.458 (ISBI 2017), 0.515 (MICCAI 2017), and 0.554 (MICCAI 2018), indicating the need for further research. LiTS remains an active benchmark and resource for research, e.g., contributing the liver-related segmentation tasks in http://medicaldecathlon.com/. In addition, both data and online evaluation are accessible via https://competitions.codalab.org/competitions/17094.

Associations of perfluoroalkyl substances with adipocytokines in umbilical cord serum: A mixtures approach

BACKGROUND

Per- and polyfluoroalkyl substances (PFAS), a kind of emerging environmental endocrine disruptors, may interfere with the secretion of adipokines and affect fetal metabolic function and intrauterine development. However, the epidemiological evidence is limited and inconsistent. We examined the associations of single and multiple PFAS exposures in utero with adipocytokine concentrations in umbilical cord serum.

METHODS

This study included 1111 mother-infant pairs from Sheyang Mini Birth Cohort Study (SMBCS), and quantified 12 PFAS and two adipokine in umbilical cord serum. Generalized linear models (GLMs) and Bayesian Kernel Machine Regression (BKMR) models were applied to estimate the associations of single- and mixed- PFAS exposure with adipokines, respectively. Furthermore, sex-stratification was done in each model to assess the sexually dimorphic effects of PFAS.

RESULTS

10 PFAS were detected with median concentrations (μg/L) ranging from 0.04 to 3.97, (except 2.7% for PFOSA and 1.7% for PFDS, which were excluded). In GLMs, for each doubling increase in PFBS, PFHpA, PFHxS, PFHpS, PFUnDA and PFDoDA, leptin decreased between 14.04% for PFBS and 22.69% for PFHpS (P < 0.05). PFAS, except for PFNA, were positively associated with adiponectin, and for each doubling of PFAS, adiponectin increased between 3.27% for PFBS and 12.28% for PFHxS (P < 0.05). In addition, infant gender modified the associations of PFAS with adipokines, especially the associations of PFBS, PFOA and PFHxS with adiponectin. Similarly, significant associations of PFAS mixtures with leptin and adiponectin were observed in the BKMR models. PFDA, PFOS, PFNA and PFHpS were identified as important contributors. In the sex-stratified analysis of BKMR models, the associations between PFAS mixtures and adipokines were more pronounced in males.

CONCLUSIONS

PFAS levels were significantly associated with adipokines in cord serum, suggesting that intrauterine mixture of PFAS exposure may be related to decreased fetal leptin level but increased fetal adiponectin level and the associations may be sex-specific.

Multiple mediation effects on association between prenatal triclosan exposure and birth outcomes

BACKGROUND

Triclosan is a broad-spectrum antimicrobial, and was thought to affect intrauterine development, but the mechanism remains unclear.

OBJECTIVE

To explore the association between prenatal triclosan exposure and birth outcomes.

METHODS

Based on 726 mother-child pairs from the Sheyang Mini Birth Cohort Study (SMBCS), we used the available (published) data of triclosan in maternal urines, the hormones including thyroid-related hormones, gonadal hormones in cord blood, and adipokines, trimethylamine-N-oxide (TMAO) and its precursors in cord blood to explore possible health effects of triclosan on birth outcomes through assessing different hormones and parameters, using Bayesian mediation analysis.

RESULTS

Maternal triclosan exposure was associated with ponderal index (β = 0.317) and head circumference (β = -0.172) in generalized linear models. In Bayesian mediation analysis of PI model, estradiol (β = 0.806) and trimethylamine (TMA, β = 0.164) showed positive mediation effects, while total thyroxine (TT4, β = -0.302), leptin (β = -2.023) and TMAO (β = -0.110) showed negative mediation effects. As for model of head circumference, positive mediation effects were observed in free thyroxine (FT4, β = 0.493), TMA (β = 0.178), and TMAO (β = 0.683), negative mediation effects were observed in TT4 (β = -0.231), testosterone (β = -0.331), estradiol (β = -1.153), leptin (β = -2.361), choline (β = -0.169), betaine (β = -0.104), acetyl-L-carnitine (β = -0.773).

CONCLUSION

The results indicated triclosan can affect intrauterine growth by interfering thyroid-related hormones, gonadal hormones, adipokines, TMAO and its precursors.

Urinary para-nitrophenol levels of pregnant women and cognitive and motor function of their children aged 2 years: Evidence from the SMBCS (China)

BACKGROUND

Urinary para-nitrophenol (PNP), an exposure biomarker of ethyl parathion (EP) and methyl parathion (MP) pesticides, was still pervasively detected in the general population even after global restriction for years. And the concern whether there is an association of PNP level with child development of the nervous system is increasing. The current study aimed to evaluate the maternal urinary PNP concentrations during late pregnancy and the associations of PNP levels with cognitive and motor function of their children at the age of 2 years.

METHODS

323 mother-child pairs from the Sheyang Mini Birth Cohort Study were included in the current study. Gas chromatography-tandem mass spectrometry was used to measure concentrations of PNP, the specific metabolite of EP and MP, in maternal urine samples during pregnancy. Developmental quotients (DQs) scores measured with Gesell Developmental Scales were employed to evaluate cognitive and motor function of children aged 2 years. Generalized linear models were performed to analyze the associations of PNP concentrations in pregnant women's urine samples with cognitive and motor function of their children.

RESULTS

Maternal PNP was detected in all urine samples with a median of 4.11 μg/L and a range from 0.57 μg/L to 109.13 μg/L, respectively. Maternal urinary PNP concentrations showed a negative trend with DQ of motor area [regression coefficient (β) = - 1.35; 95 % confidence interval (95 %CI): - 2.37, - 0.33; P < 0.01], and the children whose mothers were in the fourth quartile exposure group performed significantly worse compared to the reference group (β = - 1.11; 95 %CI: - 1.80, - 0.42; P < 0.01). As for average DQ score, children with their mothers' urinary PNP concentrations in the third quartile group had higher scores than those in the first quartile group (β = 0.39; 95 %CI: 0.03, 0.75; P = 0.04). In sex-stratified analyses, a negative trend between maternal urinary PNP concentrations and DQ scores in motor area of children was only observed in boys (β = - 1.62; 95 %CI: - 2.80, - 0.43; P < 0.01). Boys in the third quartile group had higher DQ average scores than those in the lowest quartile as reference (β = 0.53; 95 %CI: 0.02, 1.04; P = 0.04).

CONCLUSIONS

The mothers from SMBCS may be widely exposed to EP and/or MP, which were associated with the cognitive and motor function of their children aged 2 years in a sex-specific manner. Our results might provide epidemiology evidence on the potential effects of prenatal exposure to EP and/or MP on children's cognitive and motor function.