BioSAM: Generating SAM Prompts From Superpixel Graph for Biological Instance Segmentation.

Proposal-free instance segmentation methods have significantly advanced the field of biological image analysis. Recently, the Segment Anything Model (SAM) has shown an extraordinary ability to handle challenging instance boundaries. However, directly applying SAM to biological images that contain instances with complex morphologies and dense distributions fails to yield satisfactory results. In this work, we propose BioSAM, a new biological instance segmentation framework generating SAM prompts from a superpixel graph. Specifically, to avoid over-merging, we first generate sufficient superpixels as graph nodes and construct an initialized graph. We then generate initial prompts from each superpixel and aggregate them through a graph neural network (GNN) by predicting the relationship of superpixels to avoid over-segmentation. We employ the SAM encoder embeddings and the SAM-assisted superpixel similarity as new features for the graph to enhance its discrimination capability. With the graph-based prompt aggregation, we utilize the aggregated prompts in SAM to refine the segmentation and generate more accurate instance boundaries. Comprehensive experiments on four representative biological datasets demonstrate that our proposed method outperforms state-of-the-art methods.

Diversity and Evolution of NLR Genes in Citrus Species.

NLR genes are crucial components of the effector-triggered immunity (ETI) system, responsible for recognizing pathogens and initiating immune responses. Although NLR genes in many plant species have been extensively studied, the diversity of NLR genes in citrus remains largely unknown. Our analysis revealed significant variations in the copy numbers of NLR genes among these species. Gene duplication and recombination were identified as the major driving forces behind this diversity. Additionally, horizontal gene transfer (HGT) emerged as the principal mechanism responsible for the increase in NLR gene copy number in A. buxifolia. The citrus NLR genes were classified into four categories: TIR-NBS-LRR (TNL), CC-NBS-LRR (CNL), RPW8-NBS-LRR (RNL), and NL. Our findings indicate that TNL, RNL, and CNL genes originated from NL genes through the acquisition of TIR and RPW8 domains, along with CC motifs, followed by the random loss of corresponding domains. Phylogenetic analysis suggested that citrus NLR genes originated alongside the species and underwent adaptive evolution, potentially playing crucial roles in the global colonization of citrus. This study provides important insights into the diversity of citrus NLR genes and serves as a foundational dataset for future research aimed at breeding disease-resistant citrus varieties.

Neural Degradation Representation Learning for All-in-One Image Restoration.

Existing methods have demonstrated effective performance on a single degradation type. In practical applications, however, the degradation is often unknown, and the mismatch between the model and the degradation will result in a severe performance drop. In this paper, we propose an all-in-one image restoration network that tackles multiple degradations. Due to the heterogeneous nature of different types of degradations, it is difficult to process multiple degradations in a single network. To this end, we propose to learn a neural degradation representation (NDR) that captures the underlying characteristics of various degradations. The learned NDR adaptively decomposes different types of degradations, similar to a neural dictionary that represents basic degradation components. Subsequently, we develop a degradation query module and a degradation injection module to effectively approximate and utilize the specific degradation based on NDR, enabling the all-in-one restoration ability for multiple degradations. Moreover, we propose a bidirectional optimization strategy to effectively drive NDR to learn the degradation representation by optimizing the degradation and restoration processes alternately. Comprehensive experiments on representative types of degradations (including noise, haze, rain, and downsampling) demonstrate the effectiveness and generalizability of our method. Code is available at https://github.com/mdyao/NDR-Restore.

Stimulating Diffusion Model for Image Denoising via Adaptive Embedding and Ensembling.

Image denoising is a fundamental problem in computational photography, where achieving high perception with low distortion is highly demanding. Current methods either struggle with perceptual quality or suffer from significant distortion. Recently, the emerging diffusion model has achieved state-of-the-art performance in various tasks and demonstrates great potential for image denoising. However, stimulating diffusion models for image denoising is not straightforward and requires solving several critical problems. For one thing, the input inconsistency hinders the connection between diffusion models and image denoising. For another, the content inconsistency between the generated image and the desired denoised image introduces distortion. To tackle these problems, we present a novel strategy called the Diffusion Model for Image Denoising (DMID) by understanding and rethinking the diffusion model from a denoising perspective. Our DMID strategy includes an adaptive embedding method that embeds the noisy image into a pre-trained unconditional diffusion model and an adaptive ensembling method that reduces distortion in the denoised image. Our DMID strategy achieves state-of-the-art performance on both distortion-based and perception-based metrics, for both Gaussian and real-world image denoising. The code is available at https://github.com/Li-Tong-621/DMID.

Unsupervised Domain Adaptation for EM Image Denoising with Invertible Networks.

Electron microscopy (EM) image denoising is critical for visualization and subsequent analysis. Despite the remarkable achievements of deep learning-based non-blind denoising methods, their performance drops significantly when domain shifts exist between the training and testing data. To address this issue, unpaired blind denoising methods have been proposed. However, these methods heavily rely on image-to-image translation and neglect the inherent characteristics of EM images, limiting their overall denoising performance. In this paper, we propose the first unsupervised domain adaptive EM image denoising method, which is grounded in the observation that EM images from similar samples share common content characteristics. Specifically, we first disentangle the content representations and the noise components from noisy images and establish a shared domain-agnostic content space via domain alignment to bridge the synthetic images (source domain) and the real images (target domain). To ensure precise domain alignment, we further incorporate domain regularization by enforcing that: the pseudo-noisy images, reconstructed using both content representations and noise components, accurately capture the characteristics of the noisy images from which the noise components originate, all while maintaining semantic consistency with the noisy images from which the content representations originate. To guarantee lossless representation decomposition and image reconstruction, we introduce disentanglement-reconstruction invertible networks. Finally, the reconstructed pseudo-noisy images, paired with their corresponding clean counterparts, serve as valuable training data for the denoising network. Extensive experiments on synthetic and real EM datasets demonstrate the superiority of our method in terms of image restoration quality and downstream neuron segmentation accuracy. Our code is publicly available at https://github.com/sydeng99/DADn.

Non-Serial Quantization-Aware Deep Optics for Snapshot Hyperspectral Imaging.

Deep optics has been endeavoring to capture hyperspectral images of dynamic scenes, where the optical encoder plays an essential role in deciding the imaging performance. Our key insight is that the optical encoder of a deep optics system is expected to keep fabrication-friendliness and decoder-friendliness, to be faithfully realized in the implementation phase and fully interacted with the decoder in the design phase, respectively. In this paper, we propose the non-serial quantization-aware deep optics (NSQDO), which consists of the fabrication-friendly quantization-aware model (QAM) and the decoder-friendly non-serial manner (NSM). The QAM integrates the quantization process into the optimization and adaptively adjusts the physical height of each quantization level, reducing the deviation of the physical encoder from the numerical simulation through the awareness of and adaptation to the quantization operation of the DOE physical structure. The NSM bridges the encoder and the decoder with full interaction through bidirectional hint connections and flexibilize the connections with a gating mechanism, boosting the power of joint optimization in deep optics. The proposed NSQDO improves the fabrication-friendliness and decoder-friendliness of the encoder and develops the deep optics framework to be more practical and powerful. Extensive synthetic simulation and real hardware experiments demonstrate the superior performance of the proposed method.

Effects of Simvastatin on Inflammatory Response and Biological Behaviour of Adamantinomatous Craniopharyngioma.

INTRODUCTION: The aim of this study was to investigate the autoinflammatory effect and biological behaviour of simvastatin (SIM) on adamantinomatous craniopharyngioma (ACP) cells. METHODS: Craniopharyngiomas imaging, intraoperative observations, and tumour histopathology were employed to investigate the correlation between esters and craniopharyngiomas. Filipin III fluorescent probe verified the validity of SIM on the alternations of synthesized cholesterol in craniopharyngioma cells. The cell counting kit-8 (CCK8) assay detected the impacts of SIM on cell proliferation and determined the IC50 value of tumour cells. Reverse transcription polymerase chain reaction (RT-PCR) measured the expression of inflammatory factors. Flow cytometry technique detected the cell cycle and apoptosis, and cell scratch assay judged the cell migration. Meanwhile, Western blot was adopted to determine the expression of proteins related to inflammation, proliferation, and apoptosis signalling pathways. RESULTS: In the ACP tumour parenchyma, many cholesterol crystalline clefts were observed, and the deposition of esters was closely associated with craniopharyngioma inflammation. After SIM intervention, a reduction in cholesterol synthesis within ACP was noted. RT-PCR analysis revealed SIM inhibited the transcription of inflammatory factors in ACP cells. Western blot analysis demonstrated SIM inhibited nuclear factor-kappa B p65 activation expression while promoted the expressions of Cl-caspase-3 and P38 MAPK. CCK8 assay indicated a decrease in ACP cell activity upon SIM treatment. Scratch assay signified that SIM hindered ACP cell migration. Flow cytometry results suggested that the drug promoted ACP cell apoptosis. CONCLUSION: SIM suppressed the inflammatory response to craniopharyngiomas by inhibiting craniopharyngioma cholesterol synthesis, inhibited proliferation of ACP cells, and promoted their apoptosis.

Alterations of plasma circulating microRNAs in BALB/c mice with Toxocara canis visceral and cerebral larva migrans.

BACKGROUND: Human toxocariasis is a neglected parasitic disease characterised by the syndromes visceral, cerebral, and ocular larva migrans. This disease is caused by the migrating larvae of Toxocara roundworms from dogs and cats, affecting 1.4 billion people globally. Via extracellular vesicles (EVs), microRNAs have been demonstrated to play roles in host-parasite interactions and proposed as circulating biomarkers for the diagnosis and follow-up of parasitic diseases. METHODS: Small RNA-seq was conducted to identify miRNAs in the infective larvae of T. canis and plasma EV-containing preparations of infected BALB/c mice. Differential expression analysis and target prediction were performed to indicate miRNAs involved in host-parasite interactions and miRNAs associated with visceral and/or cerebral larva migrans in the infected mice. Quantitative real-time polymerase chain reaction (PCR) was used to amplify circulating miRNAs from the infected mice. RESULTS: This study reports host and parasite miRNAs in the plasma of BALB/c mice with visceral and cerebral larva migrans and demonstrates the alterations of these miRNAs during the migration of larvae from the livers through the lungs and to the brains of infected mice. After filtering unspecific changes in an irrelevant control, T. canis-derived miRNAs and T. canis infection-induced differential miRNAs are predicted to modulate genes consistently involved in mitogen-activated protein kinase (MAPK) signalling and pathways regulating axon guidance and pluripotency of stem in the infected mice with visceral and cerebral larva migrans. For these plasma circulating miRNAs predicted to be involved in host-parasite crosstalk, two murine miRNAs (miR-26b-5p and miR-122-5p) are experimentally verified to be responsive to larva migrans and represent circulating biomarker candidates for visceral and cerebral toxocariasis in BALB/c mice. CONCLUSIONS: Our findings provide novel insights into the crosstalk of T. canis and the mammalian host via plasma circulating miRNAs, and prime agents and indicators for visceral and cerebral larva migrans. A deep understanding of these aspects will underpin the diagnosis and control of toxocariasis in humans and animals.

Oxytocin attenuates hypothalamic injury-induced cognitive dysfunction by inhibiting hippocampal ERK signaling and Aß deposition.

In clinical settings, tumor compression, trauma, surgical injury, and other types of injury can cause hypothalamic damage, resulting in various types of hypothalamic dysfunction. Impaired release of oxytocin can lead to cognitive impairment and affect prognosis and long-term quality of life after hypothalamic injury. Hypothalamic injury-induced cognitive dysfunction was detected in male animals. Behavioral parameters were measured to assess the characteristics of cognitive dysfunction induced by hypothalamic-pituitary stalk lesions. Brains were collected for high-throughput RNA sequencing and immunostaining to identify pathophysiological changes in hippocampal regions highly associated with cognitive function after injury to corresponding hypothalamic areas. Through transcriptomic analysis, we confirmed the loss of oxytocin neurons after hypothalamic injury and the reversal of hypothalamic-induced cognitive dysfunction after oxytocin supplementation. Furthermore, overactivation of the ERK signaling pathway and ß-amyloid deposition in the hippocampal region after hypothalamic injury were observed, and cognitive function was restored after inhibition of ERK signaling pathway overactivation. Our findings suggest that cognitive dysfunction after hypothalamic injury may be caused by ERK hyperphosphorylation in the hippocampal region resulting from a decrease in the number of oxytocin neurons, which in turn causes ß-amyloid deposition.

WASPSYN: A Challenge for Domain Adaptive Synapse Detection in Microwasp Brain Connectomes.

The size of image volumes in connectomics studies now reaches terabyte and often petabyte scales with a great diversity of appearance due to different sample preparation procedures. However, manual annotation of neuronal structures (e.g., synapses) in these huge image volumes is time-consuming, leading to limited labeled training data often smaller than 0.001% of the large-scale image volumes in application. Methods that can utilize in-domain labeled data and generalize to out-of-domain unlabeled data are in urgent need. Although many domain adaptation approaches are proposed to address such issues in the natural image domain, few of them have been evaluated on connectomics data due to a lack of domain adaptation benchmarks. Therefore, to enable developments of domain adaptive synapse detection methods for large-scale connectomics applications, we annotated 14 image volumes from a biologically diverse set of Megaphragma viggianii brain regions originating from three different whole-brain datasets and organized the WASPSYN challenge at ISBI 2023. The annotations include coordinates of pre-synapses and post-synapses in the 3D space, together with their one-to-many connectivity information. This paper describes the dataset, the tasks, the proposed baseline, the evaluation method, and the results of the challenge. Limitations of the challenge and the impact on neuroscience research are also discussed. The challenge is and will continue to be available at https://codalab.lisn.upsaclay.fr/competitions/9169. Successful algorithms that emerge from our challenge may potentially revolutionize real-world connectomics research and further the cause that aims to unravel the complexity of brain structure and function.

Global research hotspots and trends of acute rejection after liver transplantation from 1988 to 2022: a bibliometric analysis.

Background: Acute rejection (AR) is the predominant form of rejection observed in liver transplantation and plays a crucial role in transplant immunology. This study aims to utilize bibliometric analysis to understand the status quo, hotspots, and future trends of research on AR after liver transplantation. Methods: We searched the Web of Science Core Collection (WoSCC) for studies on AR after liver transplantation published from 1988 to 2022. The Bibliometric Online Analysis Platform, VOSviewer, and CiteSpace were used for analysis of all extracted publications. Results: This study included 2,398 articles published in 456 journals by 12,568 authors from 1,965 institutions in 55 countries/regions. The United States and its affiliated institution, the University of Pittsburgh, were the most productive contributors. Transplantation (n = 12,435) was the most frequently cited journal. Neuhaus P (n = 38) was the highest output author, and Demetris AJ (n = 670) was the most co-cited author. The research hotspots of AR after liver transplantation include pathogenesis, immunosuppressive therapy, and prognosis. Emerging research directions include regulatory T cells, immunosuppression minimization, intra-patient variability (IPV) of tacrolimus, and novel non-invasive diagnostic markers. Conclusion: Our study utilized bibliometric methods to analyze the study of AR after liver transplantation over the past 35 years. With the prolonged survival of liver transplant recipients, the most active areas currently focus on individualized treatment and improving patient prognosis. Minimizing adverse reactions to immunosuppressive therapy while simultaneously avoiding an increase in the risk of AR remains a future research focus.

Identification and phylogenetic analysis of Jingmen tick virus in Jiangxi Province, China.

Jingmen tick virus (JMTV) is a newly identified segmented flavivirus that has been recognized in multiple hosts, such as humans, buffalos, bats, rodents, mosquitos and ticks. Various clinical cases and studies manifested that JMTV is a true arbovirus with wide host spectrum and showed potential threats toward public health. JMTV has been reported in multiple countries in Asia, Europe, Africa, and America. Moreover, wild boars serve as an important intermediary between humans and the wild ecological system. In China, it has been reported in nine provinces, while the prevalence and the distribution of JMTV in most regions including Jiangxi Province are still unknown. Thus, to profile the distribution of JMTV in Jiangxi Province, an epidemiological investigation was carried out from 2020 to 2022. In current study, 66 ticks were collected from 17 wild boars in Jiangxi Province. The results showed that 12 out of 66 ticks were JMTV positive, indicating JMTV is prevalent in ticks and boars in Jiangxi Province. The genome sequences of JMTV strain WY01 were sequenced to profile viral evolution of JMTV in China. Phylogenetic analysis divided JMTV strains into two genotypes, Group I and Group II. WY01 belongs to Group II and it shares the closest evolutionary relationship with the Japan strains rather than the strains from neighboring provinces in China suggesting that JMTV might have complex transmission routes. Overall, current study, for the first time, reported that JMTV is prevalent in Jiangxi Province and provided additional information concerning JMTV distribution and evolution in China.

Toward DNN of LUTs: Learning Efficient Image Restoration with Multiple Look-Up Tables.

The widespread usage of high-definition screens on edge devices stimulates a strong demand for efficient image restoration algorithms. The way of caching deep learning models in a look-up table (LUT) is recently introduced to respond to this demand. However, the size of a single LUT grows exponentially with the increase of its indexing capacity, which restricts its receptive field and thus the performance. To overcome this intrinsic limitation of the single-LUT solution, we propose a universal method to construct multiple LUTs like a neural network, termed MuLUT. Firstly, we devise novel complementary indexing patterns, as well as a general implementation for arbitrary patterns, to construct multiple LUTs in parallel. Secondly, we propose a re-indexing mechanism to enable hierarchical indexing between cascaded LUTs. Finally, we introduce channel indexing to allow cross-channel interaction, enabling LUTs to process color channels jointly. In these principled ways, the total size of MuLUT is linear to its indexing capacity, yielding a practical solution to obtain superior performance with the enlarged receptive field. We examine the advantage of MuLUT on various image restoration tasks, including super-resolution, demosaicing, denoising, and deblocking. MuLUT achieves a significant improvement over the single-LUT solution, e.g., up to 1.1dB PSNR for super-resolution and up to 2.8dB PSNR for grayscale denoising, while preserving its efficiency, which is 100× less in energy cost compared with lightweight deep neural networks. Our code and trained models are publicly available at https://github.com/ddlee-cn/MuLUT.

Growth hormone promotes the reconstruction of injured axons in the hypothalamo-neurohypophyseal system.

JOURNAL/nrgr/04.03/01300535-202410000-00026/figure1/v/2024-02-06T055622Z/r/image-tiff Previous studies have shown that growth hormone can regulate hypothalamic energy metabolism, stress, and hormone release. Therefore, growth hormone has great potential for treating hypothalamic injury. In this study, we established a specific hypothalamic axon injury model by inducing hypothalamic pituitary stalk electric lesions in male mice. We then treated mice by intraperitoneal administration of growth hormone. Our results showed that growth hormone increased the expression of insulin-like growth factor 1 and its receptors, and promoted the survival of hypothalamic neurons, axonal regeneration, and vascular reconstruction from the median eminence through the posterior pituitary. Altogether, this alleviated hypothalamic injury-caused central diabetes insipidus and anxiety. These results suggest that growth hormone can promote axonal reconstruction after hypothalamic injury by regulating the growth hormone-insulin-like growth factor 1 axis.

Correlation between white blood cell count and intestinal resection in patients with acute mesenteric vein thrombosis.

OBJECTIVE: Acute mesenteric vein thrombosis (AMVT) is an acute abdominal disease with onset, rapid progression, and extensive intestinal necrosis that requires immediate surgical resection. The purpose of this study was to determine the risk factors for nosocomial intestinal resection in patients with AMVT. METHODS: We retrospectively analysed 64 patients with AMVT diagnosed by CTA at the Affiliated Hospital of Kunming University of Science and Technology from January 2013 to December 2021. We compared patients who underwent intestinal resection (42 patients) with those who did not undergo intestinal resection (22 patients). The area under the ROC curve was evaluated, and a forest map was drawn. RESULTS: Among the 64 patients, 6 (9.38%) had a fever, 60 (93.75%) had abdominal pain, 9 (14.06%) had a history of diabetes, 8 (12.5%) had a history of deep vein thrombosis (DVT), and 25 (39.06%) had ascites suggested by B ultrasound or CT after admission. The mean age of all patients was 49.86 ± 16.25 years. The mean age of the patients in the enterectomy group was 47.71 ± 16.20 years. The mean age of the patients in the conservative treatment group (without enterectomy) was 53.95 ± 15.90 years. In the univariate analysis, there were statistically significant differences in leukocyte count (P = 0.003), neutrophil count (P = 0.001), AST (P = 0.048), total bilirubin (P = 0.047), fibrinogen (P = 0.022) and DD2 (P = 0.024) between the two groups. The multivariate logistic regression analysis showed that admission white blood cell count (OR = 1.153, 95% CI: 1.039-1.280, P = 0.007) was an independent risk factor for intestinal resection in patients with AMVT. The ROC curve showed that the white blood cell count (AUC = 0.759 95% CI: 0.620-0.897; P = 0.001; optimal threshold: 7.815; sensitivity: 0.881; specificity: 0.636) had good predictive value for emergency enterectomy for AMVT. CONCLUSIONS: Among patients with AMVT, patients with a higher white blood cell count at admission were more likely to have intestinal necrosis and require emergency enterectomy. This study is helpful for clinicians to accurately determine whether emergency intestinal resection is needed in patients with AMVT after admission, prevent further intestinal necrosis, and improve the prognosis of patients.

A bibliometric analysis of macrophages associated with non-alcoholic fatty liver disease research from 2005 to 2023.

Non-alcoholic fatty liver disease (NAFLD) is a prevalent chronic liver condition associated with the risk of progressing to decompensated cirrhosis and hepatocellular carcinoma. While macrophages play a crucial role in the development of NAFLD, their heterogeneity and plasticity allow them to undertake diverse roles in immune response, tissue repair, and maintaining tissue homeostasis. Thus, the exact involvement of macrophages in the onset and progression of NAFLD remains to be further explored. This study aims to employ bibliometric analysis to elucidate the role of macrophages in the pathogenesis of NAFLD, analyze research focal points in this domain, and speculate on future research trends. The literature search, conducted using the Web of Science Core Collection, encompassed articles and reviews related to macrophages and NAFLD published between 2005 and 2023. A bibliometric analysis of 1264 extracted publications was performed using VOSviewer 1.6.17 and Citespace 6.1. R2, evaluating parameters such as spatial and temporal distribution, authors, thematic categories, topic distribution, references, and keywords. The findings revealed a steady global increase in publications in this field, with the United States contributing the most followed by China. The University of California System produced the highest volume of publications, while the Journal of Hepatology had the highest impact factors among the top 10 publishing journals. Tacke Frank emerged as both the most prolific author and the most cited. Co-occurrence and burst analysis of keywords and references highlighted the hotspots in this research area, emphasizing the mechanisms of NAFLD pathogenesis, metabolic regulation, immune modulation, and oxidative stress. Maintaining hepatic homeostasis by liver macrophages and macrophage polarization were identified as trending research directions in this field. Based on the bibliometric analysis, continued attention toward NAFLD therapeutic research involving hepatic macrophages is anticipated. As the mechanisms underlying NAFLD pathogenesis are further elucidated, the development of more treatment approaches related to macrophage immunology and metabolic regulation may expand therapeutic options. This study offers valuable insights into the current state and future trends in the field, providing beneficial guidance to researchers aiming to make significant contributions.

Chromosome-level genome assembly of navel orange cv. Gannanzao (Citrus sinensis Osbeck cv. Gannanzao).

Navel orange cv. Gannanzao is a variant of the navel orange cv. Newhall (Citrus sinensis Osbeck cv. Newhall) that exhibits an earlier maturation, making it commercially valuable. However, the mechanisms underlying its early maturation remain obscure. To address this question, we conducted genome sequencing and de novo assembly of navel orange cv. Gannanzao. The assembled genome sequence is 334.57 Mb in length with a GC content of 31.48%. It comprises 318 contigs (N50 = 3.23 Mb) and 187 scaffolds (N50 = 31.86 Mb). The Benchmarking Universal Single-Copy Orthologs test demonstrates 94.6% completeness. The annotation revealed 23,037 gene models, 164.95 Mb of repetitive sequences, and 2,554 noncoding RNAs. A comparative analysis identified 323 fruit ripening-related genes in navel orange cv. Gannanzao genome, while navel orange cv. Newhall genome contained 345 such genes. These genes were organized into 320 orthologous gene families, with 30.3% of them exhibiting differences in gene copy numbers between the 2 genomes. Additionally, we identified 15 fruit ripening-related genes that have undergone adaptive evolution, suggesting their potential role in advancing fruit maturation in navel orange cv. Gannanzao. Whole-genome sequencing and annotation of navel orange cv. Gannanzao provides a valuable resource to unravel the early maturation mechanism of citrus and enriches the genomic resources for citrus research.

Erratum: Citrus genomic resources unravel putative genetic determinants of Huanglongbing pathogenicity.

[This corrects the article DOI: 10.1016/j.isci.2023.106024.].

Current Progress and Challenges in Large-Scale 3D Mitochondria Instance Segmentation.

In this paper, we present the results of the MitoEM challenge on mitochondria 3D instance segmentation from electron microscopy images, organized in conjunction with the IEEE-ISBI 2021 conference. Our benchmark dataset consists of two large-scale 3D volumes, one from human and one from rat cortex tissue, which are 1,986 times larger than previously used datasets. At the time of paper submission, 257 participants had registered for the challenge, 14 teams had submitted their results, and six teams participated in the challenge workshop. Here, we present eight top-performing approaches from the challenge participants, along with our own baseline strategies. Posterior to the challenge, annotation errors in the ground truth were corrected without altering the final ranking. Additionally, we present a retrospective evaluation of the scoring system which revealed that: 1) challenge metric was permissive with the false positive predictions; and 2) size-based grouping of instances did not correctly categorize mitochondria of interest. Thus, we propose a new scoring system that better reflects the correctness of the segmentation results. Although several of the top methods are compared favorably to our own baselines, substantial errors remain unsolved for mitochondria with challenging morphologies. Thus, the challenge remains open for submission and automatic evaluation, with all volumes available for download.

Space-Time Super-Resolution for Light Field Videos.

Light field (LF) cameras suffer from a fundamental trade-off between spatial and angular resolutions. Additionally, due to the significant amount of data that needs to be recorded, the Lytro ILLUM, a modern LF camera, can only capture three frames per second. In this paper, we consider space-time super-resolution (SR) for LF videos, aiming at generating high-resolution and high-frame-rate LF videos from low-resolution and low-frame-rate observations. Extending existing space-time video SR methods to this task directly will meet two key challenges: 1) how to re-organize sub-aperture images (SAIs) efficiently and effectively given highly redundant LF videos, and 2) how to aggregate complementary information between multiple SAIs and frames considering the coherence in LF videos. To address the above challenges, we propose a novel framework for space-time super-resolving LF videos for the first time. First, we propose a novel Multi-Scale Dilated SAI Re-organization strategy for re-organizing SAIs into auxiliary view stacks with decreasing resolution as the Chebyshev distance in the angular dimension increases. In particular, the auxiliary view stack with original resolution preserves essential visual details, while the down-scaled view stacks capture long-range contextual information. Second, we propose the Multi-Scale Aggregated Feature extractor and the Angular-Assisted Feature Interpolation module to utilize and aggregate information from the spatial, angular, and temporal dimensions in LF videos. The former aggregates similar contents from different SAIs and frames for subsequent reconstruction in a disparity-free manner at the feature level, whereas the latter interpolates intermediate frames temporally by implicitly aggregating geometric information. Compared to other potential approaches, experimental results demonstrate that the reconstructed LF videos generated by our framework achieve higher reconstruction quality and better preserve the LF parallax structure and temporal consistency. The implementation code is available at https://github.com/zeyuxiao1997/LFSTVSR.