Joint MRI T1 Unenhancing and Contrast-enhancing Multiple Sclerosis Lesion Segmentation with Deep Learning in OPERA Trials.

Background Deep learning-based segmentation could facilitate rapid and reproducible T1 lesion load assessments, which is crucial for disease management in multiple sclerosis (MS). T1 unenhancing and contrast-enhancing lesions in MS are those that enhance or do not enhance after administration of a gadolinium-based contrast agent at T1-weighted MRI. Purpose To develop deep learning models for automated assessment of T1 unenhancing and contrast-enhancing lesions; to investigate if joint training improved performance; to reproduce a known ocrelizumab treatment response; and to evaluate the association of baseline T1-weighted imaging metrics with clinical outcomes in relapsing MS clinical trials. Materials and Methods Joint and individual deep learning models (U-Nets) were developed retrospectively on multimodal MRI data sets from large multicenter OPERA trials of relapsing MS (August 2011 to May 2015). The joint model included cross-network connections and a combined loss function. Models were trained on OPERA I data sets with three-fold cross-validation. OPERA II data sets were the internal test set. Dice coefficients, lesion true-positive and false-positive rates, and areas under the receiver operating characteristic curve (AUCs) were used to evaluate model performance. Association of baseline imaging metrics with clinical outcomes was assessed with Cox proportional hazards models. Results A total of 796 patients (3030 visits; mean age, 37 years ± 9; 521 women) from the OPERA II trial were evaluated. The joint model achieved a mean Dice coefficient of 0.77 and 0.74, lesion true-positive rate of 0.88 and 0.86, and lesion false-positive rate of 0.04 and 0.19 for T1 contrast-enhancing and T1 unenhancing lesion segmentation, respectively. Joint training improved performance for smaller T1 contrast-enhancing lesions (≤0.06 mL; individual training AUC: 0.75; joint training AUC: 0.87; P < .001). A significant ocrelizumab treatment effect (P < .001) was seen in reducing the mean number of T1 contrast-enhancing lesions at 24, 48, and 96 weeks (manual assessment at 24 weeks: 10 lesions in 366 patients with ocrelizumab, 141 lesions in 355 patients with interferon, 93% reduction; manual assessment at 48 weeks: six lesions in 355 patients with ocrelizumab, 150 lesions in 317 patients with interferon, 96% reduction; manual assessment at 96 weeks: five lesions in 340 patients with ocrelizumab, 157 lesions in 294 patients with interferon, 97% reduction; joint model assessment at 24 weeks: 19 lesions in 365 patients with ocrelizumab, 128 lesions in 354 patients with interferon, 86% reduction; joint model assessment at 48 weeks: 14 lesions in 355 patients with ocrelizumab, 121 lesions in 317 patients with interferon, 90% reduction; joint model assessment at 96 weeks: 10 lesions in 340 patients with ocrelizumab, 144 lesions in 294 patients with interferon, 94% reduction) and the mean number of new T1 unenhancing lesions across all follow-up examinations (manual assessment: 504 lesions in 1060 visits for ocrelizumab-treated patients, 1438 lesions in 965 visits for interferon-treated patients, 68% reduction; joint model assessment: 205 lesions in 1053 visits for ocrelizumab-treated patients, 661 lesions in 957 visits for interferon-treated patients, 78% reduction). Baseline T1 unenhancing total lesion volume was associated with clinical outcomes (manual hazard ratio [HR]: 1.12, P = .02; joint model HR: 1.11, P = .03). Conclusion Joint architecture and training improved segmentation of MRI T1 contrast-enhancing multiple sclerosis lesions, and both deep learning models had sufficiently high performance to detect an ocrelizumab treatment response consistent with manual assessments. ClinicalTrials.gov: NCT01247324 and NCT01412333 © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Talbott in this issue.

Age-dependent amyloid deposition is associated with white matter alterations in cognitively normal adults during the adult life span.

INTRODUCTION: Both beta-amyloid (Ab) deposition and decline in white matter integrity, are brain alterations observed in Alzheimer's disease (AD) and start to occur by the fourth and fifth decades. However, the association between both brain alterations in asymptomatic subjects is unclear. METHODS: Amyloid positron emission tomography (PET) and diffusion tensor imaging (DTI) were obtained in 282 cognitively normal subjects (age 30-89 years). We assessed the interaction of age by abnormal amyloid PET status (Florbetapir F-18 PET >1.2 standard uptake value ratio [SUVR]) on regional mean diffusivity (MD) and global white matter hyperintensity (WMH) volume, controlled for sex, education, and hypertension. RESULTS: Subjects with abnormal amyloid PET (n = 87) showed stronger age-related increase in global WMH and regional MD, particularly within the posterior parietal regions of the white matter. DISCUSSION: Sporadic Aß deposition is associated with white matter alterations in AD predilection areas in an age-dependent manner in cognitively normal individuals.

Brain amyloid-ß burden is associated with disruption of intrinsic functional connectivity within the medial temporal lobe in cognitively normal elderly.

The medial temporal lobe is implicated as a key brain region involved in the pathogenesis of Alzheimer's disease (AD) and consequent memory loss. Tau tangle aggregation in this region may develop concurrently with cortical Aß deposition in preclinical AD, but the pathological relationship between tau and Aß remains unclear. We used task-free fMRI with a focus on the medical temporal lobe, together with Aß PET imaging, in cognitively normal elderly human participants. We found that cortical Aß load was related to disrupted intrinsic functional connectivity of the perirhinal cortex, which is typically the first brain region affected by tauopathies in AD. There was no concurrent association of cortical Aß load with cognitive performance or brain atrophy. These findings suggest that dysfunction in the medial temporal lobe may represent a very early sign of preclinical AD and may predict future memory loss.

A fast algorithm for denoising magnitude diffusion-weighted images with rank and edge constraints.

PURPOSE: To accelerate denoising of magnitude diffusion-weighted images subject to joint rank and edge constraints. METHODS: We extend a previously proposed majorize-minimize method for statistical estimation that involves noncentral χ distributions to incorporate joint rank and edge constraints. A new algorithm is derived which decomposes the constrained noncentral χ denoising problem into a series of constrained Gaussian denoising problems each of which is then solved using an efficient alternating minimization scheme. RESULTS: The performance of the proposed algorithm has been evaluated using both simulated and experimental data. Results from simulations based on ex vivo data show that the new algorithm achieves about a factor of 10 speed up over the original Quasi-Newton-based algorithm. This improvement in computational efficiency enabled denoising of large datasets containing many diffusion-encoding directions. The denoising performance of the new efficient algorithm is found to be comparable to or even better than that of the original slow algorithm. For an in vivo high-resolution Q-ball acquisition, comparison of fiber tracking results around hippocampus region before and after denoising will also be shown to demonstrate the denoising effects of the new algorithm. CONCLUSION: The optimization problem associated with denoising noncentral χ distributed diffusion-weighted images subject to joint rank and edge constraints can be solved efficiently using a majorize-minimize-based algorithm.

Stereoselective Synthesis of Arylglycine Derivatives via Palladium-Catalyzed α-Arylation of a Chiral Nickel(II) Glycinate.

A practical and efficient stereoselective synthesis of arylglycine derivatives was realized via palladium-catalyzed α-arylation of a chiral nickel(II) glycinate complex with aryl bromides. The structurally diverse arylglycine products were obtained in excellent isolated yields and with good diastereoselectivity. A simple acidic hydrolysis furnished optically pure arylglycines in high yield, and the chiral ligand (S)-BPB could be efficiently recovered and reused.

Impaired capacity for familiarity after hippocampal damage.

Recognition memory is thought to consist of two components: recollection and familiarity. Whereas it is widely agreed that the hippocampus supports recollection (remembering the episode in which an item was learned), there is uncertainty about whether it also supports familiarity (simply knowing that an item was encountered but without remembering the learning episode). We tested a counterintuitive prediction that follows from the idea that the hippocampus selectively supports recollection. Patients with hippocampal lesions should have strong experiences of familiarity as often as controls do; however, unlike controls, these experiences should not be accompanied by recollection. Accordingly, with methods that allow participants to report whether they remember an item as encountered previously or whether they simply know it is familiar, patients should express strong familiarity (in the absence of recollection) more often than controls. We indexed strong familiarity and recollection for previously studied words by obtaining confidence ratings together with Remember-Know judgments. The result was that patients provided fewer high-confidence Know responses than controls rather than more. Furthermore, the number of Know responses made by patients was substantially less than was predicted if recollection were impaired. This was true regardless of whether the prediction was based on the assumption that recollection and familiarity are independent or dependent processes. These results suggest that hippocampal lesions impair both recollection and familiarity. Unlike many previous studies of these constructs, the prediction (and the result) is independent of any particular theoretical model, and it holds even if Remember-Know judgments are not process-pure indicators of recollection and familiarity.

Different nonlinear functions in hippocampus and perirhinal cortex relating functional MRI activity to memory strength.

Findings from functional MRI (fMRI) studies of recognition memory and the medial temporal lobe often suggest qualitative differences in the contribution of the hippocampus and perirhinal cortex. This interpretation is complicated by the fact that most of the methods intended to demonstrate qualitative differences also separate strong memories from weak memories. Thus, apparent qualitative differences might reflect quantitative differences in how measured activity in medial temporal lobe structures varies with memory strength. We tested the hypothesis that the relationship between activity at the time of study and subsequent memory strength is nonlinear in hippocampus and perirhinal cortex and also distinctly different in those two structures. We found that activity in the hippocampus was characterized by a positively accelerated function and that activity in the perirhinal cortex was associated with a statistically different, negatively accelerated function. Our results do not count against the possibility that these structures differ qualitatively in their contributions to memory. Rather, our findings show how an alternative interpretation based on quantitative differences can also account for a good deal of data, and they suggest that a demonstration of qualitative differences requires more stringent criteria than are achieved in most fMRI studies.

Measles Virus-Based Vaccine Expressing Membrane-Anchored Spike of SARS-CoV-2 Inducing Efficacious Systemic and Mucosal Humoral Immunity in Hamsters.

As SARS-CoV-2 continues to evolve and COVID-19 cases rapidly increase among children and adults, there is an urgent need for a safe and effective vaccine that can elicit systemic and mucosal humoral immunity to limit the emergence of new variants. Using the Chinese Hu191 measles virus (MeV-hu191) vaccine strain as a backbone, we developed MeV chimeras stably expressing the prefusion forms of either membrane-anchored, full-length spike (rMeV-preFS), or its soluble secreted spike trimers with the help of the SP-D trimerization tag (rMeV-S+SPD) of SARS-CoV-2 Omicron BA.2. The two vaccine candidates were administrated in golden Syrian hamsters through the intranasal or subcutaneous routes to determine the optimal immunization route for challenge. The intranasal delivery of rMeV-S+SPD induced a more robust mucosal IgA antibody response than the subcutaneous route. The mucosal IgA antibody induced by rMeV-preFS through the intranasal routine was slightly higher than the subcutaneous route, but there was no significant difference. The rMeV-preFS vaccine stimulated higher mucosal IgA than the rMeV-S+SPD vaccine through intranasal or subcutaneous administration. In hamsters, intranasal administration of the rMeV-preFS vaccine elicited high levels of NAbs, protecting against the SARS-CoV-2 Omicron BA.2 variant challenge by reducing virus loads and diminishing pathological changes in vaccinated animals. Encouragingly, sera collected from the rMeV-preFS group consistently showed robust and significantly high neutralizing titers against the latest variant XBB.1.16. These data suggest that rMeV-preFS is a highly promising COVID-19 candidate vaccine that has great potential to be developed into bivalent vaccines (MeV/SARS-CoV-2).

Multi-arm U-Net with dense input and skip connectivity for T2 lesion segmentation in clinical trials of multiple sclerosis.

T2 lesion quantification plays a crucial role in monitoring disease progression and evaluating treatment response in multiple sclerosis (MS). We developed a 3D, multi-arm U-Net for T2 lesion segmentation, which was trained on a large, multicenter clinical trial dataset of relapsing MS. We investigated its generalization to other relapsing and primary progressive MS clinical trial datasets, and to an external dataset from the MICCAI 2016 MS lesion segmentation challenge. Additionally, we assessed the model's ability to reproduce the separation of T2 lesion volumes between treatment and control arms; and the association of baseline T2 lesion volumes with clinical disability scores compared with manual lesion annotations. The trained model achieved a mean dice coefficient of ≥ 0.66 and a lesion detection sensitivity of ≥ 0.72 across the internal test datasets. On the external test dataset, the model achieved a mean dice coefficient of 0.62, which is comparable to 0.59 from the best model in the challenge, and a lesion detection sensitivity of 0.68. Lesion detection performance was reduced for smaller lesions (≤ 30 µL, 3-10 voxels). The model successfully maintained the separation of the longitudinal changes in T2 lesion volumes between the treatment and control arms. Such tools could facilitate semi-automated MS lesion quantification; and reduce rater burden in clinical trials.

Medial temporal lobe function and recognition memory: a novel approach to separating the contribution of recollection and familiarity.

Human neuroimaging studies of recognition memory have often been interpreted to mean that the hippocampus supports recollection but not familiarity. This interpretation is complicated by the fact that recollection-based decisions are typically associated with stronger memories than familiarity-based decisions. Some studies of source memory controlled for this difference in memory strength and found that hippocampal activity during learning predicted subsequent item memory strength while recollection-based memory (performance on source memory questions) was held at chance. This result suggests that the hippocampus is important for familiarity. However, a difficulty with this approach is that when source memory is assessed by asking specific, task-relevant source memory questions, participants who fail to answer the prescribed questions might nevertheless have available other (task-irrelevant) source information. Accordingly, successful item memory could still be associated with recollection. The present study used a novel method to assess item memory and source memory. Instead of responding to specific source questions, participants rated their source memory strength based on any information about the learning episode that was available to them. When subsequent source memory strength was held constant at the lowest possible level, we identified regions bilaterally in hippocampus, as well as in perirhinal cortex, where activity during learning increased as subsequent item memory increased in strength. In addition, activity in cortical regions (including prefrontal cortex) was related to source memory success independently of item memory strength. These findings suggest that activity in the hippocampus is related to the encoding of familiarity-based item memory, independent of subsequent recollection-based success.

Polishing Characteristics of Cemented Carbide Using Cubic Boron Nitride Magnetic Abrasive Powders.

This paper describes the application of bonded magnetic abrasive powders (MAPs) in the magnetic abrasive finishing (MAF) process. In order to improve the poor finishing performance and short service life of MAPs in polishing super-hard materials, a double-stage atomization technique was used to successfully manufacture MAPs with a CBN as an abrasive phase. The prepared results show that CBN abrasives with their original structure were deeply and densely embedded on the surface of spherical MAPs. Based on the MAF process, a five-level and four-factor central composite design experiment was carried out to verify the developed MAPs polishing performance on the finishing of cemented carbide parts (864 Hv). Working gap, rotational speed, feed rate of a workpiece, and mesh number of MAP were considered as influence factors. The analysis data was used to understand different interactions of significant parameters. A regression model for predicting the change of surface roughness was obtained, and the optimal parameter combination was figured out through a solution of a quadratic equation in Design-Expert software. According to MAF results, the strong cutting ability of atomized CBN MAPs improved the surface roughness of cemented carbide by over 80% at the optimum parameters. The strong cutting ability of atomized CBN MAPs can produce good surface quality on the hard materials. The findings of this research can promote a large-scale application of MAF technology in the surface polishing of hard materials.

Deformation-based morphometry identifies deep brain structures protected by ocrelizumab.

BACKGROUND: Despite advancements in treatments for multiple sclerosis, insidious disease progression remains an area of unmet medical need, for which atrophy-based biomarkers may help better characterize the progressive biology. METHODS: We developed and applied a method of longitudinal deformation-based morphometry to provide voxel-level assessments of brain volume changes and identified brain regions that were significantly impacted by disease-modifying therapy. RESULTS: Using brain MRI data from two identically designed pivotal trials of relapsing multiple sclerosis (total N = 1483), we identified multiple deep brain regions, including the thalamus and brainstem, where volume loss over time was reduced by ocrelizumab (p < 0.05), a humanized anti-CD20 + monoclonal antibody approved for the treatment of multiple sclerosis. Additionally, identified brainstem shrinkage, as well as brain ventricle expansion, was associated with a greater risk for confirmed disability progression (p < 0.05). CONCLUSIONS: The identification of deep brain structures has a strong implication for developing new biomarkers of brain atrophy reduction to advance drug development for multiple sclerosis, which has an increasing focus on targeting the progressive biology.

Experimental Investigation on Process Parameters during Laser-Assisted Turning of SiC Ceramics Based on Orthogonal Method and Response Surface Methodology.

In this study, laser-assisted machining experiments are carried out on silicon carbide (SiC) ceramic materials by a turning process, and laser power, cutting depth, rotational speed, and feed speed are selected as research factors. In order to improve the surface processing quality of laser-assisted turning of SiC ceramics and obtain the smallest surface roughness, the orthogonal method and response surface method are used to investigate the effect of various factors on surface roughness. The effect of various factors on surface roughness is evaluated by variance analysis, mean analysis, main effect diagram, 3D response surface, and corresponding contour diagram. The surface roughness prediction model is established based on the response surface method, and the prediction error is 4.1% with high accuracy. The experimental results show that laser power and cutting depth are the most significant factors affecting surface roughness, and rotational speed is a significant factor affecting surface roughness. Under the optimum process conditions, the smallest surface roughness Ra obtained by the response surface method is 0.294 µm, which is lower than 0.315 µm obtained by the orthogonal method, and the surface quality is higher. Therefore, the optimal process parameters of the response surface method can obtain the smallest surface roughness and higher surface quality in laser-assisted turning of SiC ceramics.

Study on the Micro Removal Process of Inner Surface of Cobalt Chromium Alloy Cardiovascular Stent Tubes.

Due to the special manufacturing process of cobalt-chromium alloy cardiovascular stent tubes, there are serious surface defects in their inner walls, which affects the therapeutic effect after implantation. At the same time, the traditional processing technology cannot finish the inner wall of a cardiovascular stent tube. In light of the above problems, magnetic abrasive finishing (MAF) equipment for the inner wall of an ultra-fine and ultra-long cardiovascular stent tube is proposed, and MAF technology is used to improve the surface quality of its inner wall. High-performance spherical magnetic abrasive powders are used to finish the inner wall of a cobalt-chromium alloy cardiovascular stent tube with an inner diameter of 1.6 mm and an outer diameter of 1.8 mm. The effects of finishing time, tube rotational speed, feed speed of the magnetic pole, MAPs filling quantity, and MAP abrasive size on the surface roughness and material removal thickness of cobalt-chromium alloy cardiovascular stent tube are investigated. The results show that the surface roughness of the inner wall of the cobalt-chromium alloy cardiovascular stent decreases from 0.485 µm to 0.101 µm, and the material removal thickness of the defect layer is 4.3 µm. MAF technology is used to solve the problem of the poor surface quality of the inner walls of ultra-fine and ultra-long cobalt-chromium alloy cardiovascular stent tubes.

Prediction and Optimization of Surface Roughness for Laser-Assisted Machining SiC Ceramics Based on Improved Support Vector Regression.

In this paper, the surface roughness of SiC ceramics was investigated in laser-assisted machining (LAM) processes; machine learning was used to predict surface roughness and to optimize the process parameters, and therefore, to ultimately improve the surface quality of a workpiece and obtain excellent serviceability. First, single-factor turning experiments were carried out on SiC ceramics using LAM according to the material removal mechanism to investigate the variation trend of the effects of different laser powers, rotational speeds, feed rates, and cutting depths on surface roughness. Then, laser power, rotational speed, feed rate and cutting depth were selected as the four factors, and the surface roughness was used as the target value for the orthogonal experiments. The results of the single-factor experiments and the orthogonal experiments were combined to construct a prediction model based on the combination of the grey wolf optimization (GWO) algorithm and support vector regression (SVR). The coefficient of determination (R2) of the optimized prediction model reached 0.98676 with an average relative error of less than 2.624%. Finally, the GWO algorithm was used to optimize the global parameters of the prediction model again, and the optimal combination of process parameters was determined and verified by experiments. The actual minimum surface roughness (Ra) value was 0.418 µm, and the relative error was less than 1.91% as compared with the predicted value of the model. Therefore, the prediction model based on GWO-SVR can achieve accurate prediction of the surface roughness of SiC ceramics in LAM and can obtain the optimum surface roughness using parameter optimization.

Experimental Research and Optimization of Ti-6Al-4V Alloy Microgroove Machining Based on Waterjet-Guided High-Power Laser.

In order to improve the tribological properties of Ti-6Al-4V alloy and further broaden the application scope of titanium alloy materials in the industrial field, a preparation method of a waterjet-guided high-power laser processing surface microgroove was studied. In this paper, a multifocus coupling lens was innovatively designed to replace the spherical lens in the traditional waterjet-guided laser coupling device, which avoids the gas explosion phenomenon in the coupling of the high-power laser and waterjet, and realizes the high-quality coupling of the high-power laser and water beam fiber. Then, with the microgroove morphology as the response target, the single-factor test and response surface test of the water-guided laser processing microgroove were carried out. Based on the experimental results, an approximate mathematical model of the response surface between the process parameters and the microgroove topography target was constructed, and the quantitative relationship between the waterjet-guided laser processing parameters and the target response was studied. At the same time, the optimal combination of process parameters was obtained by multiobjective optimization, so as to effectively improve the microgroove morphology. This technology provides method guidance and a decision-making reference for subsequent waterjet-guided laser processing of titanium alloy surface functional microstructures.

Study on screening of fermentation agents and optimization of the fermentation process for pharyngitis tablet residue.

This study aimed to screen an appropriate starter and determine the optimal fermentation process parameters to optimize the fermentation process of nutrient components and bioactive contents in pharyngitis tablet residue. This study included two experiments. In experiment I, single-factor experimental design was used to study the effects of different biological starters (cellulase preparation; Lactobacillus Plantarum and Bacillus subtilis preparation; mixture of cellulase + Lactobacillus Plantarum and Bacillus subtilis) on the nutritional values and bioactive ingredient contents in pharyngitis tablet residue. In experiment II, orthogonal design experiment was adopted to study the effects of initial water content (45, 50, and 55%), fermentation temperature (35, 37, and 40°C), and fermentation time (24, 48, and 72 h) on the changes in nutrient components, biologically activity contents, and toxin contents of the residue after optimal fermentation agent treatment. Cellulase preparation was found to be the optimal starter. The optimal fermentation conditions were: initial water content, 55%; fermentation temperature, 37°C; and fermentation time, 72 h. The contents of aflatoxin B1 and vomit toxin were in line with Chinese feed hygiene production standards. The fermentation quality of pharyngitis tablet residue can be improved by using the optimal starter and fermentation conditions.

Cerebral cortical folding analysis with multivariate modeling and testing: Studies on gender differences and neonatal development.

This paper presents a novel statistical framework for human cortical folding pattern analysis that relies on a rich multivariate descriptor of folding patterns in a region of interest (ROI). The ROI-based approach avoids problems faced by spatial normalization-based approaches stemming from the deficiency of homologous features between typical human cerebral cortices. Unlike typical ROI-based methods that summarize folding by a single number, the proposed descriptor unifies multiple characteristics of surface geometry in a high-dimensional space (hundreds/thousands of dimensions). In this way, the proposed framework couples the reliability of ROI-based analysis with the richness of the novel cortical folding pattern descriptor. This paper presents new mathematical insights into the relationship of cortical complexity with intra-cranial volume (ICV). It shows that conventional complexity descriptors implicitly handle ICV differences in different ways, thereby lending different meanings to "complexity". The paper proposes a new application of a nonparametric permutation-based approach for rigorous statistical hypothesis testing with multivariate cortical descriptors. The paper presents two cross-sectional studies applying the proposed framework to study folding differences between genders and in neonates with complex congenital heart disease. Both studies lead to novel interesting results.

Outbreak of neonatal gastroenteritis associated with astrovirus serotype 1 at a hospital in Inner Mongolia, China.

This report describes for the first time an outbreak of acute gastroenteritis among neonates associated with human astrovirus (HAstV) serotype 1b at a maternity hospital in Inner Mongolia, China. Of 40 specimens, 28 were astrovirus positive and rotavirus, calicivirus, and adenovirus negative. Poor hygiene likely contributed to the spread and persistence of HAstV in the neonatal care room.

[Start-up of Simultaneous ANAMMOX and Denitrification Process and Changes in Microbial Community Characteristics].

To understand the relationship between nitrogen and carbon removal performance and the microbial community during start-up of simultaneous ANAMMOX and denitrification (SAD), nitrogen and carbon removal performance and microbial community change were studied by gradually increasing the influent COD concentration. The results showed that with the increase of the influent COD concentration, NH4+-N and NO2--N effluent remained stable, and the average removal rate was more than 98%. The removal rate of TN increased gradually, and the rate was 95.6% in the third stage, which was 6.8% higher than that of ANAMMOX in theory. ΔNO3--N/ΔNH4+-N decreased significantly from 0.15-0.17 to 0.03-0.07. The contribution rate of ANAMMOX to nitrogen removal decreased gradually, denitrification for nitrogen removal increased gradually, and the COD removal rate increased. Sludge activity analysis showed that the denitrification activity of sludge increased significantly and the ANAMMOX activity decreased slightly after the start-up of SAD. High throughput sequencing results showed that the dominant phyla of microorganisms in the reactor were Chloroflexi, Planctomycetes, Firmicutes, Armatimonadetes, and Proteobacteria. The characteristics of the microbial community in the reactor were closely related to the performance of SAD in nitrogen and carbon removal. The main functional microorganisms related to nitrogen and carbon removal were ANAMMOX bacteria, anaerobic digestive bacteria, and denitrifying bacteria. The abundance of ANAMMOX bacteria decreased after the start-up of SAD, and the anaerobic digestive bacteria and heterotrophic denitrifying bacteria increased significantly.