Enhancing genome-wide populus trait prediction through deep convolutional neural networks.

As a promising model, genome-based plant breeding has greatly promoted the improvement of agronomic traits. Traditional methods typically adopt linear regression models with clear assumptions, neither obtaining the linkage between phenotype and genotype nor providing good ideas for modification. Nonlinear models are well characterized in capturing complex nonadditive effects, filling this gap under traditional methods. Taking populus as the research object, this paper constructs a deep learning method, DCNGP, which can effectively predict the traits including 65 phenotypes. The method was trained on three datasets, and compared with other four classic models-Bayesian ridge regression (BRR), Elastic Net, support vector regression, and dualCNN. The results show that DCNGP has five typical advantages in performance: strong prediction ability on multiple experimental datasets; the incorporation of batch normalization layers and Early-Stopping technology enhancing the generalization capabilities and prediction stability on test data; learning potent features from the data and thus circumventing the tedious steps of manual production; the introduction of a Gaussian Noise layer enhancing predictive capabilities in the case of inherent uncertainties or perturbations; fewer hyperparameters aiding to reduce tuning time across datasets and improve auto-search efficiency. In this way, DCNGP shows powerful predictive ability from genotype to phenotype, which provide an important theoretical reference for building more robust populus breeding programs.

Evaluation of the synergistic effect of eravacycline and tigecycline against carbapenemase-producing carbapenem-resistant Klebsiella pneumoniae.

BACKGROUND: Carbapenem-resistant Klebsiella pneumoniae (CRKP) poses a substantial healthcare challenge. This study assessed the in vitro efficacy of selected antibiotic combinations against CRKP infections. METHODS: Our research involved the evaluation of 40 clinical isolates of CRKP, with half expressing Klebsiella pneumoniae carbapenemase (KPC) and half producing Metallo-ß-lactamase (MBL), two key enzymes contributing to carbapenem resistance. We determined the minimum inhibitory concentrations (MICs) of four antibiotics: eravacycline, tigecycline, polymyxin-B, and ceftazidime/avibactam. Synergistic interactions between these antibiotic combinations were examined using checkerboard and time-kill analyses. RESULTS: We noted significant differences in the MICs of ceftazidime/avibactam between KPC and MBL isolates. Checkerboard analysis revealed appreciable synergy between combinations of tigecycline (35%) or eravacycline (40%) with polymyxin-B. The synergy rates for the combination of tigecycline or eravacycline with polymyxin-B were similar among the KPC and MBL isolates. These combinations maintained a synergy rate of 70.6% even against polymyxin-B resistant isolates. In contrast, combinations of tigecycline (5%) or eravacycline (10%) with ceftazidime/avibactam showed significantly lower synergy than combinations with polymyxin-B (P < 0.001 and P = 0.002, respectively). Among the MBL CRKP isolates, only one exhibited synergy with eravacycline or tigecycline and ceftazidime/avibactam combinations, and no synergistic activity was identified in the time-kill analysis for these combinations. The combination of eravacycline and polymyxin-B demonstrated the most promising synergy in the time-kill analysis. CONCLUSION: This study provides substantial evidence of a significant synergy when combining tigecycline or eravacycline with polymyxin-B against CRKP strains, including those producing MBL. These results highlight potential therapeutic strategies against CRKP infections.

Pharmacological inhibition of the Src homology phosphatase 2 confers partial protection in a mouse model of alcohol-associated liver disease.

Alcohol-associated liver disease (ALD) is a leading factor of liver-related death worldwide. ALD has various manifestations that include steatosis, hepatitis, and cirrhosis and is currently without approved pharmacotherapies. The Src homology phosphatase 2 (Shp2) is a drug target in some cancers due to its positive regulation of Ras-mitogen-activated protein kinase signaling and cell proliferation. Shp2 pharmacological inhibition yields beneficial outcomes in animal disease models, but its impact on ALD remains unexplored. This study aims to investigate the effects of Shp2 inhibition and its validity using a preclinical mouse model of ALD. We report that the administration of SHP099, a potent and selective allosteric inhibitor of Shp2, partially ameliorated ethanol-induced hepatic injury, inflammation, and steatosis in mice. Additionally, Shp2 inhibition was associated with reduced ethanol-evoked activation of extracellular signal-regulated kinase (ERK), oxidative, and endoplasmic reticulum (ER) stress in the liver. Besides the liver, excessive alcohol consumption induces multi-organ injury and dysfunction, including the intestine. Notably, Shp2 inhibition diminished ethanol-induced intestinal inflammation and permeability, abrogated the reduction in tight junction protein expression, and the activation of ERK and stress signaling in the ileum. Collectively, Shp2 pharmacological inhibition mitigates the deleterious effects of ethanol in the liver and intestine in a mouse model of ALD. Given the multifactorial aspects underlying ALD pathogenesis, additional studies are needed to decipher the utility of Shp2 inhibition alone or as a component in a multitherapeutic regimen to combat this deadly malady.

Multimodal fused deep learning for drug property prediction: Integrating chemical language and molecular graph.

Accurately predicting molecular properties is a challenging but essential task in drug discovery. Recently, many mono-modal deep learning methods have been successfully applied to molecular property prediction. However, mono-modal learning is inherently limited as it relies solely on a single modality of molecular representation, which restricts a comprehensive understanding of drug molecules. To overcome the limitations, we propose a multimodal fused deep learning (MMFDL) model to leverage information from different molecular representations. Specifically, we construct a triple-modal learning model by employing Transformer-Encoder, Bidirectional Gated Recurrent Unit (BiGRU), and graph convolutional network (GCN) to process three modalities of information from chemical language and molecular graph: SMILES-encoded vectors, ECFP fingerprints, and molecular graphs, respectively. We evaluate the proposed triple-modal model using five fusion approaches on six molecule datasets, including Delaney, Llinas2020, Lipophilicity, SAMPL, BACE, and pKa from DataWarrior. The results show that the MMFDL model achieves the highest Pearson coefficients, and stable distribution of Pearson coefficients in the random splitting test, outperforming mono-modal models in accuracy and reliability. Furthermore, we validate the generalization ability of our model in the prediction of binding constants for protein-ligand complex molecules, and assess the resilience capability against noise. Through analysis of feature distributions in chemical space and the assigned contribution of each modal model, we demonstrate that the MMFDL model shows the ability to acquire complementary information by using proper models and suitable fusion approaches. By leveraging diverse sources of bioinformatics information, multimodal deep learning models hold the potential for successful drug discovery.

Clinical manifestations and risk factors for COVID-19 and its severity in patients with hematological malignancies.

BACKGROUND: Patients with hematological malignancies (HM) were at a high risk of developing severe disease from coronavirus disease 2019 (COVID-19). We aimed to assess the clinical outcome of COVID-19 in hospitalized patients with HM. METHODS: Adult patients with HM who were hospitalized with a laboratory-confirmed COVID-19 between May, 2021 and November, 2022 were retrospectively identified. Primary outcome was respiratory failure requiring mechanical ventilation or mortality within 60 days after hospitalization. We also analyzed associated factors for de-isolation (defined as defervescence with a consecutive serial cycle threshold value > 30) within 28 days. RESULTS: Of 152 eligible patients, 22 (14.5%) developed respiratory failure or mortality in 60 days. Factors associated with developing respiratory failure that required mechanical ventilation or mortality included receipt of allogeneic hematopoietic stem-cell transplantation (allo-HSCT) (adjusted hazards ratio [aHR], 5.10; 95% confidence interval [CI], 1.64-15.85), type 2 diabetes mellitus (aHR, 2.47; 95% CI, 1.04-5.90), lymphopenia at admission (aHR, 6.85; 95% CI, 2.45-19.15), and receiving <2 doses of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccines (aHR, 3.00; 95% CI, 1.19-7.60). Ninety-nine (65.1%) patients were de-isolated in 28 days, against which two hazardous factors were identified: receipt of B-cell depletion therapies within one year prior to COVID-19 (aHR, 0.55, 95% CI, 0.35-0.87) and lymphopenia upon admission (aHR, 0.65; 95% CI, 0.43-1.00). CONCLUSION: We found a high rate of respiratory failure and mortality among patients with HM who contracted the SARS-CoV-2. Factors associated with developing respiratory failure or mortality in 60 days included receipt of allo-HSCT, type 2 diabetes mellitus and lymphopenia upon admission. Having received ≥2 doses of vaccination conferred protection against clinical progression.

Status and Prospects of Research on Deep Learning-based De Novo Generation of Drug Molecules.

Traditional molecular de novo generation methods, such as evolutionary algorithms, generate new molecules mainly by linking existing atomic building blocks. The challenging issues in these methods include difficulty in synthesis, failure to achieve desired properties, and structural optimization requirements. Advances in deep learning offer new ideas for rational and robust de novo drug design. Deep learning, a branch of machine learning, is more efficient than traditional methods for processing problems, such as speech, image, and translation. This study provides a comprehensive overview of the current state of research in de novo drug design based on deep learning and identifies key areas for further development. Deep learning-based de novo drug design is pivotal in four key dimensions. Molecular databases form the basis for model training, while effective molecular representations impact model performance. Common DL models (GANs, RNNs, VAEs, CNNs, DMs) generate drug molecules with desired properties. The evaluation metrics guide research directions by determining the quality and applicability of generated molecules. This abstract highlights the foundational aspects of DL-based de novo drug design, offering a concise overview of its multifaceted contributions. Consequently, deep learning in de novo molecule generation has attracted more attention from academics and industry. As a result, many deep learning-based de novo molecule generation types have been actively proposed.

Discovery of potent small molecule inhibitors of histone lysine methyltransferase NSDs.

Nuclear receptor binding SET domain (NSD) proteins are a class of histone lysine methyltransferases and implicated in multiple cancer types with aberrant expression and involvement of cancer related signaling pathways. In this study, a series of small-molecule compounds including compound 2 and 3 are identified against the SET domain of NSDs through structure-based virtual screening. Our lead compound 3 exhibits potent inhibitory activities in vitro towards the NSD2-SET and NSD3-SET with an IC50 of 0.81 µM and 0.84 µM, respectively, and efficiently inhibits histone H3 lysine 36 dimethylation and decreases the expression of NSDs-targeted genes in non-small cell lung cancer cells at 100 nM. Compound 3 suppresses cell proliferation and reduces the clonogenicity in H460 and H1299 non-small cell lung cancer cells, and induces s-phase cell cycle arrest and apoptosis. These data establish our compounds as a valuable tool-kit for the study of the biological roles of NSDs in cancer.

Research Progress in Tumor Diagnosis Based on Raman Spectroscopy.

BACKGROUND: Cancer is a major disease that threatens human life and health. Raman spectroscopy can provide an effective detection method. OBJECTIVE: The study aimed to introduce the application of Raman spectroscopy to tumor detection. We have introduced the current mainstream Raman spectroscopy technology and related application research. METHODS: This article has first introduced the grim situation of malignant tumors in the world. The advantages of tumor diagnosis based on Raman spectroscopy have also been analyzed. Secondly, various Raman spectroscopy techniques applied in the medical field are introduced. Several studies on the application of Raman spectroscopy to tumors in different parts of the human body are discussed. Then the advantages of combining deep learning with Raman spectroscopy in the diagnosis of tumors are discussed. Finally, the related problems of tumor diagnosis methods based on Raman spectroscopy are pointed out. This may provide useful clues for future work. CONCLUSION: Raman spectroscopy can be an effective method for diagnosing tumors. Moreover, Raman spectroscopy diagnosis combined with deep learning can provide more convenient and accurate detection results.

Effect of early dexamethasone on outcomes of COVID-19: A quasi-experimental study using propensity score matching.

BACKGROUND: The RECOVERY trial demonstrated that the use of dexamethasone is associated with a 36% lower 28-day mortality in hospitalized patients with COVID-19 on invasive mechanical ventilation. Nevertheless, the optimal timing to start dexamethasone remains uncertain. METHODS: We conducted a quasi-experimental study at National Taiwan University Hospital (Taipei, Taiwan) using propensity score matching to simulate a randomized controlled trial to receive or not to receive early dexamethasone (6 mg/day) during the first 7 days following the onset of symptoms. Treatment was standard protocol-based, except for the timing to start dexamethasone, which was left to physicians' decision. The primary outcome is 28-day mortality. Secondary outcomes include secondary infection within 60 days and fulfilling the criteria of de-isolation within 20 days. RESULTS: A total of 377 patients with COVID-19 were enrolled. Early dexamethasone did not decrease 28-day mortality in all patients (adjusted odds ratio [aOR], 1.03; 95% confidence interval [CI], 0.97-1.10) or in patients who required O2 for severe/critical disease at admission (aOR, 1.05; 95%CI, 0.94-1.18); but is associated with a 24% increase in superinfection in all patients (aOR, 1.24; 95% CI, 1.12-1.37) and a 23% increase in superinfection in patients of O2 for several/critical disease at admission (aOR, 1.23; 95% CI, 1.02-1.47). Moreover, early dexamethasone is associated with a 42% increase in likelihood of delayed clearance of SARS-CoV-2 virus (adjusted hazard ratio, 1.42; 95% CI, 1.01-1.98). CONCLUSION: An early start of dexamethasone (within 7 days after the onset of symptoms) could be harmful to hospitalized patients with COVID-19.

Electrical tuning of branched flow of light.

Branched flows occur ubiquitously in various wave systems, when the propagating waves encounter weak correlated scattering potentials. Here we report the experimental realization of electrical tuning of the branched flow of light using a nematic liquid crystal (NLC) system. We create the physical realization of the weakly correlated disordered potentials of light via the inhomogeneous orientations of the NLC. We demonstrate that the branched flow of light can be switched on and off as well as tuned continuously through the electro-optical properties of NLC film. We further show that the branched flow can be manipulated by the polarization of the incident light due to the optical anisotropy of the NLC film. The nature of the branched flow of light is revealed via the unconventional intensity statistics and the rapid fidelity decay along the light propagation. Our study unveils an excellent platform for the tuning of the branched flow of light which creates a testbed for fundamental physics and offers a new way for steering light.

PRMT1 Integrates Immune Microenvironment and Fatty Acid Metabolism Response in Progression of Hepatocellular Carcinoma.

Background: Protein arginine methyltransferase (PRMT) family members have important roles in cancer processes. However, its functions in the regulation of cancer immunotherapy of hepatocellular carcinoma (HCC) are incompletely understood. This study aimed to investigate the roles of PRMT1 in HCC. Methods: Single-cell RNA sequencing (scRNA-seq) and clinicopathological data were obtained and used to explore the diagnostic and prognostic value, cellular functions and roles in immune microenvironment regulation of PRMT1 in HCC. The functions of PRMT1 were explored using Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO), as well as gene set enrichment analysis (GSEA). TIMER and CIBERSORT were used to analyze the relationships between PRMT1 expression and immune cell infiltration. The STRING database was used to construct a protein-protein interaction (PPI) network. Results: PRMT1 was aberrantly expressed in HCC, which high expression was associated with tumor progression, worse overall survival (OS) and disease-free survival (DFS) of patients with HCC. PRMT1 was also associated with immune cell infiltration. Moreover, it was specifically expressed in immune cells, including exhausted CD8 T cells, B cells, and mono/macro cells in patients with immunotherapy. The expression of immune checkpoints was significantly increased in the high-PRMT1 expression groups of HCC patients. Regarding biological mechanisms, cell viability, migration and invasion, and the expression of genes related to fatty acid metabolism were suppressed in PRMT1 knockdown HCC cells. Moreover, genes co-expressed with PRMT1 were involved in the fatty acid metabolic process and enriched in fatty and drug-induced liver disease. Conclusion: Taken together, these results indicate that PRMT1 might exert its oncogenic effects via immune microenvironment regulation and fatty acid metabolism in HCC. Our finding will provide a foundation for further studies and indicate a potential clinical therapeutic target for liver cancer.

Preventing and controlling intra-hospital spread of COVID-19 in Taiwan - Looking back and moving forward.

COVID-19 has exposed major weaknesses in the healthcare settings. The surge in COVID-19 cases increases the demands of health care, endangers vulnerable patients, and threats occupational safety. In contrast to a hospital outbreak of SARS leading to a whole hospital quarantined, at least 54 hospital outbreaks following a COVID-19 surge in the community were controlled by strengthened infection prevention and control measures for preventing transmission from community to hospitals as well as within hospitals. Access control measures include establishing triage, epidemic clinics, and outdoor quarantine stations. Visitor access restriction is applied to inpatients to limit the number of visitors. Health monitoring and surveillance is applied to healthcare personnel, including self-reporting travel declaration, temperature, predefined symptoms, and test results. Isolation of the confirmed cases during the contagious period and quarantine of the close contacts during the incubation period are critical for containment. The target populations and frequency of SARS-CoV-2 PCR and rapid antigen testing depend on the level of transmission. Case investigation and contact tracing should be comprehensive to identify the close contacts to prevent further transmission. These facility-based infection prevention and control strategies help reduce hospital transmission of SARS-CoV-2 to a minimum in Taiwan.

Achievements of COVID-19 vaccination programs: Taiwanese perspective.

The coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is a global health crisis. The specific characteristics of aerosol transmission in the latent period and the contagiousness of SARS-CoV-2 lead to rapid spread of infection in the community. Vaccination is the most effective method for preventing infection and severe outcomes. As of December 1, 2022, 88% of the Taiwanese population had received at least two doses of COVID-19 vaccines. Heterologous vaccination with ChAdOx1-mRNA-based or ChAdOx1-protein-based vaccines has been found to elicit higher immunogenicity than homologous vaccination with ChAdOx1-ChAdOx1 vaccines. A longitudinal cohort study revealed that 8-12-week intervals between the two heterologous vaccine doses of the primary series led to good immunogenicity and that the vaccines were safe. A third booster dose of mRNA vaccine is being encouraged to evoke effective immune responses against variants of concern. A novel domestic recombinant protein subunit vaccine (MVC-COV1901) was manufactured and authorized for emergency use in Taiwan. It has shown a good safety profile, with promising neutralizing antibody titers against SARS-CoV-2. Given the global pandemic due to emerging novel variants of SARS-CoV-2, booster COVID-19 vaccines and appropriate intervals between booster doses need to be investigated.

Antiviral therapy of coronavirus disease 2019 (COVID-19).

The coronavirus disease 2019 (COVID-19) pandemic has reached a turning point. The non-pharmaceutical interventions for preventing COVID-19 are lifting. Vaccination uptake is increasing in general, but this strategy is continuously challenged by the rapid evolution of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Of note, the Omicron subvariants spread globally for at least one year, and the most recently developed subvariants show strong immune evasion to preexisting immunity, either from previous infection, vaccination or both. Therefore, early and appropriate antiviral agents to treat patients at risk for severe COVID-19 or death is crucial to decrease morbidities and mortalities, to restore the healthcare capacities and to facilitate a return to the new normal. Current antiviral therapy for COVID-19 consist of neutralizing monoclonal antibodies (mAbs) and direct antiviral agents. Each agent has been proved for early ambulatory treatment of COIVD-19, but suffer from variable effectiveness and limitations due to patients' comorbidities, drug properties, or antiviral resistance. Besides, some specific mAbs are indicated for prophylaxis of COVID-19 before or after close contact with confirmed COVID-19 patients. This review article summarizes the evidence and unmet needs of the currently available antiviral agents for management of COVID-19 in the context of the Omicron subvariants.

Coronavirus disease 2019 (COVID-19) universal admission screening in patients and companions in Taiwan from May 2021 to June 2022: A nationwide multicenter study.

OBJECTIVE: Universal admission screening and follow-up symptom-based testing for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) may play critical roles in controlling nosocomial transmission. We describe the performance of test strategies for inpatients and their companions during various disease incidences in Taiwan. DESIGN: Retrospective population-based cohort study. SETTING: The study was conducted across 476 hospitals in Taiwan. METHODS: The data for both testing strategies by reverse transcription-polymerase chain reaction for SARS-CoV-2 in newly admitted patients and their companions during May 2021-June 2022 were extracted and analyzed. RESULTS: The positivity rate of universal admission screening was 0.76% (14,640 of 1,928,676) for patients and 0.37% (5,372 of 1,438,944) for companions. The weekly community incidences of period 1 (May 2021-June 2021), period 2 (July 2021-March 2022), and period 3 (April 2022-June 2022) were 6.57, 0.27, and 1,261, respectively, per 100,000 population. The positivity rates of universal admission screening for patients and companions (4.39% and 2.18%) in period 3 were higher than those in periods 1 (0.29% and 0.04%) and 2 (0.03% and 0.003%) (all P < .01). Among the 22,201 confirmed cases, 9.86% were identified by symptom-based testing. The costs and potential savings of universal admission screening for patients and companions achieved a breakeven point when the test strategy was implemented in a period with weekly community incidences of 27 and 358 per 100,000 population, respectively. CONCLUSIONS: Universal admission screening and follow-up symptom-based testing is important for reducing nosocomial transmission. Implementing universal admission screening at an appropriate time would balance the benefits with costs and potential unintended harms.

Immunogenicity and safety of heterologous booster with protein-based COVID-19 vaccine (NVX-CoV2373) in healthy adults: A comparative analysis with mRNA vaccines.

BACKGROUND: Information on the protein-based severe acute respiratory syndrome (SARS-CoV-2) vaccine-NVX-CoV2373 (Novavax), as a heterologous booster remains limited. We investigated the immunogenicity and adverse events of NVX-CoV2373 as a second booster and compared them with those of mRNA vaccines in healthy adults. METHODS: Healthcare workers who had received an mRNA vaccine (mRNA-1273 or BNT-162b2) as the first booster (third dose) 12 weeks prior were recruited. Participants voluntarily received either NVX-CoV2373 or an mRNA vaccine as a second booster. Participants with a history of SARS-CoV-2 infection were excluded. The primary outcomes included serum anti-SARS-CoV-2 spike protein (SP) and neutralizing antibody titers against B.1.1.7 (Alpha), B.1.1.529 (Omicron) BA2, and BA5 variants on the 28th day after the boost. Secondary outcomes included new SARS-CoV-2 infections and adverse events reported during the study period. RESULTS: A total of 160 participants were enrolled in this study. Compared with the mRNA vaccination group (n = 59), the NVX-CoV2373 vaccination group (n = 101) had significantly lower anti-SARS-CoV-2 SP antibody titers and neutralizing antibody titers against all variants tested after the boost. During the study period, higher rates of new SARS-CoV-2 infections and a lower incidence of adverse events were observed in the NVX-CoV2373 vaccination group. No significant differences in cellular immune responses were observed between the two groups. CONCLUSION: Compared to a homologous mRNA booster vaccination, heterologous boosters with NVX-CoV2373 showed lower antibody responses, a higher incidence of new SARS-CoV-2 infections, and fewer adverse events.

CXCL9 as a Reliable Biomarker for Discriminating Anti-IFN-γ-Autoantibody-Associated Lymphadenopathy that Mimics Lymphoma.

The diagnosis of adult-onset immunodeficiency syndrome associated with neutralizing anti-interferon γ autoantibodies (AIGA) presents substantial challenges to clinicians and pathologists due to its nonspecific clinical presentation, absence of routine laboratory tests, and resemblance to certain lymphoma types, notably nodal T follicular helper cell lymphoma, angioimmunoblastic type (nTFHL-AI). Some patients undergo lymphadenectomy for histopathological examination to rule out lymphoma, even in the absence of a preceding clinical suspicion of AIGA. This study aimed to identify reliable methods to prevent misdiagnosis of AIGA in this scenario through a retrospective case-control analysis of clinical and pathological data, along with immune gene transcriptomes using the NanoString nCounter platform, to compare AIGA and nTFHL-AI. The investigation revealed a downregulation of the C-X-C motif chemokine ligand 9 (CXCL9) gene in AIGA, prompting an exploration of its diagnostic utility. Immunohistochemistry (IHC) targeting CXCL9 was performed on lymph node specimens to assess its potential as a diagnostic biomarker. The findings exhibited a significantly lower density of CXCL9-positive cells in AIGA compared to nTFHL-AI, displaying a high diagnostic accuracy of 92.3% sensitivity and 100% specificity. Furthermore, CXCL9 IHC demonstrated its ability to differentiate AIGA from various lymphomas sharing similar characteristics. In conclusion, CXCL9 IHC emerges as a robust biomarker for differentiating AIGA from nTFHL-AI and other similar conditions. This reliable diagnostic approach holds the potential to avert misdiagnosis of AIGA as lymphoma, providing timely and accurate diagnosis.

Spine MRI image segmentation method based on ASPP and U-Net network.

The spine is one of the most important structures in the human body, serving to support the body, organs, protect nerves, etc. Medical image segmentation for the spine can help doctors in their clinical practice for rapid decision making, surgery planning, skeletal health diagnosis, etc. The current difficulty is mainly the poor segmentation accuracy of skeletal Magnetic Resonance Imaging (MRI) images. To address the problem, we propose a spine MRI image segmentation method, Atrous Spatial Pyramid Pooling (ASPP)-U-shaped network (UNet), which combines an ASPP structure with a U-Net network. This approach improved the network feature extraction by introducing an ASPP structure into the U-Net network down-sampling structure. The medical image segmentation models are trained and tested on publicly available datasets and obtained the Dice coefficient and Mean Intersection over Union coefficients with 0.866 and 0.755, respectively. The experimental results show that ASPP-UNet has higher accuracy for spine MRI image segmentation compared with other mainstream networks.