Biology subjects

Liu, R.

Publications and source records attributed to Liu, R..

At least 271 records · Page 15Linked to original sources

bioRxiv2020

Gluconeogenic Enzyme PCK1 Deficiency Is Critical for CHK2 O-GlcNAcylation and Hepatocellular Carcinoma Growth upon Glucose Deprivation

Elevated hexosamine-biosynthesis pathway (HBP) activity and O-GlcNAcylation are emerging hallmarks of hepatocellular carcinoma (HCC). Inhibiting O-GlcNAcylation could be a promising anti-cancer strategy. Here, we investigate this possibility and demonstrate that deficiency of phosphoenolpyruvate carboxykinase 1 (PCK1), a rate-limiting enzyme in gluconeogenesis, promotes O-GlcNAcylation and hepatoma cell proliferation under low-glucose conditions. PCK1 loss results in oxaloacetate accumulation and AMPK inactivation, promoting uridine diphosphate-N-acetylglucosamine (UDP-GlcNAc) synthesis and CHK2 threonine 378 O-GlcNAcylation and counteracting its ubiquitination and degradation. O-GlcNAcylation also promotes CHK2-dependent Rb phosphorylation and HCC cell proliferation. Therefore, blocking HBP-mediated O-GlcNAcylation suppresses tumor progression in liver-specific Pck1-knockout mice. We reveal a link between PCK1 depletion and hyper-O-GlcNAcylation that underlies HCC oncogenesis and suggest therapeutic targets for HCC that act by inhibiting O-GlcNAcylation.

cancer biology↗

bioRxiv2020

PecanPy: a fast, efficient, and parallelized Python implementation of node2vec

Learning low-dimensional representations (embeddings) of nodes in large graphs is key to applying machine learning on massive biological networks. Node2vec is the most widely used method for node embedding. However, its original Python and C++ implementations scale poorly with network density, failing for dense biological networks with hundreds of millions of edges. We have developed PecanPy, a new Python implementation of node2vec that uses cache-optimized compact graph data structures and precomputing/parallelization to result in fast, high-quality node embeddings for biological networks of all sizes and densities. PecanPy software and documentation are available at https://github.com/krishnanlab/pecanpy.

bioinformatics↗

bioRxiv2020

Restoration of β-globin expression with optimally designed lentiviral vector for β-thalassemia treatment in Chinese patients

{beta}-thalassemia is one of the most prevalent genetic diseases worldwide. The current treatment for {beta}-thalassemia is allogeneic hematopoietic stem cell transplantation (HSCT), which is limited due to lack of matched donors. Gene therapy has been developed as an alternative therapeutic option for transfusion-ependent {beta}-thalassemia (TDT). However, successful gene therapy for {beta}-thalassemia patients in China has not been reported. Here we present the results of preclinical studies of an optimally designed LV named LentiHBBT87Q in hematopoietic stem cells (HSCs) derived from Chinese TDT patients. LentiHBBT87Q was selected from a series of LVs with optimized backbone and de novo cloning strategy. It contains an exogenous T87Q {beta}-globin gene (HBBT87Q) driven by a specific reconstituted locus control region (rLCR) and efficiently express HBB mRNA and HBB protein in erythroblasts derived from cord blood (CB) HSCs. To facilitate clinical transformation, we manufactured clinical grade LentiHBBT87Q (cLentiHBBT87Q) and optimized its transduction procedure. Importantly, transduction of cLentiHBBT87Qrestored expression of HBB monomer and adult hemoglobin (HbA) tetramer to relatively normal level in erythroblasts from bone marrow (BM) HSCs of Chinese TDT patients, that carry the most common mutation types and cover various genotypes, including {beta}0/{beta}0. Furthermore, viral integration sites (VIS) of cLentiHBBT87Q were similar to other LVs safely used in previous clinical trials and the associated risk of tumorigenesis was not observed in cLentiHBBT87Q transduced HSCs through comprehensive analysis. Taken together, we have engineered the cLentiHBBT87Q that can restore {beta}-globin expression in the HSCs-derived erythroblasts of Chinese TDT patients with minimal risk on tumorigenesis, providing a favorable starting point for future clinical application.

cell biology↗

bioRxiv2020

Trantrace: a reliable, traceable, and collaborative solution to streamline the translation of emerging biomedical resources

BackgroundPrecision medicine is gaining popularity in routine health care, which heavily relies on the interpretation of emerging biomedical databases and professional guidelines. Translation of those biomedical materials, that is crucial to deliver research discovery to health interventions in non-native English-speaking countries, remains a large amount of work for biomedical practitioners. ResultsWe presented an application called Trantrace to facilitate the routine of large-scale long-running translation. Especially for the case that many people joined multiple translation projects, it has a rigorous task division and cooperation process: assignment, translation, review, release, and iteration. Trantrace empowers users to build own projects with different permissions, version control all the translation operation without specific skills, and further improve the translation efficiency through real-time task management tools. ConclusionsBy providing a working platform for collaborative translation and enabling long-time version correction, the Trantrace contributes towards the creation of a multilingual biomedical knowledge commons, which would be a valuable aid for personalized therapies. The source code is freely accessible at https://github.com/sgi-drylab/trantrace.

bioinformatics↗

bioRxiv2020

Chloroquine and hydroxychloroquine as ACE2 blockers to inhibit viropexis of COVID-19 Spike psedotype virus

BackgroundThe novel coronavirus disease (2019-nCoV) has been affecting global health since the end of 2019 and there is no sign that the epidemic is abating. The major issue for controlling the infectious is lacking efficient prevention and therapeutic approaches. Chloroquine (CQ) and Hydroxychloroquine (HCQ) have been reported to treat the disease, but the underlying mechanism remains controversial. PurposeThe objective of this study is to investigate whether CQ and HCQ could be ACE2 blockers and used to inhibit 2019-nCoV virus infection. MethodsIn our study, we used CCK-8 staining, flow cytometry and immunofluorescent staining to evaluate the toxicity and autophagy of CQ and HCQ, respectively, on ACE2 high-expressing HEK293T cells (ACE2h cells). We further analyzed the binding character of CQ and HCQ to ACE2 by molecular docking and surface plasmon resonance (SPR) assays, 2019-nCoV spike pseudotyped virus was also used to observe the viropexis effect of CQ and HCQ in ACE2h cells. ResultsResults showed that HCQ is slightly more toxic to ACE2h cells than CQ. Both CQ and HCQ could bind to ACE2 with KD =(7.31{+/-}0.62)e-7 M and (4.82{+/-}0.87)e-7 M, respectively. They exhibit equivalent suppression effect for the entrance of 2019-nCoV spike pseudotyped virus into ACE2h cells. ConclusionsCQ and HCQ both inhibit the entrance 2019-nCoV into cells by blocking the binding of the virus with ACE2. Our findings provide novel insights into the molecular mechanism of CQ and HCQ treatment effect on virus infection.

biochemistry↗

bioRxiv2020

A generative modeling approach for interpreting population-level variability in brain structure

Understanding how neural structure varies across individuals is critical for characterizing the effects of disease, learning, and aging on the brain. However, disentangling the different factors that give rise to individual variability is still an outstanding challenge. In this paper, we introduce a deep generative modeling approach to find different modes of variation across many individuals. To do this, we start by training a variational autoencoder on a collection of auto-fluorescence images from a little over 1,700 mouse brains at 25 micron resolution. To then tap into the learned factors and validate the models expressiveness, we developed a novel bi-directional technique to interpret the latent space-by making structured perturbations to both, the high-dimensional inputs of the network, as well as the low-dimensional latent variables in its bottleneck. Our results demonstrate that through coupling generative modeling frameworks with structured perturbations, it is possible to probe the latent space to provide insights into the representations of brain structure formed in deep neural networks.

neuroscience↗

bioRxiv2020

An enormous Paris polyphylla genome sheds light on genome size evolution and polyphyllin biogenesis

The monocot family Melanthiaceae with varying genome sizes in a range of 230-fold is an ideal model to study the genome size fluctuation in plants. Its family member Paris genus demonstrates an evolutionary trend of bearing huge genomes characterized by an average c-value of 49.22 pg. Here, we report a 70.18 Gb genome assembly out of the 82.55 Gb genome of Paris polyphylla var. yunnanensis (PPY), which represents the biggest sequenced genome to date. We annotate 69.53% repetitive sequences in this genome and 62.50% of which are long-terminal repeat (LTR) transposable elements. Further evolution analysis indicates that the giant genome likely results from the joint effect of common and species-specific expansion of different LTR superfamilies, which might contribute to the environment adaptation after speciation. Moreover, we identify the candidate pathway genes for the biogenesis of polyphyllins, the PPY-specific medicinal saponins, by complementary approaches including genome mining, comprehensive analysis of 31 next-generation RNA-seq data and 55.23 Gb single-molecule circular consensus sequencing (CCS) RNA-seq reads, and correlation of the transcriptome and phytochemical data of five different tissues at four growth stages. This study not only provides significant insights into plant genome size evolution, but also paves the way for the following polyphyllin synthetic biology.

plant biology↗

bioRxiv2020

Tumor-on-a-chip platform to interrogate the role of macrophages in tumor progression

Tumor-infiltrating leukocytes, in particular macrophages, play an important role in tumor behavior and clinical outcome. The spectrum of macrophage subtypes ranges from antitumor "M1"-type to protumor "M2"-type macrophages. Tumor-associated macrophages (TAMs) typically display phenotypic features of both M1 and M2, and the population distribution is thought to be dynamic and evolve as the tumor progresses. However, our understanding of how TAMs impact the tumor microenvironment remains limited by the lack of appropriate 3D in vitro models that can capture cell to cell dynamics at high spatial and temporal resolution. Using our recently developed micro-physiological "tumor-on-a-chip" (TOC) device, we present here our findings on the impact of defined macrophage subsets on tumor behavior. The TOC device design contains three adjacent and connected chambers in which both the upper and lower chambers are loaded with tumor cells while the central chamber contains a dynamic, perfused, living microvascular network. Introduction of human pancreatic or colorectal cancer cells together with M1-polorized macrophages significantly inhibited tumor growth and tumor-induced angiogenesis. Protein analysis and antibody-based neutralization studies confirmed that these effects were mediated through production of chemokines CXCL9, CXCL10, and CXCL11. By contrast, M2-macrophages mediated increased tumor cell migration into the vascularized chamber and did not inhibit tumor growth or angiogenesis. In fact, single-cell RNA-sequencing showed that M2 macrophages further segregated endothelial cells into two distinct subsets, corresponding to static cells in vessels versus active cells involved in angiogenesis. The impact of M2 macrophages was mediated mostly by production of MMP7 and ANGPT2. In summary, our data demonstrate the utility of the TOC device to mechanistically probe biological questions in a 3D in vitro microenvironment. Insight BoxMacrophages in the tumor microenvironment are key determinants of tumor behavior and clinical outcome. The macrophage subset composition and its functional impact change as tumors progress or during treatment, but adequate models to study this are lacking. We developed a tumor-on-a-chip model of perfused 3D tumor growth to probe the impact of defined macrophage subsets. Our data is consistent with previously described macrophage activity in the tumor microenvironment, and provides potential new molecular targets. Herein, we demonstrate feasibility of probing immuno-oncology questions in a 3D in vitro microenvironment and at a spatiotemporal resolution.

cancer biology↗

bioRxiv2020

mbkmeans: fast clustering for single cell data using mini-batch k-means

Single-cell RNA-Sequencing (scRNA-seq) is the most widely used high-throughput technology to measure genome-wide gene expression at the single-cell level. One of the most common analyses of scRNA-seq data detects distinct subpopulations of cells through the use of unsupervised clustering algorithms. However, recent advances in scRNA-seq technologies result in current datasets ranging from thousands to millions of cells. Popular clustering algorithms, such as k-means, typically require the data to be loaded entirely into memory and therefore can be slow or impossible to run with large datasets. To address this problem, we developed the mbkmeans R/Bioconductor package, an open-source implementation of the mini-batch k-means algorithm. Our package allows for on-disk data representations, such as the common HDF5 file format widely used for single-cell data, that do not require all the data to be loaded into memory at one time. We demonstrate the performance of the mbkmeans package using large datasets, including one with 1.3 million cells. We also highlight and compare the computing performance of mbkmeans against the standard implementation of k - means. Our software package is available in Bioconductor at https://bioconductor.org/packages/mbkmeans.

bioinformatics↗

bioRxiv2020

Mapping fetal brain development based on automated segmentation and 4D brain atlasing

Fetal brain MRI has become an important tool for in utero assessment of brain development and disorders. However, quantitative analysis of fetal brain MRI remains difficult, partially due to the limited tools for automated preprocessing and the lack of normative brain templates. In this paper, we proposed an automated pipeline for fetal brain extraction, super-resolution reconstruction, and fetal brain atlasing to quantitatively map in utero fetal brain development during mid-to-late gestation in a Chinese population. First, we designed a U-net convolutional neural network for automated fetal brain extraction, which achieved an average accuracy of 97%. We then generated a developing fetal brain atlas, using an iterative linear and nonlinear registration approach. Based on the 4D spatiotemporal atlas, we quantified the morphological development of the fetal brain between 23-36 weeks of gestation. The proposed pipeline enabled the fully-automated volumetric reconstruction for clinically available fetal brain MRI data, and the 4D fetal brain atlas provided normative templates for quantitative analysis of potential fetal brain abnormalities, especially in the Chinese population.

neuroscience↗

bioRxiv2020

Tumor neoantigen heterogeneity impacts bystander immune inhibition of pancreatic cancer growth

The threshold for immunogenic clonal fraction in a heterogeneous solid tumor required to induce effective bystander killing of non-immunogenic subclones is unknown. Pancreatic cancer poses crucial challenges for immune therapeutic interventions due to low mutational burden and consequent lack of neoantigens. Here, we designed a model to incorporate artificial neoantigens into genes of interest in cancer cells and to test the potential of said antigens to actuate bystander killing. By precisely controlling the abundance of a neoantigen in the tumor, we studied the impact of neoantigen frequency on immune response and immune escape. Our results showed that a single, strong, widely expressed neoantigen could lead to a robust antitumor response when at least 80% of cancer cells express the neoantigen. Further, immunological assays revealed induction of T-cell responses against a non-target self-antigen on KRAS oncoprotein, when we inoculated animals with a high frequency of tumor cells expressing a test neoantigen. Using nanoparticle-based gene therapy, we successfully altered the tumor microenvironment by perturbing interleukin-12 and interleukin-10 gene expression. The subsequent remodeling of the microenvironment reduced the threshold of neoantigen frequency at which bioluminescent signal intensity for tumor burden decreased 1.5-logfold, marking a robust tumor growth inhibition, from 83% to as low as 29%. Our results thus suggest that bystander killing is rather inefficient in immunologically cold tumors like pancreatic cancer and requires an extremely high abundance of neoantigens. However, the bystander killing mediated antitumor response can be rescued, when supported by adjuvant immune therapy.

cancer biology↗

bioRxiv2020

Predicting cell fate commitment of embryonic differentiation by single-cell graph entropy

Cell fate commitment occurs during early embryonic development, that is, the embryonic differentiation sometimes undergoes a critical phase transition or "tipping point" of cell fate commitment, at which there is a drastic or qualitative shift of the cell populations. In this study, we presented a novel computational approach, the single-cell graph entropy (SGE), to explore the gene-gene associations among cell populations based on single-cell RNA sequencing (scRNA-seq) data. Specifically, by transforming the sparse and fluctuating gene expression data to the stable local network entropy, the SGE score quantitatively characterizes the criticality of gene regulatory networks among cell populations, and thus can be employed to predict the tipping point of cell fate or lineage commitment at the single cell level. The proposed SGE method was applied to five scRNA-seq datasets. For all these datasets of embryonic differentiation, SGE effectively captures the signal of the impending cell fate transitions, which cannot be detected by gene expressions. Some "dark" genes that are non-differential but sensitive to SGE values were revealed. The successful identification of critical transition for all five datasets demonstrates the effectiveness of our method in analyzing scRNA-seq data from a network perspective, and the potential of SGE to track the dynamics of cell differentiation.

bioinformatics↗

bioRxiv2020

Visible light optical coherence tomography angiography (vis-OCTA) and local microvascular retinal oximetry in human retina

We report herein the first visible light optical coherence tomography angiography (vis-OCTA) for human retinal imaging. Compared to the existing vis-OCT systems, we devised a spectrometer with a narrower bandwidth to increase the spectral power density for OCTA imaging, while retaining the major spectral contrast in the blood. We achieved a 100 kHz A-line rate, the fastest acquisition speed reported so far for human retinal vis-OCT. We rigorously optimized the imaging protocol such that a single acquisition takes <6 seconds with a field of view (FOV) of 3x7.8 mm2. The angiography enables accurate localization of microvasculature down to the capillary level and thus enables oximetry at vessels < 100 m in diameter. We demonstrated microvascular hemoglobin oxygen saturation (sO2) at the feeding and draining vessels at the perifoveal region. The longitudinal repeatability was assessed by <5% coefficient of variation (CV). The unique capabilities of our vis-OCTA system may allow studies on the role of microvascular oxygen in various retinal pathologies.

bioengineering↗

bioRxiv2020

Development of a Coupled Simulation Toolkit for Computational Radiation Biology Based on Geant4 and CompuCell3D

Understanding and designing clinical radiation therapy is one of the most important areas of state-of-the-art oncological treatment regimens. Decades of research have gone into developing sophisticated treatment devices and optimization protocols for schedules and dosages. In this paper, we presented a comprehensive computational platform that facilitates building of the sophisticated multi-cell-based model of how radiation affects the biology of living tissue. We designed and implemented a coupled simulation method, including a radiation transport model, and a cell biology model, to simulate the tumor response after irradiation. The radiation transport simulation was implemented through Geant4 which is an open-source Monte Carlo simulation platform that provides many flexibilities for users, as well as low energy DNA damage simulation physics, Geant4-DNA. The cell biology simulation was implemented using CompuCell3D (CC3D) which is a cell biology simulation platform. In order to couple Geant4 solver with CC3D, we developed a "bridging" module that extracts tumor cellular geometry of the CC3D simulation (including specification of the individual cells) and ported it to the Geant4 for radiation transport simulation. The cell dose and cell DNA damage distribution in multicellular system were obtained using Geant4. The tumor response was simulated using cell-based tissue models based on CC3D. By merging two powerful and widely used modeling platforms, CC3D and Geant4, we delivered a novel tool that can give us the ability to simulate the dynamics of biological tissue in the presence of ionizing radiation, which provides a powerful framework for quantifying the biological consequences of radiation therapy. The developed tool has an advantage on that it has strong extensibility due to the exploitability of two modeling platforms. In this introductory methods paper, we described our modeling platform in detail and showed how it can be applied to study the application of radiotherapy to a vascularized tumor.

biophysics↗

bioRxiv2020

Autocrine insulin pathway signaling regulates actin dynamics in cell wound repair

Cells are exposed to frequent mechanical and/or chemical stressors that can compromise the integrity of the plasma membrane and underlying cortical cytoskeleton. The molecular mechanisms driving the immediate repair response launched to restore the cell cortex and circumvent cell death are largely unknown. Using microarrays and drug-inhibition studies to assess gene expression, we find that initiation of cell wound repair in the Drosophila model is dependent on translation, whereas transcription is required for subsequent steps. We identified 253 genes whose expression is up-regulated (80) or down-regulated (173) in response to laser wounding. A subset of these genes were validated using RNAi knockdowns and exhibit aberrant actomyosin ring assembly and/or actin remodeling defects. Strikingly, we find that the canonical insulin signaling pathway controls actin dynamics through the actin regulators Girdin and Chickadee (profilin), and its disruption leads to abnormal wound repair. Our results provide new insight for understanding how cell wound repair proceeds in healthy individuals and those with diseases involving wound healing deficiencies.

developmental biology↗

bioRxiv2020

FEP-based screening prompts drug repositioning against COVID-19

Coronavirus disease 2019 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has become a global crisis. There is no therapeutic treatment specific for COVID-19. It is highly desirable to identify potential antiviral agents against SARS-CoV-2 from existing drugs available for other diseases and, thus, repurpose them for treatment of COVID-19. In general, a drug repurposing effort for treatment of a new disease, such as COVID-19, usually starts from a virtual screening of existing drugs, followed by experimental validation, but the actual hit rate is generally rather low with traditional computational methods. Here we report a new virtual screening approach with accelerated free energy perturbation-based absolute binding free energy (FEP-ABFE) predictions and its use in identifying drugs targeting SARS-CoV-2 main protease (Mpro). The accurate FEP-ABFE predictions were based on the use of a new restraint energy distribution (RED) function designed to accelerate the FEP-ABFE calculations and make the practical FEP-ABFE-based virtual screening of the existing drug library possible for the first time. As a result, out of twenty-five drugs predicted, fifteen were confirmed as potent inhibitors of SARS-CoV-2 Mpro. The most potent one is dipyridamole (Ki=0.04 M) which has showed promising therapeutic effects in subsequently conducted clinical studies for treatment of patients with COVID-19. Additionally, hydroxychloroquine (Ki=0.36 M) and chloroquine (Ki=0.56 M) were also found to potently inhibit SARS-CoV-2 Mpro for the first time. We anticipate that the FEP-ABFE prediction-based virtual screening approach will be useful in many other drug repurposing or discovery efforts. Significance StatementDrug repurposing effort for treatment of a new disease, such as COVID-19, usually starts from a virtual screening of existing drugs, followed by experimental validation, but the actual hit rate is generally rather low with traditional computational methods. It has been demonstrated that a new virtual screening approach with accelerated free energy perturbation-based absolute binding free energy (FEP-ABFE) predictions can reach an unprecedently high hit rate, leading to successful identification of 16 potent inhibitors of SARS-CoV-2 main protease (Mpro) from computationally selected 25 drugs under a threshold of Ki = 4 M. The outcomes of this study are valuable for not only drug repurposing to treat COVID-19, but also demonstrating the promising potential of the FEP-ABFE prediction-based virtual screening approach.

biophysics↗

bioRxiv2020

EASE: EM-Assisted Source Extraction from calcium imaging data

Combining two-photon calcium imaging (2PCI) and electron microscopy (EM) provides arguably the most powerful current approach for connecting function to structure in neural circuits. Recent years have seen dramatic advances in obtaining and processing CI and EM data separately. In addition, several joint CI-EM datasets (with CI performed in vivo, followed by EM reconstruction of the same volume) have been collected. However, no automated analysis tools yet exist that can match each signal extracted from the CI data to a cell segment extracted from EM; previous efforts have been largely manual and focused on analyzing calcium activity in cell bodies, neglecting potentially rich functional information from axons and dendrites. There are two major roadblocks to solving this matching problem: first, dense EM reconstruction extracts orders of magnitude more segments than are visible in the corresponding CI field of view, and second, due to optical constraints and non-uniform brightness of the calcium indicator in each cell, direct matching of EM and CI spatial components is nontrivial. In this work we develop a pipeline for fusing CI and densely-reconstructed EM data. We model the observed CI data using a constrained nonnegative matrix factorization (CNMF) framework, in which segments extracted from the EM reconstruction serve to initialize and constrain the spatial components of the matrix factorization. We develop an efficient iterative procedure for solving the resulting combined matching and matrix factorization problem and apply this procedure to joint CI-EM data from mouse visual cortex. The method recovers hundreds of dendritic components from the CI data, visible across multiple functional scans at different depths, matched with densely-reconstructed three-dimensional neural segments recovered from the EM volume. We publicly release the output of this analysis as a new gold standard dataset that can be used to score algorithms for demixing signals from 2PCI data. Finally, we show that this database can be exploited to (1) learn a mapping from 3d EM segmentations to predict the corresponding 2d spatial components estimated from CI data, and (2) train a neural network to denoise these estimated spatial components. This neural network denoiser is a stand-alone module that can be dropped in to enhance any existing 2PCI analysis pipeline.

neuroscience↗

bioRxiv2020

Accurate Diagnosis of Colorectal Cancer Based on Histopathology Images Using Artificial Intelligence

BackgroundAccurate and robust pathological image analysis for colorectal cancer (CRC) diagnosis is time-consuming and knowledge-intensive, but is essential for CRC patients treatment. The current heavy workload of pathologists in clinics/hospitals may easily lead to unconscious misdiagnosis of CRC based on their daily image analyses. MethodsBased on a state-of-the-art transfer-learned deep convolutional neural network in artificial intelligence (AI), we proposed a novel patch aggregation strategy for clinic CRC prediction/diagnosis using weakly labeled pathological whole slide image (WSI) patches. This approach was trained and validated using an unprecedented and enormously large number of 170,099 patches, >14,680 WSIs, from >9,631 subjects that covered diverse and representative clinical cases from multi-independent-sources across China, U.S., and Germany. ResultsOur innovative AI tool was consistently nearly perfectly agreed with (average Kappa-statistic 0.896) and even often better than most of the experienced expert pathologists when tested in diagnosing CRC WSIs from multi-centers. The average area under the receiver operating characteristics curve (AUC) of AI was greater than that of the pathologists (0.981 vs 0.970) and achieved the best performance among the application of other AI methods to CRC diagnosis. Our AI-generated heatmap highlights the image regions of cancer tissue/cells. ConclusionsThis first-ever generalizable AI system can handle large amounts of WSIs consistently and robustly without potential bias due to fatigue commonly experienced by clinical pathologists. Hence, it will drastically alleviate the heavy clinical burden of daily pathology diagnosis, and improve the treatment for CRC patients. This tool is generalizable to other cancer diagnosis based on image recognition.

pathology↗