Search

Achieving generalization for deep learning models has usually suffered from the bottleneck of annotated sample scarcity. As a common way of tackling this issue, few-shot learning focuses on "episodes", i.e. sampled tasks that help the model acquire generalizable knowledge onto unseen categories - better the episodes, the higher a model's generalisability. Despite extensive research, the characteristics of episodes and their potential effects are relatively less explored. A recent paper discussed that different episodes exhibit different prediction difficulties, and coined a new metric "hardness" to quantify episodes, which however is too wide-range for an arbitrary dataset and thus remains impractical for realistic applications. In this paper therefore, we for the first time conduct an algebraic analysis of the critical factors influencing episode hardness supported by experimental demonstrations, that reveal episode hardness to largely depend on classes within an episode, and importantly propose an efficient pre-sampling hardness assessment technique named Inverse-Fisher Discriminant Ratio (IFDR). This enables sampling hard episodes at the class level via class-level (cl) sampling scheme that drastically decreases quantification cost. Delving deeper, we also develop a variant called class-pair-level (cpl) sampling, which further reduces the sampling cost while guaranteeing the sampled distribution. Finally, comprehensive experiments conducted on benchmark datasets verify the efficacy of our proposed method. Codes are available at: https://github.com/PRIS-CV/class-level-sampling.

The aim of this study was to prepare active intelligent gluten protein films using wheat gluten protein (WG) and apple pectin (AP) as film-forming matrices, and blueberry anthocyanin extract (BAE) as a natural indicator. SEM and FT-IR analyses demonstrated the successful immobilization of BAE in the film matrix by hydrogen bonding interactions and its compatibility with WG and AP. The resultant WG-AP/BAE indicator films demonstrated notable antioxidant activity, color stability, barrier qualities, pH and ammonia response sensitivity, and mechanical properties. Among them, WG-AP/BAE5 exhibited the best mechanical properties (TS: 0.83 MPa and EB: 242.23%) as well as the lowest WVP (3.92 × 10 -8  g.m/m 2 .Pa.s), and displayed high sensitivity to volatile ammonia. In addition, WG-AP/BAE5 showed a color shift from purplish red to green to yellowish green, demonstrating the monitoring of shrimp freshness in real time. Consequently, this study offers a firm scientific foundation for the development of active intelligent gluten protein films and their use in food freshness assessments., Competing Interests: Declaration of competing interest None to declare., (Copyright © 2024 Elsevier Ltd. All rights reserved.)

Professor Takeyori Saheki's team at Kagoshima University, Japan, published a paper in Nature Genetics in June 1999, pinpointing the pathogenic gene for adult-onset type Ⅱ citrullinemia as SLC25A13 and naming the protein product encoded by this gene as citrin. Over the past 25 years, the researches have made positive progress on the pathophysiological mechanism, clinical phenotype, molecular diagnosis, treatment, and prognosis of citrin deficiency (CD) as an autosomal recessive genetic disease. Currently, three age-dependent clinical phenotypes of CD have been found, namely neonatal intrahepatic cholestasis caused by citrin deficiency, failure to thrive and dyslipidemia caused by citrin deficiency, and adult-onset type Ⅱ citrullinemia. Although relevant internal medicine drugs are being researched and developed while liver transplantation has been used for the treatment of CD patients, scientific dietary therapy remains the foundation, core, and key for the management of this disease. Furthermore, CD management involves the full life cycle of patients, requiring the joint efforts of basic and clinical medicine as well as systematic articulation at multi-levels, such as the parents, family, and society. By full-cycle, multidisciplinary, and systematic management, it is an achievable goal for CD patients to learn, work, and live in a healthy manner.

The quest for novel antibacterial agents is imperative in the face of escalating antibiotic resistance. Naturally occurring tetrahydro-β-carboline (THβC) alkaloids have been highlighted due to their significant biological derivatives. However, these structures have been little explored for antibacterial drugs development. In this study, a series of 1,2,3,4-THβC derivatives were synthesized and assessed for their antibacterial prowess against both gram-positive and gram-negative bacteria. The compounds exhibited moderate to good antibacterial activity, with some compounds showing superior efficacy against gram-positive bacteria, especially methicillin-resistant Staphylococcus aureus (MRSA), to that of Gentamicin. Among these analogs, compound 3k emerged as a hit compound, demonstrating rapid bactericidal action and a significant post-antibacterial effect, with significant cytotoxicity towards human LO2 and HepG2 cells. In addition, compound 3k (10 mg/kg) showed comparable anti-MRSA efficacy to Ciprofloxacin (2 mg/kg) in a mouse model of abdominal infection. Overall, the present findings suggested that THβC derivatives based on the title compounds hold promising applications in the development of antibacterial drugs., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 Elsevier Ltd. All rights reserved.)

Four new clerodane diterpenoids, namely tinocapills A-D (1-4), and one known analogue (5) were isolated from the roots of Tinospora capillipes in the present study. The structures of these new compounds, including their absolute configurations, were determined through a combination of detailed spectroscopic analysis and theoretical statistical approaches, including electronic circular dichroism (ECD) analyses and quantum mechanical (QM)-NMR methods. Additionally, the stereostructure of 5 was confirmed via X-ray diffraction analysis. Furthermore, all these isolates were evaluated for their antibacterial and anti-inflammatory activities. Compounds 1, 2 and 5 demonstrated antibacterial activity against methicillin-resistant Staphylococcus aureus (MRSA) with MICs ranging from 4-64 μg/mL, and compounds 3 and 4 exhibited potential anti-inflammatory effects by suppressing LPS-induced TNF-α and NO releases in RAW264.7 cells., (© 2024 Wiley-VHCA AG, Zurich, Switzerland.)

The main challenge for fine-grained few-shot image classification is to learn feature representations with higher inter-class and lower intra-class variations, with a mere few labelled samples. Conventional few-shot learning methods however cannot be naively adopted for this fine-grained setting - a quick pilot study reveals that they in fact push for the opposite (i.e., lower inter-class variations and higher intra-class variations). To alleviate this problem, prior works predominately use a support set to reconstruct the query image and then utilize metric learning to determine its category. Upon careful inspection, we further reveal that such unidirectional reconstruction methods only help to increase inter-class variations and are not effective in tackling intra-class variations. In this paper, we introduce a bi-reconstruction mechanism that can simultaneously accommodate for inter-class and intra-class variations. In addition to using the support set to reconstruct the query set for increasing inter-class variations, we further use the query set to reconstruct the support set for reducing intra-class variations. This design effectively helps the model to explore more subtle and discriminative features which is key for the fine-grained problem in hand. Furthermore, we also construct a self-reconstruction module to work alongside the bi-directional module to make the features even more discriminative. We introduce the snapshot ensemble method in the episodic learning strategy - a simple trick to further improve model performance without increasing training costs. Experimental results on three widely used fine-grained image classification datasets, as well as general and cross-domain few-shot image datasets, consistently show considerable improvements compared with other methods.

This study investigated the effects of Lycium barbarum pulp (LBP) on the properties of mixed dough and gluten protein. The results showed that appropriate addition of LBP (5 %) significantly improved the performance of the dough, promoted the aggregation of gluten protein, enhanced the water binding ability, and delayed the gelatinization of starch during cooking. Compared with the control group, the peak temperature (T p ) of the LBP sample gradually increased from 63.23 °C to 65.56 °C, the expansion force reduced by about 21.56 %, the absolute Zeta potential lowered by about 18.4 %, and the α -helix content and β -folding increased by 32.36 % and 10.23 %, respectively, indicating the more orderly and stable overall structure. However, LBP did not change the crystal configuration of starch and still showed typical type A line diffraction. Moreover, the addition of LBP increased the polyphenol content, which further improved the antioxidant properties and provided the possibility to improve the health potential of the flour., Competing Interests: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (© 2024 The Authors.)

Objective: To investigate the postoperative outcomes and visual quality following laser blended vision (LBV) surgery for the correction of myopia with presbyopia. Methods: This is a prospective study. A total of 50 patients (100 eyes) who underwent LBV surgery for myopia with presbyopia at Beijing Tongren Hospital, Capital Medical University, between August 2021 and March 2022 were included. Uncorrected distance visual acuity (UDVA), uncorrected near visual acuity (UNVA), best-corrected distance visual acuity (BCDVA), manifest refraction spherical equivalent (MRSE) were measured preoperatively and at 1, 3, 6, 12, and 24 months postoperatively for the dominant eye, non-dominant eye, and both eyes. Effectiveness Index (EI) and Safety Index (SI) were calculated to evaluate refractive outcomes. Accommodation function and contrast sensitivity were assessed to evaluate functional vision. Objective visual quality was assessed using higher-order aberrations (HOAs) and a dual-channel visual quality analysis system. Subjective visual quality was evaluated using a questionnaire. Results: At 24 months postoperatively, the EI and SI for the dominant eye were 1.04±0.23 and 1.14±0.14, respectively, while for the non-dominant eye, they were 0.85±0.21 and 1.06±0.08, respectively. Although there were statistically significant differences in UNVA between the eyes at 1 and 3 months postoperatively (all P <0.05), no significant differences were observed from 6 months onward (all P >0.05). At 6 months postoperatively, the UNVA for the dominant and non-dominant eyes were 0.07±0.13 and 0.03±0.13, respectively, with no significant difference ( P >0.05). MRSE showed no significant differences at any follow-up time points up to 24 months. Accommodation function did not decrease compared to preoperative levels at any postoperative follow-up. Increases in HOAs were primarily due to spherical aberrations, and there were no significant differences in objective visual quality before and after surgery. Contrast sensitivity improved significantly at 24 months postoperatively compared to preoperative levels. At 24 months postoperatively, 11 out of 12 patients reported good visual quality in the questionnaire. Conclusions: LBV surgery gradually achieves good binocular blended vision within 6 months postoperatively and demonstrates good safety, efficacy, and stability at 24 months. It maintains good functional vision and visual quality with high postoperative patient satisfaction.

The proficiency of phyllosphere microbiomes in efficiently utilizing plant-provided nutrients is pivotal for their successful colonization of plants. The methylotrophic capabilities of Methylobacterium/Methylorubrum play a crucial role in this process. However, the precise mechanisms facilitating efficient colonization remain elusive. In the present study, we investigate the significance of methanol assimilation in shaping the success of mutualistic relationships between methylotrophs and plants. A set of strains originating from Methylorubrum extorquens AM1 are subjected to evolutionary pressures to thrive under low methanol conditions. A mutation in the phosphoribosylpyrophosphate synthetase gene is identified, which converts it into a metabolic valve. This valve redirects limited C1-carbon resources towards the synthesis of biomass by up-regulating a non-essential phosphoketolase pathway. These newly acquired bacterial traits demonstrate superior colonization capabilities, even at low abundance, leading to increased growth of inoculated plants. This function is prevalent in Methylobacterium/Methylorubrum strains. In summary, our findings offer insights that could guide the selection of Methylobacterium/Methylorubrum strains for advantageous agricultural applications., (© 2024. The Author(s).)

Deficiency of citrin, the liver-type aspartate-glutamate carrier, arises from biallelic mutations of the gene SLC25A13. Although citrin deficiency (CD) is associated with higher risk of hepatocellular carcinoma (HCC) in adult patients, this association remains inconclusive in pediatric cases. The patient in this paper had been diagnosed to have CD by SLC25A13 analysis at the age 10 months, and then in response to dietary therapy, her prolonged jaundice and marked hepatosplenomegaly resolved gradually. However, she was referred to the hospital once again due to recurrent abdominal distention for 2 weeks at her age 4 years and 9 months, when prominently enlarged liver and spleen were palpated, along with a strikingly elevated serum alpha-fetoprotein (AFP) level of 27605 ng/mL as well as a large mass in the right liver lobe and a suspected tumor thrombus within the portal vein on enhanced computed tomography. After 4 rounds of adjuvant chemotherapy, right hepatic lobectomy and portal venous embolectomy were performed at her age 5 years and 3 months, and metastatic hepatoblastoma was confirmed by histopathological analysis. Afterwards, the patient underwent 5 additional cycles of chemotherapy and her condition remained stable for 7 months after surgery. Unfortunately, hepatoblastoma recurred in the left lobe at the age 5 years and 10 months, which progressed rapidly into liver failure, and led to death at the age 6 years and 1 month. As far as we know, this is the the first case of hepatoblastoma in a patient with CD, raising the possibility of an association between these two conditions., (© 2024. The Author(s).)

Despite great strides made on fine-grained visual classification (FGVC), current methods are still heavily reliant on fully-supervised paradigms where ample expert labels are called for. Semi-supervised learning (SSL) techniques, acquiring knowledge from unlabeled data, provide a considerable means forward and have shown great promise for coarse-grained problems. However, exiting SSL paradigms mostly assume in-category (i.e., category-aligned) unlabeled data, which hinders their effectiveness when re-proposed on FGVC. In this paper, we put forward a novel design specifically aimed at making out-of-category data work for semi-supervised FGVC. We work off an important assumption that all fine-grained categories naturally follow a hierarchical structure (e.g., the phylogenetic tree of "Aves" that covers all bird species). It follows that, instead of operating on individual samples, we can instead predict sample relations within this tree structure as the optimization goal of SSL. Beyond this, we further introduced two strategies uniquely brought by these tree structures to achieve inter-sample consistency regularization and reliable pseudo-relation. Our experimental results reveal that (i) the proposed method yields good robustness against out-of-category data, and (ii) it can be equipped with prior arts, boosting their performance thus yielding state-of-the-art results.

This study explored the potential of methanol as a sustainable feedstock for biomanufacturing, focusing on Methylobacterium extorquens, a well-established representative of methylotrophic cell factories. Despite this bacterium's long history, its untapped photosynthetic capabilities for production enhancement have remained unreported. Using genome-scale flux balance analysis, it was hypothesized that introducing photon fluxes could boost the yield of 3-hydroxypropionic acid (3-HP), an energy- and reducing equivalent-consuming chemicals. To realize this, M. extorquens was genetically modified by eliminating the negative regulator of photosynthesis, leading to improved ATP levels and metabolic activity in non-growth cells during a two-stage fermentation process. This modification resulted in a remarkable 3.0-fold increase in 3-HP titer and a 2.1-fold increase in its yield during stage (II). Transcriptomics revealed that enhanced light-driven methanol oxidation, NADH transhydrogenation, ATP generation, and fatty acid degradation were key factors. This development of photo-methylotrophy as a platform technology introduced novel opportunities for future production enhancements., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2023 Elsevier Ltd. All rights reserved.)

The problem of sketch semantic segmentation is far from being solved. Despite existing methods exhibiting near-saturating performances on simple sketches with high recognisability, they suffer serious setbacks when the target sketches are products of an imaginative process with high degree of creativity. We hypothesise that human creativity, being highly individualistic, induces a significant shift in distribution of sketches, leading to poor model generalisation. Such hypothesis, backed by empirical evidences, opens the door for a solution that explicitly disentangles creativity while learning sketch representations. We materialise this by crafting a learnable creativity estimator that assigns a scalar score of creativity to each sketch. It follows that we introduce CreativeSeg, a learning-to-learn framework that leverages the estimator in order to learn creativity-agnostic representation, and eventually the downstream semantic segmentation task. We empirically verify the superiority of CreativeSeg on the recent "Creative Birds" and "Creative Creatures" creative sketch datasets. Through a human study, we further strengthen the case that the learned creativity score does indeed have a positive correlation with the subjective creativity of human. Codes are available at https://github.com/PRIS-CV/Sketch-CS.

As powerful as fine-grained visual classification (FGVC) is, responding your query with a bird name of "Whip-poor-will" or "Mallard" probably does not make much sense. This however commonly accepted in the literature, underlines a fundamental question interfacing AI and human - what constitutes transferable knowledge for human to learn from AI? This paper sets out to answer this very question using FGVC as a test bed. Specifically, we envisage a scenario where a trained FGVC model (the AI expert) functions as a knowledge provider in enabling average people (you and me) to become better domain experts ourselves. Assuming an AI expert trained using expert human labels, we anchor our focus on asking and providing solutions for two questions: (i) what is the best transferable knowledge we can extract from AI, and (ii) what is the most practical means to measure the gains in expertise given that knowledge? We propose to represent knowledge as highly discriminative visual regions that are expert-exclusive and instantiate it via a novel multi-stage learning framework. A human study of 15,000 trials shows our method is able to consistently improve people of divergent bird expertise to recognise once unrecognisable birds. We further propose a crude but benchmarkable metric TEMI and therefore allow future efforts in this direction to be comparable to ours without the need of large-scale human studies.

Objective: This study focuses on Na(+)-taurocholate cotransporting polypeptide (NTCP) deficiency to analyze and investigate the value of the serum bile acid profile for facilitating the diagnosis and differential diagnosis. Methods: Clinical data of 66 patients with cholestatic liver diseases (CLDs) diagnosed and treated in the Department of Pediatrics of the First Affiliated Hospital of Jinan University from early April 2015 to the end of December 2021 were collected, including 32 cases of NTCP deficiency (16 adults and 16 children), 16 cases of neonatal intrahepatic cholestasis caused by citrin deficiency (NICCD), 8 cases of Alagille syndrome, and 10 cases of biliary atresia. At the same time, adult and pediatric healthy control groups (15 cases each) were established. The serum bile acid components of the study subjects were qualitatively and quantitatively analyzed by ultra-high performance liquid chromatography-tandem mass spectrometry. The data were plotted and compared using statistical SPSS 19.0 and GraphPad Prism 5.0 software. The clinical and bile acid profiles of children with NTCP deficiency and corresponding healthy controls, as well as differences between NTCP deficiency and other CLDs, were compared using statistical methods such as t-tests, Wilcoxon rank sum tests, and Kruskal-Wallis H tests. Results: Compared with the healthy control, the levels of total conjugated bile acids, total primary bile acids, total secondary bile acids, glycocholic acid, taurocholic acid, and glycochenodeoxycholic acid were increased in NTCP deficiency patients ( P < 0.05). Compared with adults with NTCP deficiency, the levels of total conjugated bile acids and total primary bile acids were significantly increased in children with NTCP deficiency ( P < 0.05). The serum levels of taurochenodeoxycholic acid, glycolithocholate, taurohyocholate, and tauro-α-muricholic acid were significantly increased in children with NTCP deficiency, but the bile acid levels such as glycodeoxycholic acid, glycolithocholate, and lithocholic acid were decreased ( P < 0.05). The serum levels of secondary bile acids such as lithocholic acid, deoxycholic acid, and hyodeoxycholic acid were significantly higher in children with NTCP deficiency than those in other CLD groups such as NICCD, Alagille syndrome, and biliary atresia ( P < 0.05). Total primary bile acids/total secondary bile acids, total conjugated bile acids/total unconjugated bile acids, taurocholic acid, serum taurodeoxycholic acid, and glycodeoxycholic acid effectively distinguished children with NTCP deficiency from other non-NTCP deficiency CLDs. Conclusion: This study confirms that serum bile acid profile analysis has an important reference value for facilitating the diagnosis and differential diagnosis of NTCP deficiency. Furthermore, it deepens the scientific understanding of the changing characteristics of serum bile acid profiles in patients with CLDs such as NTCP deficiency, provides a metabolomic basis for in-depth understanding of its pathogenesis, and provides clues and ideas for subsequent in-depth research.

Aim: To analyze the differences, agreements, and correlation among total corneal power parameters generated by different instruments after myopic keratorefractive surgery., Methods: The prospective cross-sectional study included patients who underwent myopic keratorefractive surgery and received measurements of corneal power 3mo after surgery. Automated keratometer was used for the measurement of simulated keratometry (SimK), swept-source optical coherence tomography (SS-OCT) based biometer for total keratometry (TK), anterior segment-OCT for real keratometry (RK), and Scheimpflug keratometer for the true net power (TNP), the total corneal refractive power (TCRP) and equivalent K-readings (EKR). The differences among these parameters were analyzed, and the agreements and correlation between SimK and other total corneal power parameters were investigated., Results: A total of 70 eyes of 70 patients after myopic keratorefractive surgery were included. The evaluated corneal power parameters were as follows: SimK 38.32±1.93 D, TK 37.54±2.12 D, RK 36.64±2.09 D, TNP 36.56±1.97 D, TCRP 36.70±2.01 D, and EKR 37.55±2.00 D. Pairwise comparison showed that there were significant differences ( P <0.001) among all parameters except for between TK and EKR, RK and TNP, RK and TCRP ( P =1.000, 1.000, 1.000, respectively). The limits of agreement between SimK and TK, RK, TNP, TCPR, and EKR were 1.08, 1.08, 1.43, 1.48, and 1.73 D, respectively. All parameters showed good correlation with SimK, and the correlation coefficients were 0.995, 0.994, 0.983, 0.982, and 0.975., Conclusion: Among the corneal power parameters after myopic keratorefractive surgery, the value of SimK is the largest, followed by TK and EKR, with TCRP, RK, and TNP being the smallest. The differences among the parameters may be attributable to the different calculation principles. Correct understanding and evaluation of corneal power parameters can provide a theoretical basis for taking advantage of the total corneal power to improve the accuracy of intraocular lens calculation after keratorefractive surgery., (International Journal of Ophthalmology Press.)

Motion recognition provides movement information for people with physical dysfunction, the elderly and motion-sensing games production, and is important for accurate recognition of human motion. We employed three classical machine learning algorithms and three deep learning algorithm models for motion recognition, namely Random Forests (RF), K-Nearest Neighbors (KNN) and Decision Tree (DT) and Dynamic Neural Network (DNN), Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN). Compared with the Inertial Measurement Unit (IMU) worn on seven parts of body. Overall, the difference in performance among the three classical machine learning algorithms in this study was insignificant. The RF algorithm model performed best, having achieved a recognition rate of 96.67%, followed by the KNN algorithm model with an optimal recognition rate of 95.31% and the DT algorithm with an optimal recognition rate of 94.85%. The performance difference among deep learning algorithm models was significant. The DNN algorithm model performed best, having achieved a recognition rate of 97.71%. Our study validated the feasibility of using multidimensional data for motion recognition and demonstrated that the optimal wearing part for distinguishing daily activities based on multidimensional sensing data was the waist. In terms of algorithms, deep learning algorithms based on multi-dimensional sensors performed better, and tree-structured models still have better performance in traditional machine learning algorithms. The results indicated that IMU combined with deep learning algorithms can effectively recognize actions and provided a promising basis for a wider range of applications in the field of motion recognition.

Objective: The purpose of this study was to assess the safety and efficacy of a second allogeneic hematopoietic stem cell transplantation (allo-HSCT) with reduced-intensity conditioning (RIC) in patients with hematological malignancies who had relapsed after the first allo-HSCT. Methods: Between April 2018 and June 2021, 44 patients with hematological malignancies (B-ALL 23, T-ALL/T-LBL 4, AML15, and MDS 2) were enrolled and retrospectively examined. Unrelated donors ( n =12) or haploidentical donors ( n =32) were used. Donors were replaced in all patients for the second allo-HSCT. Hematological and immunological germline predisposition genes and hematopoietic and immune function tests were used to select the best-related donor. Total body irradiation (TBI) /fludarabine (FLU) -based ( n =38), busulfan (BU) /FLU-based ( n =4), total marrow irradiation (TMI) /FLU-based ( n =1), and BU/cladribine-based ( n =1) were the RIC regimens used. For graft versus host disease (GVHD) prevention, cyclosporine, mycophenolate mofetil, short-term methotrexate, and ATG were used. Eighteen (40.9%) of 44 patients with gene variations for which targeted medications are available underwent post-transplant maintenance therapy. Results: The median age was 25 years old (range: 7-55). The median interval between the first and second HSCT was 19.5 months (range: 6-77). Before the second allo-HSCT, 33 (75%) of the patients were in complete remission (CR), whereas 11 (25%) were not. All patients had long-term engraftment. The grade Ⅱ-Ⅳ GVHD and severe acute GVHD rates were 20.5% and 9.1%, respectively. Chronic GVHD was found in 20.5% of limited patterns and 22.7% of severe patterns. CMV and EBV reactivation rates were 29.5% and 6.8%, respectively. Hemorrhage cystitis occurred in 15.9% of cases, grade Ⅰ or Ⅱ. The 1-yr disease-free survival (DFS), overall survival (OS), and cumulative recurrence incidence (RI) rates of all patients were 72.5% (95% CI , 54.5%-84.3%), 80.6% (95% CI , 63.4%-90.3%), and 25.1% (95% CI , 13.7%-43.2%), respectively, with a median follow-up of 14 (2-39) months. There were eight deaths (seven relapses and one infection). The rate of non-relapse mortality (NRM) was only 2.3%. The CR patients' 1-yr RI rate was significantly lower than the NR patients (16.8% vs 48.1%, P =0.026). The DFS rate in CR patients was greater than in NR patients, although there was no statistical difference (79.9% vs 51.9%, P =0.072). Univariate analysis revealed that CR before the second allo-HSCT was an important prognostic factor. Conclusion: With our RIC regimens, donor change, and post-transplant maintenance therapy, the second allo-HSCT in relapsed hematological malignancies after the first allo-HSCT is a safe and effective treatment with high OS and DFS and low NRM and relapse rate. The most important factor influencing the prognosis of the second allo-HSCT is the patient's illness condition before the transplant.

Objective: To compare the accuracy of different corneal curvature parameters in assessing the corneal refractive status and tracking corneal power changes after small incision lenticule extraction (SMILE). Methods: This prospective cross-sectional study tracked and recorded total corneal curvature parameters measured by different instruments before and three months after SMILE for myopia. These parameters, including total keratometry (TK) from the IOLMaster 700, total corneal refractive power (TCRP) from the Pentacam AXL, real keratometry (RK) from the CASIA 2, and corrected parameters calculated using the Haigis, Shammas, and Maloney methods, were compared with data obtained using the clinical history method (CHM). Surgically induced changes in TK, TCRP, and RK were analyzed and compared with those in spherical equivalent on the corneal plane (ΔSEco). Results: The study included 40 eyes (40 participants). After SMILE, the difference was smallest between TK [(0.08±0.38) D] and CHM values ( P >0.05). However, TCRP, RK, K Haigis , K Shammas , and K Maloney were significantly different from CHM data ( P <0.05). The width of the 95% limits of agreement of TK (1.49 D) was narrowest, followed by that of RK (1.57 D). Pearson analysis showed that each parameter had a good correlation with CHM data. The differences between the changes in TK, TCRP and RK caused by surgery and ΔSEco were (0.03±0.39) D, (0.17±0.43) D, and (-0.19±0.46) D, respectively. The width of the 95% limits of agreement of ΔTK (1.54 D) was narrowest, and the correlation coefficient of ΔTK (0.951) was highest. Conclusion: The parameter TK of the IOLMaster 700 can provide accurate and objective corneal power evaluation after SMILE.

Objective: To study the diagnostic value of different detection markers in histological categories of endocervical adenocarcinoma (ECA), and their assessment of patient prognosis. Methods: A retrospective study of 54 patients with ECA in the Cancer Hospital, Chinese Academy of Medical Sciences from 2005-2010 were performed. The cases of ECA were classified into two categories, namely human papillomavirus-associated adenocarcinoma (HPVA) and non-human papillomavirus-associated adenocarcinoma (NHPVA), based on the 2018 international endocervical adenocarcinoma criteria and classification (IECC). To detect HR-HPV DNA and HR-HPV E6/E7 mRNA in all patients, we used whole tissue section PCR (WTS-PCR) and HPV E6/E7 mRNA in situ hybridization (ISH) techniques, respectively. Additionally, we performed Laser microdissection PCR (LCM-PCR) on 15 randomly selected HR-HPV DNA-positive cases to confirm the accuracy of the above two assays in identifying ECA lesions. Receiver operating characteristic (ROC) curves were used to analyze the efficacy of markers to identify HPVA and NHPVA. Univariate and multifactorial Cox proportional risk model regression analyses were performed for factors influencing ECA patients' prognoses. Results: Of the 54 patients with ECA, 30 were HPVA and 24 were NHPVA. A total of 96.7% (29/30) of HPVA patients were positive for HR-HPV DNA and 63.3% (19/30) for HR-HPV E6/E7 mRNA, and 33.3% (8/24) of NHPVA patients were positive for HR-HPV DNA and HR-HPV E6/E7 mRNA was not detected (0/24), and the differences were statistically significant ( P <0.001). LCM-PCR showed that five patients were positive for HR-HPV DNA in the area of glandular epithelial lesions and others were negative, which was in good agreement with the E6/E7 mRNA ISH assay ( Kappa =0.842, P =0.001). Analysis of the ROC results showed that the AUC of HR-HPV DNA, HR-HPV E6/E7 mRNA, and p16 to identify HPVA and NHPVA were 0.817, 0.817, and 0.692, respectively, with sensitivities of 96.7%, 63.3%, and 80.0% and specificities of 66.7%, 100.0%, and 58.3%, respectively. HR-HPV DNA identified HPVA and NHPVA with higher AUC than p16 ( P =0.044). The difference in survival rates between HR-HPV DNA (WTS-PCR assay) positive and negative patients was not statistically significant ( P =0.156), while the difference in survival rates between HR-HPV E6/E7 mRNA positive and negative patients, and p16 positive and negative patients were statistically significant (both P <0.05). Multifactorial Cox regression analysis showed that International Federation of Obstetrics and Gynecology (FIGO) staging ( HR =19.875, 95% CI: 1.526-258.833) and parametrial involvement ( HR =14.032, 95% CI: 1.281-153.761) were independent factors influencing the prognosis of patients with ECA. Conclusions: HR-HPV E6/E7 mRNA is more reflective of HPV infection in ECA tissue. The efficacy of HR-HPV E6/E7 mRNA and HR-HPV DNA (WTS-PCR assay) in identifying HPVA and NHPVA is similar, with higher sensitivity of HR-HPV DNA and higher specificity of HR-HPV E6/E7 mRNA. HR-HPV DNA is more effective than p16 in identifying HPVA and NHPVA. HPV E6/E7 mRNA and p16 positive ECA patients have better survival rates than negative.

An esophageal fistula can be caused by an esophageal tumor as well as the surgery, radiotherapy (RT), or chemoradiotherapy used to treat the tumor. The most dangerous complications are massive hemoptysis and asphyxia. This report describes a 58-year-old man with a >1-month history of dysphagia and hemoptysis. Contrast-enhanced computed tomography revealed a tumor in the upper esophagus and a tracheoesophageal fistula. Esophagography revealed a large lesion measuring approximately 8 cm in length. Esophagogastroduodenoscopy showed an ulcerated tumor with raised margins originating 22 cm from the incisors, and histologic examination of a biopsy specimen indicated squamous cell carcinoma. The tumor was finally classified as stage IVA (T4bN0M0) esophageal squamous cell carcinoma. Massive hemoptysis occurred after the patient was admitted to the hospital. Therefore, we applied staged dose-escalated RT in three stages (6.0 Gy in 5 fractions, 7.5 Gy in 5 fractions, and 46.8 Gy in 26 fractions) to decrease the rate of tumor shrinkage brought on by RT and give the normal tissue enough time to close the fistula. Finally, the hemoptysis resolved and the patient's symptoms were significantly improved. Contrast-enhanced chest computed tomography revealed shrinkage of the tumor. In conclusion, staged dose-escalated RT can be applied for esophageal fistula closure.

A two-dimensional phononic crystal sensor model with high-quality factor and excellent sensitivity for sensing acetone solutions and operating at 25-45 kHz is proposed. The model for filling solution cavities is based on reference designs of quasi-crystal and gradient cavity structures. The transmission spectrum of sensor is simulated by the finite element method. High-quality factor of 45,793.06 and sensitivity of 80,166.67 Hz are obtained for the acetone concentration with 1-9.1%, and quality factor of 61,438.09 and sensitivity of 24,400.00 Hz are obtained for the acetone concentration range of 10-100%, which indicated the sensor could still achieve high sensitivity and quality factor at operating frequencies from 25 to 45 kHz. To verify the application of the sensor to sensing other solutions, the sensitivity for sound velocity and density is calculated as 24.61 m -1 and 0.7764 m 3 /(kg × s), respectively. It indicates the sensor is sensitive to acoustic impedance changes of the solution and equally suitable for sensing other solutions. The simulation results reveal the phononic crystal sensor possessed high-performance in composition capture in pharmaceutical production and petrochemical industry, which can provide theoretical reference for the design of new biochemical sensors for reliable detection of solution concentration., (© 2023. The Author(s).)

Counties are the key spatial units in achieving the reduction and control of carbon emissions. It is of great significance to study and reveal the spatial-temporal evolution characteristics and influencing mechanism of carbon emissions for realizing the "carbon peak and carbon neutral" goal. In this study, the spatial-temporal evolution and heterogeneity of carbon emissions at the county level in China from 2000 to 2017 were analyzed by using mathematical statistics and panel data regression modeling, and the influencing mechanism was explored. The results showed that: ① from 2000 to 2017, the annual growth rate of carbon emissions was 7.12%, which experienced the three stages of "sharp rise, slow rise, and high fluctuation" and finally stabilized at approximately 90×10 8 t. At the county scale, there was a significant positive spatial autocorrelation. ② The general panel regression model showed that GDP, construction land area, population, per capita GDP, and per capita deposit balance of financial institutions were significantly correlated with carbon emissions, and the former three had the strongest promoting effect on carbon emissions. ③ The goodness of fit of the geographically and temporally weighted regression model was high, and the direction and intensity of the other impact factors changed greatly in spatial-temporal characteristics, except that GDP showed a stable promoting effect globally. The results showed that carbon emission levels and main influencing factors varied among counties in China. This study revealed the heterogeneity of carbon emissions at the county level, which is helpful to optimize the spatial-temporal implementation path of the "dual carbon" target.

Sketch is a well-researched topic in the vision community by now. Sketch semantic segmentation in particular, serves as a fundamental step towards finer-level sketch interpretation. Recent works use various means of extracting discriminative features from sketches and have achieved considerable improvements on segmentation accuracy. Common approaches for this include attending to the sketch-image as a whole, its stroke-level representation or the sequence information embedded in it. However, they mostly focus on only a part of such multi-facet information. In this paper, we for the first time demonstrate that there is complementary information to be explored across all these three facets of sketch data, and that segmentation performance consequently benefits as a result of such exploration of sketch-specific information. Specifically, we propose the Sketch-Segformer, a transformer-based framework for sketch semantic segmentation that inherently treats sketches as stroke sequences other than pixel-maps. In particular, Sketch-Segformer introduces two types of self-attention modules having similar structures that work with different receptive fields (i.e., whole sketch or individual stroke). The order embedding is then further synergized with spatial embeddings learned from the entire sketch as well as localized stroke-level information. Extensive experiments show that our sketch-specific design is not only able to obtain state-of-the-art performance on traditional figurative sketches (such as SPG, SketchSeg-150K datasets), but also performs well on creative sketches that do not conform to conventional object semantics (CreativeSketch dataset) thanks for our usage of multi-facet sketch information. Ablation studies, visualizations, and invariance tests further justifies our design choice and the effectiveness of Sketch-Segformer. Codes are available at https://github.com/PRIS-CV/Sketch-SF.

Generalized Few-shot Semantic Segmentation (GFSS) aims to segment each image pixel into either base classes with abundant training examples or novel classes with only a handful of (e. g., 1-5) training images per class. Compared to the widely studied Few-shot Semantic Segmentation (FSS), which is limited to segmenting novel classes only, GFSS is much under-studied despite being more practical. Existing approach to GFSS is based on classifier parameter fusion whereby a newly trained novel class classifier and a pre-trained base class classifier are combined to form a new classifier. As the training data is dominated by base classes, this approach is inevitably biased towards the base classes. In this work, we propose a novel Prediction Calibration Network (PCN) to address this problem. Instead of fusing the classifier parameters, we fuse the scores produced separately by the base and novel classifiers. To ensure that the fused scores are not biased to either the base or novel classes, a new Transformer-based calibration module is introduced. It is known that the lower-level features are useful of detecting edge information in an input image than higher-level features. Thus, we build a cross-attention module that guides the classifier's final prediction using the fused multi-level features. However, transformers are computationally demanding. Crucially, to make the proposed cross-attention module training tractable at the pixel level, this module is designed based on feature-score cross-covariance and episodically trained to be generalizable at inference time. Extensive experiments on PASCAL- 5 i and COCO- 20 i show that our PCN outperforms the state-the-the-art alternatives by large margins.

Free-hand sketches are highly illustrative, and have been widely used by humans to depict objects or stories from ancient times to the present. The recent prevalence of touchscreen devices has made sketch creation a much easier task than ever and consequently made sketch-oriented applications increasingly popular. The progress of deep learning has immensely benefited free-hand sketch research and applications. This paper presents a comprehensive survey of the deep learning techniques oriented at free-hand sketch data, and the applications that they enable. The main contents of this survey include: (i) A discussion of the intrinsic traits and unique challenges of free-hand sketch, to highlight the essential differences between sketch data and other data modalities, e.g., natural photos. (ii) A review of the developments of free-hand sketch research in the deep learning era, by surveying existing datasets, research topics, and the state-of-the-art methods through a detailed taxonomy and experimental evaluation. (iii) Promotion of future work via a discussion of bottlenecks, open problems, and potential research directions for the community.

Source-Free Domain Adaptation (SFDA) is becoming topical to address the challenge of distribution shift between training and deployment data, while also relaxing the requirement of source data availability during target domain adaptation. In this paper, we focus on SFDA for semantic segmentation, in which pseudo labeling based target domain self-training is a common solution. However, pseudo labels generated by the source models are particularly unreliable on the target domain data due to the domain shift issue. Therefore, we propose to use Bayesian Neural Network (BNN) to improve the target self-training by better estimating and exploiting pseudo-label uncertainty. With the uncertainty estimation of BNNs, we introduce two novel self-training based components: Uncertainty-aware Online Teacher-Student Learning (UOTSL) and Uncertainty-aware FeatureMix (UFM). Extensive experiments on two popular benchmarks, GTA 5 → Cityscapes and SYNTHIA → Cityscapes, show the superiority of our proposed method with mIoU gains of 3.6% and 5.7% over the state-of-the-art respectively.

Background: Progressive familial intrahepatic cholestasis type 3 (PFIC3) is an autosomal recessive disease caused by pathogenic variants of the gene ABCB4. This study aimed to investigate the ABCB4 genotypic and the clinical phenotypic features of PFIC3 patients., Methods: The clinical and molecular genetic data of 13 new pediatric patients with PFIC3 as well as 82 reported ones in the PubMed and CNKI databases were collected and analyzed., Results: The 13 new PFIC3 patients included six females and seven males, and the main presentations were hepatomegaly, splenomegaly, jaundice, and pruritus, as well as increased levels of gamma-glutamyl transpeptidase (GGT). Fourteen new ABCB4 variants were detected, including eight diagnosed to be likely-pathogenic and six, pathogenic. Among all the 95 PFIC3 cases, hepatomegaly was observed in 85.3% (81/95), pruritus in 67.4% (64/95), splenomegaly in 52.6% (50/95), jaundice in 48.4% (46/95), portal hypertension in 34.7% (33/95) and GGT elevation in 100% (88/88) of the patients. Positive responses at varied degrees to oral ursodeoxycholic acid (UDCA) treatment were observed in 66.1% (39/59) of the patients, among whom 38.5% (15/39) fully recovered in terms of the laboratory changes. Although the condition remained stable in 53 patients (58.9%, 53/90), the clinical outcomes were not promising in the rest 37 cases (41.1%, 37/90), including 7 died, 27 having undergone while another 3 waiting for liver transplantation. A total of 96 ABCB4 variants were detected in the 95 patients. PFIC3 patients with biallelic null variants exhibited earlier onset ages [10.5 (2, 18) vs. 19 (8, 60) months, p = 0.007], lower UDCA response rate [18.2% (2/11) vs. 77.1% (37/48), p = 0.001], and more unpromising clinical outcomes [80% (12/15) vs. 33.3% (25/75), p = 0.001], compared with those with non-biallelic null variants., Conclusions: PFIC3 presented with hepatomegaly, pruritus, splenomegaly and jaundice with increased serum GGT level as a biochemistry hallmark. Although varying degrees of improvement in response to UDCA therapy were observed, 41.1% of PFIC3 patients exhibited unfavorable prognosis. ABCB4 genotypes of biallelic null variants were associated with severer PFIC3 phenotypes. Moreover, the 14 novel variants in this study expanded the ABCB4 mutation spectrum, and provided novel molecular biomarkers for diagnosis of PFIC3 patients., (© 2022. The Author(s).)

Background and Aims: Although the significance of primary congenital hypothyroidism (CH) is supported by an increasing amount of evidence, the clinical and genetic characteristics of this condition are still poorly understood. This study aimed to explore the underlying genetic etiologies in a cohort of primary CH patients., Subjects and Methods: The clinical data of 33 patients with primary CH were collected and analyzed via a cross-sectional study. Genetic analysis was performed by high-throughput sequencing and Sanger verification, and the pathogenicity of the novel missense variants was predicted using a variety of comprehensive bioinformatic tools., Results: Among the 33 patients, 22 (22/33, 66.7%) harbored pathogenic variants in the causative genes of thyroid dysgenesis or dyshormonogenesis, with DUOX2 (15/33, 45.5%) topping the list, followed by TG , TPO , DUOXA2 and PAX8 . Four novel genetic variants were detected, including a pathogenic frameshift and three likely pathogenic missense variants. Positive neonatal screening for TSH, neonatal jaundice and abnormal thyroid morphology were the main positive findings among all cases. Although 31 of the total 33 CH patients exhibited normal anthropometric and social performance, the other 2 had poor prognosis in this study., Conclusions: This study reported 33 new CH patients bearing four novel genetic variants, which enriched the variant spectrum of CH genes. In this cohort, genetic factors causing thyroid dyshormonogenesis were the main etiologies of CH development. Most patients exhibited a favorable prognosis; however, systematic management remains a challenge in achieving improved clinical outcomes for CH patients.

Fine-grained image analysis (FGIA) is a longstanding and fundamental problem in computer vision and pattern recognition, and underpins a diverse set of real-world applications. The task of FGIA targets analyzing visual objects from subordinate categories, e.g., species of birds or models of cars. The small inter-class and large intra-class variation inherent to fine-grained image analysis makes it a challenging problem. Capitalizing on advances in deep learning, in recent years we have witnessed remarkable progress in deep learning powered FGIA. In this paper we present a systematic survey of these advances, where we attempt to re-define and broaden the field of FGIA by consolidating two fundamental fine-grained research areas - fine-grained image recognition and fine-grained image retrieval. In addition, we also review other key issues of FGIA, such as publicly available benchmark datasets and related domain-specific applications. We conclude by highlighting several research directions and open problems which need further exploration from the community.

Fine-grained visual classification (FGVC) is much more challenging than traditional classification tasks due to the inherently subtle intra-class object variations. Recent works are mainly part-driven (either explicitly or implicitly), with the assumption that fine-grained information naturally rests within the parts. In this paper, we take a different stance, and show that part operations are not strictly necessary - the key lies with encouraging the network to learn at different granularities and progressively fusing multi-granularity features together. In particular, we propose: (i) a progressive training strategy that effectively fuses features from different granularities, and (ii) a consistent block convolution that encourages the network to learn the category-consistent features at specific granularities. We evaluate on several standard FGVC benchmark datasets, and demonstrate the proposed method consistently outperforms existing alternatives or delivers competitive results. Codes are available at https://github.com/PRIS-CV/PMG-V2.

Background: Infantile hypotonia with psychomotor retardation and characteristic facies type 3(IHPRF3) (OMIM #616,900) is an autosomal recessive disorder caused by biallelic pathogenic variants of the TBCK gene, and to date, this disease was reported rather limitedly in number and all described cases were Caucasians., Case Presentation: This paper reported the clinical and genetic features of a Chinese patient with IHPRF3. The patient was a 15-month-old male with global developmental delay, profound hypotonia, and typical facial dysmorphic features including mildly coarse facial appearance, hypertelorism, tented upper lip, exaggerated Cupid's bow, macroglossia and arched eyebrows. Magnetic Resonance Imaging (MRI) analysis of the brain revealed slightly widened bilateral ventricles and subarachnoid space. On genetic analysis, the patient was homozygous for a novel TBCK variant c.247C > T(p.Arg83Ter). The parents were both carriers without any positive symptoms or signs. With an extremely low frequency (0.0000082) in Exome Aggregation Consortium, the variant has not been reported in any other databases or official literatures, and was diagnosed to be pathogenic according to the American College of Medical Genetics and Genomics(ACMG) standards and guidelines. Neurorehabilitation training did not work well and the long-term prognosis remained to be observed., Conclusions: This study reported the clinical and molecular features of the first non-Caucasian patient with IHPRF3 arising from a novel homozygous TBCK mutation, which provided a novel molecular marker for the definite diagnosis of IHPRF3 patients and for its genetic counseling and prenatal diagnosis in the affected families., (© 2022. The Author(s).)

Objective: To screen and perform preliminary clinical validation of biomarkers of activity based on positron emission tomography/computed tomography (PET-CT) and transcriptomics in sputum-negative pulmonary tuberculosis lesion tissue. Methods: Nine patients with sputum-negative pulmonary tuberculosis treated surgically at the Shanghai Public Health Clinical Center for Thoracic Surgery from January 1, 2017 to December 31, 2019 were retrospectively collected as the discovery group, including four males and five females, aged 20-57 years (mean 36 years). All of the patients underwent PET-CT scanning before surgery, and the resected specimens were postoperatively classified according to preoperative PET-CT. The resected specimens were divided into areas with increased fluorodeoxyglucose (FDG) metabolism (SUVmax>3) and areas with normal FDG metabolism (SUVmax ≤ 3) according to the preoperative PET-CT performance. After sample processing, total RNA was extracted from the tissues of different regions, and then whole gene transcriptome sequencing was performed. Bioinformatics analysis of the two sets of data was performed to discover the expression profiles of the differences in whole gene transcriptome data between the two regions and to screen for candidate biomarkers. Eighty patients with sputum-negative pulmonary tuberculosis admitted to Shanghai Public Health Clinical Center from January 1, 2019 to January 1, 2021 were retrospectively collected as the validation group, including 37 males and 43 females, aged 20-62 years, with an average age of 39 years. The validation group was divided into a group with increased SUV ( n =40) and a group without lesions on CT imaging ( n =40). Enzyme-linked immunosorbent assay (ELISA) was used to determine the protein levels of candidate biomarkers in the peripheral plasma of patients. The effect of biomarkers was assessed using subject operating characteristic (ROC) curves. Student's t -test was used to determine whether the difference in protein levels between the two groups was statistically significant. Results: Bioinformatics analysis revealed that the expression levels of C1QB , CCL19 , CCL5 and HLA-DMB correlated with the metabolic activity of sputum-negative tuberculosis lesion tissue. Further screening and validation by the validation group confirmed that the difference in C1QB protein levels in the peripheral plasma of patients was statistically significant between the group with increased SUV and the group without lesions on CT imaging [(3.55±0.34) mg/L vs. (2.75±0.21) mg/L, t =4.12, P <0.001]. And the ROC curve showed that the area under the curve for C1QB protein levels was 0.731, which had potential clinical value. Conclusion: The C1QB protein level can be used to assess the activity of lesions in patients with sputum-negative tuberculosis and is a potential biomarker.

Na + -taurocholate cotransporting polypeptide deficiency (NTCPD) is an autosomal recessive disorder arising from biallelic SLC10A1 mutations. As a newly-described inborn error of bile acid metabolism, the epidemiology of this condition remains largely unclear in Chinese population so far. In this study, a total of 2,828 peripheral blood samples were collected from 12 cities in Guangdong, a province with the largest population in China, and the four prevalent SLC10A1 variants c.800C > T (p.Ser267Phe), c.263T > C (p.Ile88Thr), c.595A > C (p.Ser199Arg) and c.665T > C (p.Leu222Ser) were screened for by using polymerase chain reaction (PCR)- restriction fragment length polymorphism (RFLP). As a result, 663 mutated SLC10A1 alleles were detected, and the mutated allele frequency was calculated to be 11.72% (663/5,656), with a carrier frequency 20.69% (1/5) and a theoretical morbidity rate 1.37% (1/73) of NTCPD in Guangdong province. The variant c.800C > T (p.Ser267Phe) exhibited highest allele frequency among the four prevalent variants (χ 2 = 1501.27, p < 0.0001) as well as higher allele frequency in the peripheral region than that within the Pearl River Delta (χ 2 = 4.834, p < 0.05). The results suggested that NTCPD might be a disorder rather common in Guangdong province. The findings depicted the molecular epidemiologic features of NTCPD, providing preliminary but significant laboratory evidences for the subsequent NTCPD diagnosis and management in Guangdong population., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2022 Li, Chen, Lin, Lin, Yaqub and Song.)

In this paper, we aim to explore the fine-grained perception ability of deep models for the newly proposed scene sketch semantic segmentation task. Scene sketches are abstract drawings containing multiple related objects. It plays a vital role in daily communication and human-computer interaction. The study has only recently started due to a main obstacle of the absence of large-scale datasets. The currently available dataset SketchyScene is composed of clip art-style edge maps, which lacks abstractness and diversity. To drive further research, we contribute two new large-scale datasets based on real hand-drawn object sketches. A general automatic scene sketch synthesis process is developed to assist with new dataset composition. Furthermore, we propose to enhancing local detail perception in deep models to realize accurate stroke-oriented scene sketch segmentation. Due to the inherent differences between hand-drawn sketches and natural images, extreme low-level local features of strokes are incorporated to improve detail discrimination. Stroke masks are also integrated into model training to guide the learning attention. Extensive experiments are conducted on three large-scale scene sketch datasets. Our method achieves state-of-the-art performance under four evaluation metrics and yields meaningful interpretability via visual analytics.

We present the first one-shot personalized sketch segmentation method. We aim to segment all sketches belonging to the same category provisioned with a single sketch with a given part annotation while (i) preserving the parts semantics embedded in the exemplar, and (ii) being robust to input style and abstraction. We refer to this scenario as personalized. With that, we importantly enable a much-desired personalization capability for downstream fine-grained sketch analysis tasks. To train a robust segmentation module, we deform the exemplar sketch to each of the available sketches of the same category. Our method generalizes to sketches not observed during training. Our central contribution is a sketch-specific hierarchical deformation network. Given a multi-level sketch-strokes encoding obtained via a graph convolutional network, our method estimates rigid-body transformation from the target to the exemplar, on the upper level. Finer deformation from the exemplar to the globally warped target sketch is further obtained through stroke-wise deformations, on the lower-level. Both levels of deformation are guided by mean squared distances between the keypoints learned without supervision, ensuring that the stroke semantics are preserved. We evaluate our method against the state-of-the-art segmentation and perceptual grouping baselines re-purposed for the one-shot setting and against two few-shot 3D shape segmentation methods. We show that our method outperforms all the alternatives by more than 10% on average. Ablation studies further demonstrate that our method is robust to personalization: changes in input part semantics and style differences.

Most existing studies on unsupervised domain adaptation (UDA) assume that each domain's training samples come with domain labels (e.g., painting, photo). Samples from each domain are assumed to follow the same distribution and the domain labels are exploited to learn domain-invariant features via feature alignment. However, such an assumption often does not hold true-there often exist numerous finer-grained domains (e.g., dozens of modern painting styles have been developed, each differing dramatically from those of the classic styles). Therefore, forcing feature distribution alignment across each artificially-defined and coarse-grained domain can be ineffective. In this paper, we address both single-source and multi-source UDA from a completely different perspective, which is to view each instance as a fine domain. Feature alignment across domains is thus redundant. Instead, we propose to perform dynamic instance domain adaptation (DIDA). Concretely, a dynamic neural network with adaptive convolutional kernels is developed to generate instance-adaptive residuals to adapt domain-agnostic deep features to each individual instance. This enables a shared classifier to be applied to both source and target domain data without relying on any domain annotation. Further, instead of imposing intricate feature alignment losses, we adopt a simple semi-supervised learning paradigm using only a cross-entropy loss for both labeled source and pseudo labeled target data. Our model, dubbed DIDA-Net, achieves state-of-the-art performance on several commonly used single-source and multi-source UDA datasets including Digits, Office-Home, DomainNet, Digit-Five, and PACS.

Na + -taurocholate cotransporting polypeptide deficiency (NTCPD) is a newly described disorder arising from biallelic mutations of the SLC10A1 gene. As a result of a lack of compelling evidence from case-control studies, its genotypic and phenotypic features remain open for in-depth investigation. This study aimed to explore the genotypic and clinical phenotypic characteristics of paediatric patients with NTCPD. The SLC10A1 genotypes of all NTCPD patients were confirmed by screening for the prevalent variant c.800C>T and Sanger sequencing when necessary. The clinical presentations and laboratory changes were collected, reviewed and analysed, and then qualitatively and quantitatively compared with the relevant controls. A total of 113 paediatric NTCPD patients were diagnosed while c.374dupG and c.682&#95;683delCT were detected as two novel pathogenic mutations. Hypercholanemia was observed in 99.12% of the patients. Indirect hyperbilirubinemia in affected neonates exhibited higher positive rates in comparison to controls. Moreover, transient cholestatic jaundice, elevated liver enzymes and 25-hydroxyvitamin D (Vit D) deficiency during early infancy were more commonly observed in patients than in controls. All NTCPD patients exhibited favourable clinical outcomes as a result of symptomatic and supportive treatment. The findings enriched the SLC10A1 mutation spectrum and provided comprehensive insights into the phenotypic characteristics of NTCPD. NTCPD should be considered and SLC10A1 gene should be analysed in patients with above age-dependent clinical features. Furthermore, over investigation and intervention should be avoided in the management of NTCPD patients., (© 2021 The Authors. Liver International published by John Wiley & Sons Ltd.)