Search

Recently, algorithm-hardware (HW) co-exploration for neural networks (NNs) has become the key to obtaining high-quality solutions. However, previous efforts for field-programmable gate arrays (FPGAs) focus on neural architecture search (NAS) while lacking HW architecture search (HAS), thus limiting the full potential of co-design. Although expanding the scope of HAS offers performance potential, the exponentially increased joint search space presents a formidable challenge. To address this, we propose a deep and efficient framework NAF, which jointly searches for Networks and Accelerators for FPGAs in a balanced co-search space. First, we adjust the NAS space and then introduce a block-level bitwidth search on the software side. Meanwhile, we design a HW-friendly quantization algorithm to facilitate HW efficiency and accuracy. Second, we design a dataflow-configurable HW unit with computation and memory access optimizations for quantized multiplication. Based on this, we incorporate critical heterogeneous multicore architecture exploration on the HW side. Third, to enable rapid HW feedback in the enlarged HAS space, we perform resource and performance modeling and design a fast HW generation algorithm based on the genetic algorithm. Specifically, we apply optimization techniques, like mapping space pruning, greedy bandwidth allocation, and coarse-grained search, to speed up this process. We validate NAF in edge and cloud scenarios. Experimental results show that NAF efficiently explores a significantly larger joint space and provides high-quality solutions. Compared with previous state-of-the-art co-design works, the searched convolutional neural network-accelerator pairs improve the throughput by $2.07\times \sim 7.10\times $ and energy efficiency by $1.41\times \sim 2.27\times $ under similar accuracy on the ImageNet dataset.

To investigate the current situation of mental psychology and quality of life (QoL) in patients with inflammatory bowel disease (IBD) in China, and analyze the influencing factors. A unified questionnaire was developed to collect clinical data on IBD patients from 42 hospitals in 22 provinces from September 2021 to May 2022. Multivariate Logistic regression analysis was conducted, and independent influencing factors were screened out to construct nomogram. The consistency index (C‐index), receiver operating characteristic (ROC) curve, area under the ROC curve (AUC), calibration curve, and decision curve analysis (DCA) were used to evaluate the discrimination, accuracy, and clinical utility of the nomogram model. A total of 2478 IBD patients were surveyed, including 1371 patients with ulcerative colitis (UC) and 1107 patients with Crohn's disease (CD). Among them, 25.5%, 29.7%, 60.2%, and 37.7% of IBD patients had anxiety, depression, sleep disturbance and poor QoL, respectively. The proportion of anxiety, depression, and poor QoL in UC patients was significantly higher than that in CD patients (all p< 0.05), but there was no difference in sleep disturbance between them (p= 0.737). Female, higher disease activity and the first visit were independent risk factors for anxiety, depression and sleep disturbance in IBD patients (all p< 0.05). The first visit, higher disease activity, abdominal pain and diarrhea symptoms, anxiety, depression and sleep disturbance were independent risk factors for the poor QoL of patients (all p< 0.05). The AUC value of the nomogram prediction model for predicting poor QoL was 0.773 (95% CI: 0.754–0.792). The calibration diagram of the model showed that the calibration curve fit well with the ideal curve, and DCA showed that the nomogram model could bring clinical benefits. IBD patients have higher anxiety, depression, and sleep disturbance, which affect their QoL. The nomogram prediction model we constructed has high accuracy and performance when predicting QoL.

Vitamin D (VitD) is potentially immunomodulatory, so here we aimed to explore the relationships between serum VitD levels, immune checkpoint inhibitor (ICI) efficacy, and immune-related adverse events (irAEs). Serum 25-hydroxyvitamin D [25(OH)D] levels were quantified before and after ICI treatment in prospectively enrolled patients with advanced lung cancers. Of 77 enrolled patients, 29 developed 42 irAEs. Baseline 25(OH)D levels of partial response (PRs) patients were significantly higher than non-PR patients (19.39±7.16 vs. 16.28±5.99 ng/mL, P=0.04). The area under the curve of 25(OH)D >15.73 ng/mL to identify PR was 0.63 (95% CI, 0.51–0.76, P=0.047), and baseline 25(OH)D levels >15.73 ng/mL (odds ratio: 2.93, 95% CI, 1.10–7.79, P=0.03) and prior targeted therapy (odds ratio: 0.30, 95% CI, 0.10–0.92, P=0.04) were independent predictors of PR as best efficacy by multivariable logistic regression. With respect to irAEs, baseline 25(OH)D levels were higher in grade 1 irAE patients than in grade 2/3/4 irAE patients (20.07±8.64 vs. 15.22±2.30 ng/mL, P=0.02). However, the area under the curve was only 0.56 (95% CI, 0.42–0.70, P=0.39) for a baseline 25(OH)D of 20.99 ng/mL for predicting irAE occurrence. There was a direct monotonic relationship and U-shaped relationship between baseline 25(OH)D levels and ICI efficacy and irAE occurrence, respectively. Overall survival was significantly different between VitD sufficient, insufficient, and deficient patients (log-rank P=0.01), which remained after adjustment in Cox proportional hazards regression models. Baseline 25(OH)D levels seem to be associated with ICI efficacy and prognosis, it might be helpful to assess the baseline VitD status, and supplementation with VitD might bring some benefit to enhance ICI efficacy and reduce moderate-severe irAEs.

Recovering the high-resolution three-dimensional (3D) surface of an object from a single frame image has been the ultimate goal long pursued in fringe projection profilometry (FPP). The color fringe projection method is one of the technologies with the most potential towards such a goal due to its three-channel multiplexing properties. However, the associated color imbalance, crosstalk problems, and compromised coding strategy remain major obstacles to overcome. Inspired by recent successes of deep learning for FPP, we propose a single-shot absolute 3D shape measurement with deep-learning-based color FPP. Through “learning” on extensive data sets, the properly trained neural network can “predict” the high-resolution, motion-artifact-free, crosstalk-free absolute phase directly from one single color fringe image. Compared with the traditional approach, our method allows for more accurate phase retrieval and more robust phase unwrapping. Experimental results demonstrate that the proposed approach can provide high-accuracy single-frame absolute 3D shape measurement for complicated objects.

The digitization of the complete shape of real objects has essential applications in fields of intelligent manufacturing, industrial detection, and reverse modeling. In order to build the full geometric models of rigid objects, the object must be moved relative to the measurement system (or the scanner must be moved relative to the object) to obtain and integrate views of the object from all sides, which not only complicates the system configuration but makes the whole process time-consuming. In this Letter, we present a high-resolution real-time 360° three-dimensional (3D) model reconstruction method that allows one to rotate an object manually and see a continuously updated 3D model during the scanning process. A multi-view fringe projection profilometry system acquires high-precision depth information about a handheld object from different perspectives and, meanwhile, the multiple views are aligned and merged together in real time. Our system employs stereo phase unwrapping and an adaptive depth constraint that can unwrap the phase of dense fringe images robustly without increasing the number of captured patterns. We then develop an efficient coarse-to-fine registration strategy to match the 3D surface segments rapidly. Our experiments demonstrate that our method can reconstruct the high-precision complete 3D model of complex objects under arbitrary rotation without any instrument assist and expensive pre/post-processing.

Structured illumination microscopy (SIM) is one of the powerful super-resolution modalities in bioscience with the advantages of full-field imaging and high photon efficiency. However, artifact-free super-resolution image reconstruction requires precise knowledge about the illumination parameters. The sample- and environment-dependent on-the-fly experimental parameters need to be retrieved a posteriori from the acquired data, posing a major challenge for real-time, long-term live-cell imaging, where low photobleaching, phototoxicity, and light dose are a must. In this work, we present an efficient and robust SIM algorithm based on principal component analysis (PCA-SIM). PCA-SIM is based on the observation that the ideal phasor matrix of a SIM pattern is of rank one, leading to the low complexity, precise identification of noninteger pixel wave vector and pattern phase while rejecting components that are unrelated to the parameter estimation. We demonstrate that PCA-SIM achieves non-iteratively fast, accurate (below 0.01-pixel wave vector and 0.1%of 2πrelative phase under typical noise level), and robust parameter estimation at low SNRs, which allows real-time super-resolution imaging of live cells in complicated experimental scenarios where other state-of-the-art methods inevitably fail. In particular, we provide the open-source MATLAB toolbox of our PCA-SIM algorithm and associated datasets. The combination of iteration-free reconstruction, robustness to noise, and limited computational complexity makes PCA-SIM a promising method for high-speed, long-term, artifact-free super-resolution imaging of live cells.

Conventional endoscopy and endoscopic ultrasonography (EUS) are used to estimate the invasion depth of early-stage gastric cancers (EGCs), but estimates made by either technique are often inaccurate. We developed a model to determine the invasion depth of EGCs using conventional endoscopy and EUS findings, with pathology results as the reference. We performed a retrospective study of 195 patients (205 lesions) diagnosed with gastric cancers who underwent endoscopy and EUS followed by resection. Based on pathology analyses, lesions (n= 205) were assigned to categories of: mucosa invasion or minute invasion into the submucosal layer less than 500 μm from the muscularis mucosae (M-SM1) or penetration of 500 μm or more (≥SM2). The lesions were randomly assigned to derivation (138 lesions) and validation sets (67 lesions). A depth predictive model was proposed in the derivation set using multivariate logistic regression analyses. The discriminative power of this model was assessed in both sets. Remarkable redness (OR 5.42; 95% CI 1.32–22.29), abrupt cutting of converging folds (OR 8.58; 95% CI 1.65–44.72), lesions location in the upper third of the stomach (OR 10.26; 95% CI 2.19–48.09), and deep invasion based on EUS findings (OR 16.53; 95% CI 4.48–61.15) significantly associated with ≥SM2 invasion. A model that incorporated these 4 variables discriminated between M-SM1 and ≥SM2 lesions with the area under the ROC curve of 0.865 in the derivation set and 0.797 in the validation set. In the derivation set, a cut-off score of 8 identified lesions as ≥SM2 with 54% sensitivity and 97% specificity. The model correctly predicted the invasion depth 89.86% of lesions; it overestimated the depth of 2.17% of lesions. We developed a model to identify EGCs with invasion depth ≥SM2 based on endoscopy and EUS findings. This model might reduce overestimation of gastric tumor depth and prevent unnecessary gastrectomy.

It is well accepted that inflammatory bowel disease (IBD), including ulcerative colitis and Crohn’s disease, is one of high risk factors for colorectal cancer (CRC). This supports the notion that inflammation plays a key role in cancer development. Epidemiologic studies have shown that vitamin D (Vit D) deficiency is associated with IBD and various types of cancer including colorectal cancer. Clinical trials and mouse models have shown that Vit D is beneficial in preventing or ameliorating inflammation and carcinogenesis. Mechanism studies show that Vit D played wide-spectrum roles in anti-inflammation, immunoregulation, and anti-cancer activities. In this review, we summarized and updated the findings on Vit D, IBD and colitis-associated colorectal cancer (CAC) with the emphasis on the effects of Vit D on anti-inflammation, EMT and cancer stem cells (CSC).

SEPT9gene methylation was validated as a biomarker for colorectal cancer (CRC) for >10 years and available as the Epi proColon test as an aid in CRC detection for >6 years. It was proven to be an accurate, reliable, fast, and convenient molecular test. In this opportunistic screening study, we validated a further simplified SEPT9gene methylation assay in 1031 subjects in Chinese hospitals. The sensitivity for CRC detection was 76.6% at a specificity of 95.9%, and the results showed a satisfactory detection rate for each CRC stage, including early stages. The new SEPT9assay, with enhanced technical simplicity, convenience, and lower cost, did not differ in performance compared with Epi proColon 2.0, the commercialized SEPT9assay. The CRC detection sensitivity was further enhanced when the assay was combined with carcinoembryonic antigen (sensitivity, 86.4%) or fecal immunochemical test (sensitivity, 94.4%), suggesting that the combined tests may be an effective option for future opportunistic screening. In brief, our study has validated a new SEPT9assay and combined testing as an aid in cancer detection, providing a new approach for opportunistic CRC screening.

To prospectively assess the utility of multidetector row computed tomography angiography (MDCTA) in the diagnosis of active gastrointestinal bleeding (GIB).

IL‐23 expressed in inflamed mucosa of IBD could induce IEL, NK, and T cell activation, proinflammatory cytokine secretion, and Th17 cell differentiation. This study analyzed IL‐23p19 expression in inflamed mucosa of IBD and the role in the induction of IEL and NK cell activation as well as Th17 cell differentiation. Expression of IL‐23p19 was performed by immunohistochemistry and quantitative real‐time PCR. Expression of IL‐23R was assessed by flow cytometry. Cytolytic activities of IEL and NK cells by IL‐23 were determined by a standard 51Cr‐release assay. Cytokine levels were analyzed by ELISA and quantitative real‐time PCR. Expression of IL‐23p19 was increased significantly in inflamed mucosa of CD compared with that in UC and healthy controls. Double‐staining confirmed that IL‐23p19+cells were mainly CD68+macrophages/DCs. IL‐23R+cells were increased significantly in PB‐ and LP‐CD4+and ‐CD8+T and NK cells. IL‐23 markedly promoted IBD IEL and NK cell activation and cytotoxicity and triggered IBD PB‐ and LP‐T cells to secrete significantly higher levels of IFN‐γ, TNF, IL‐2, and IL‐17A compared with controls. Importantly, IL‐23 promoted IBD PB‐ or LP‐CD4+T cells to differentiate into Th17 cells, characterized by increased expression of IL‐17A and RORC. Anti‐TNF treatment could markedly reduce IL‐23 expression and Th17 cell infiltration in inflamed mucosa of CD patients. These data indicate that IL‐23 is highly expressed in inflamed mucosa of IBD and plays an important role in the induction of IEL, NK, and T cell activation, proinflammatory cytokine secretion, and Th17 cell differentiation. Targeted therapy directed against IL‐23p19 may have a therapeutic role in treatment of IBD.

Background. Patients diagnosed with ulcerative colitis (UC) associated with primary sclerosis cholangitis (PSC) and sessile serrated adenoma (SSA) are rare. The present study aimed to identify the potential causative gene mutation in a patient with UC associated with PSC and SSA. Methods. DNA was extracted from the blood sample and tissue sample of SSA, followed by the whole exome sequencing (WES) analysis. Bioinformatics analysis was utilized to predict the deleteriousness of the identified variants. Multiple sequence alignment and conserved protein domain analyses were performed using online software. Sanger sequencing was used to validate the identified variants. Expression and diagnostic analysis of identified mutated genes was performed in the GSE119600 dataset (peripheral blood samples of PSC and UC) and GSE43841 dataset (tumor samples of SSA). Results. In the present study, a total of 842 single nucleotide variants (SNVs) in 728 genes were identified in the blood sample. Two variants, integrin beta 4 (ITGB4) (c.C2503G; p.P835A) and a mucin 3A (MUC3A) (c.C1019T; p.P340L), were further analyzed. MUC3A was associated with inflammatory bowel disease. Sanger sequence in blood revealed that the ITGB4 mutation was fully cosegregated with the result of WES in the patient. Additionally, a variant, tumor protein p53 gene (TP53) (c.86delA; p.N29Tfs∗15) was identified in the tissue sample of SSA. Compared to that in normal controls, ITGB4 was upregulated in both UC and PSC, MUC3A was, respectively, upregulated and downregulated in PSC and UC, and TP53 was downregulated in SSA. ITGB4 and TP53 had a potential diagnostic value for UC, PSC and SSA. Conclusions. The present study demonstrated that the ITGB4 (c.C2503G; p.P835A) and MUC3A (c.C1019T; p.P340L) mutations may be the potential causative variants in a patient with UC associated with PSC and SSA. TP53 (c.86delA; p.N29Tfs∗15) mutation may be associated with SSA in this patient.