SEARCH · Search bioRxiv

Results for “Pathology”

Search indexed bioRxiv preprints in genomics, neuroscience, cell biology and bioinformatics. Read source abstracts and check manuscript versions; preprints are not peer reviewed.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 37 records · Page 2Linked to original sources

bioRxiv2018

Green autofluorescence intensity of skin and fingernails: A novel biomarker for non-invasive evaluation of pathological state of blood vessels

Stroke and myocardial infarction (MI) are two leading causes of death around the world. It is of great significance to establish novel and non-invasive approaches for evaluating pathological state of blood vessels, so that early interventions may be carried out to prevent incidence of stroke or MI. Our recent studies have suggested that altered Pattern of Autofluorescence (AF) of skin and fingernails are novel biomarkers of acute ischemic stroke (AIS) and MI. In particular, our studies have shown characteristic increases in the green AF intensity of the fingernails and certain regions of the skin of AIS patients and MI patients. By determining the skins green AF of the Healthy Group, the Low-Risk Group for Developing AIS, and the High-Risk Group for Developing AIS, our current study has indicated that the green AF intensity in the fingernails and certain regions of the skin, including the right and left Dorsal Index Fingers, Ventroforefingers, Dorsal Antebrachium and Index Fingernails as well as right Dorsal Centremetacarpus, is highly correlated with the risk to develop AIS. There is also evidence suggesting that increased oxidative stress may account for the increased AF intensity in the Low-Risk Group for Developing AIS and the High-Risk Group for Developing AIS. These findings have suggested that the green AF intensity of the fingernails and certain regions of the skin is a novel biomarker for non-invasive evaluation of the pathological state of blood vessels and the risk for developing AIS or MI.

pathology↗

bioRxiv2016

Modelling avoidance in pathologically anxious humans using reinforcement-learning

Serious and debilitating symptoms of anxiety are the most common mental health problem worldwide, accounting for around 5% of all adult years lived with disability in the developed world. Avoidance behaviour -avoiding social situations for fear of embarrassment, for instance-is a core feature of such anxiety. However, as for many other psychiatric symptoms, the biological mechanisms underlying avoidance remain unclear. Reinforcement-learning models provide formal and testable characterizations of the mechanisms of decision-making; here, we examine avoidance in these terms. One hundred and one healthy and pathologically anxious individuals completed an approach-avoidance go/no-go task under stress induced by threat of unpredictable shock. We show an increased reliance in the anxious group on a parameter of our reinforcement-learning model that characterizes a prepotent (Pavlovian) bias to withhold responding in the face of negative outcomes. This was particularly the case when the anxious individuals were under stress. This formal description of avoidance within the reinforcement-learning framework provides a new means of linking clinical symptoms with biophysically plausible models of neural circuitry and, as such, takes us closer to a mechanistic understanding of pathological anxiety.

animal behavior and cognition↗

bioRxiv2017

Mice lacking NF-κB1 exhibit marked DNA damage responses and more severe gastric pathology in response to parenteral tamoxifen administration.

BackgroundTamoxifen (TAM) has recently been shown to cause acute gastric atrophy and metaplasia in mice. We have previously demonstrated that the outcome of Helicobacter felis infection, which induces similar gastric lesions in mice, is altered by deletion of specific NF-{kappa}B subunits. Nfkb1-/- mice developed more severe gastric atrophy than wild-type (WT) mice 6 weeks after H. felis infection. In contrast, Nfkb2-/- mice were protected from this pathology. We therefore hypothesized that gastric lesions induced by TAM may be similarly regulated by signaling via NF-{kappa}B subunits.\n\nMethodsGroups of 5 female C57BL/6 (WT), Nfkb1-/-, Nfkb2-/- and c-Rel-/- mice were administered 150mg/kg TAM by IP injection. 72 hours later, gastric corpus tissues were taken for quantitative histological assessment. In addition, groups of 6 female WT and Nfkb1-/- mice were exposed to 12Gy {gamma}-irradiation. Gastric epithelial apoptosis was quantified 6 and 48 hours after irradiation.\n\nResultsTAM induced gastric epithelial lesions in all strains of mice, but this was more severe in Nfkb1-/- mice than WT mice. Nfkb1-/- mice exhibited more severe parietal cell loss than WT mice, had increased gastric epithelial expression of Ki67 and had an exaggerated gastric epithelial DNA damage response as quantified by {gamma}H2AX. To determine investigate whether the difference in gastric epithelial DNA damage response of Nfkb1-/- mice was unique to TAM induced DNA damage, or a generic consequence of DNA damage, we also assessed gastric epithelial apoptosis following {gamma}-irradiation. 6 hours after {gamma}-irradiation, gastric epithelial apoptosis was increased in the gastric corpus and antrum of Nfkb1-/- mice.\n\nConclusionsNF-{kappa}B1 mediated signaling regulates the development of gastric mucosal pathology following TAM administration. This is associated with an exaggerated gastric epithelial DNA damage response. This aberrant response appears to reflect a more generic sensitization of the gastric mucosa of Nfkb1-/- mice to DNA damage.

immunology↗

bioRxiv2017

P53 is not necessary for DUX4 pathology

AbstractsFSHD is a genetically dominant myopathy caused by mutations that cause expression of the normally silent DUX4 gene. This transcription factor has been shown to interfere with myogenesis when misexpressed at very low levels in myoblasts, and to cause cell death when overexpressed at high levels. A previous report using adeno-associated virus to deliver high levels of DUX4 to mouse skeletal muscle demonstrated severe pathology that was suppressed on a p53 knockout background, implying that DUX4 acted through the p53 pathway. Here, we investigate the p53-dependence of DUX4 using both in vitro cellular and the transgenic iDUX4[2.7] mouse models. We find that inhibiting p53 has no effect on the cytoxicity of DUX4 in vitro. When crossed onto the p53 null background, we find no suppression of the male-specific lethality or skin phenotypes of the DUX4 transgene, and find that primary myoblasts from this mouse are still killed by DUX4 expression. These data challenge the notion that the p53 pathway is central to the pathogenicity of DUX4.\n\nSummary StatementDUX4 is thought to mediate cytopathology through p53. Here, DUX4 is shown to kill primary myoblasts and promote pathological phenotypes in the iDUX4[2.7] mouse model on the p53-null background, calling into question this notion.

genetics↗

bioRxiv2017

A common neural network among state, trait, and pathological anxiety from whole-brain functional connectivity

Anxiety is one of the most common mental states of humans. Although it drives us to avoid frightening situations and to achieve our goals, it may also impose significant suffering and burden if it becomes extreme. Because we experience anxiety in a variety of forms, previous studies investigated neural substrates of anxiety in a variety of ways. These studies revealed that individuals with high state, trait, or pathological anxiety showed altered neural substrates. However, no studies have directly investigated whether the different dimensions of anxiety share a common neural substrate, despite its theoretical and practical importance. Here, we investigated a neural network of anxiety shared by different dimensions of anxiety in a unified analytical framework using functional magnetic resonance imaging (fMRI). We analyzed different datasets in a single scale, which was defined by an anxiety-related neural network derived from whole brain. Through the fMRI task for provoking anxiety, we found a common neural network of state anxiety across participants (1,638 trials obtained from 10 participants). Then, using the resting-state fMRI in combination with the participants trait anxiety scale (879 participants from the Human Connectome Project), we demonstrated that trait anxiety also shared the same neural network as state anxiety. Furthermore, the common neural network between state and trait anxiety could detect patients with obsessive-compulsive disorder, which is characterized by pathological anxiety-driven behaviors (174 participants from multi-site datasets). Our findings provide direct evidence that different dimensions of anxiety are not completely independent but have a substantial biological inter-relationship. Our results also provide a biologically defined dimension of anxiety, which may promote further investigation of various human characteristics, including psychiatric disorders, from the perspective of anxiety.

neuroscience↗

bioRxiv2017

Soluble amyloid-beta buffering by plaques in Alzheimer disease dementia versus high-pathology controls

An unanswered question regarding Alzheimer disease dementia (ADD) is whether amyloid-beta (A{beta}) plaques sequester toxic soluble A{beta} species early in the pathological progression. We previously reported that the concentration of soluble A{beta} aggregates from patients with mild dementia was higher than soluble A{beta} aggregates from patients with modest A{beta} plaque burden but no dementia. The ratio of soluble A{beta} aggregate concentration to A{beta} plaque area fully distinguished these groups of patients. We hypothesized that initially plaques may serve as a reservoir or sink for toxic soluble A{beta} aggregates, sequestering them from other targets in the extracellular space and thereby preventing their toxicity. To initially test a generalized version of this hypothesis, we have performed binding assessments using biotinylated synthetic A{beta}1-42 peptide. A{beta}1-42-biotin peptide was incubated on unfixed frozen sections from non-demented high plaque pathology controls and patients with dementia of the Alzheimer type. The bound peptide was measured using ELISA and confocal microscopy. We observed no quantitative difference in A{beta} binding between the groups using either method. Further testing of the buffering hypothesis using various forms of synthetic and human derived soluble A{beta} aggregates will be required to definitively address the role of plaque buffering as it relates to ADD.

neuroscience↗

bioRxiv2018

Convolving Pre-Trained Convolutional Neural Networks at Various Magnifications to Extract Diagnostic Features for Digital Pathology

Deep learning is an area of artificial intelligence that has received much attention in the past few years due to both an increase in computational power with the increased use of graphics processing units (GPUs) for computational analyses and the performance of these class of algorithms on visual recognition tasks. They have found utility in applications ranging from image search to facial recognition for security and social media purposes. Their continued success has propelled their use across many new domains including the medical field, in areas of radiology and pathology in particular, as these fields are thought to be driven by visual recognition tasks. In this paper, we present an application of deep learning, termed transfer learning, using ResNet50, a pre-trained convolutional neural network (CNN) to act as a feature-detector at various magnifications to identify low and high level features in digital pathology images of various breast lesions for the purpose of classifying them correctly into the labels of normal, benign, in-situ, or invasive carcinoma as provided in the ICIAR 2018 Breast Cancer Histology Challenge (BACH).

bioinformatics↗

bioRxiv2018

OptoGranules reveal the evolution of stress granules to ALS-FTD pathology

Stress granules are non-membranous assemblies of mRNA and protein that form in response to a variety of stressors. Genetic, pathologic, biophysical and cell biological studies have implicated disturbances in the dynamics of membrane-less organelles, such as stress granules, as a pathobiological component of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD)1-12. This confluence of evidence has inspired the hypothesis that these diseases reflect an underlying disturbance in the dynamics and material properties of stress granules; however, this concept has remained largely untestable in available models of stress granule assembly, which require the confounding variable of exogenous stressors. Here we demonstrate the development and use of a light-inducible stress granule system, termed OptoGranules, which permits discrete, experimental control of the dynamics and material properties of stress granules in living cells in the absence of exogenous stressors. The nucleator in this system is Opto-G3BP1, a light-sensitive chimeric protein assembled from the intrinsically disordered region (IDR) and RNA-binding domain of G3BP1 combined with the light-sensitive oligomerization domain of Arabidopsis thaliana cryptochrome 2 (CRY2) photolyase homology region (PHR). Upon stimulation with blue light, Opto-G3BP1 initiates the rapid assembly of dynamic, cytoplasmic, liquid granules that are composed of canonical stress granule components, including G3BP1, PABP, TIA1, TIAR, eIF4G, eIF3{eta}, ataxin 2, GLE1, TDP-43 and polyadenylated RNA. With this system, we demonstrate that persistent or repetitive assembly of stress granules is cytotoxic and is accompanied by the evolution of stress granules to neuronal cytoplasmic inclusions that recapitulate the pathology of ALS-FTD.

neuroscience↗

bioRxiv2018

Evolutionary history of Alzheimer Disease causing protein family Presenilins with pathological implications.

Presenilin proteins are type II transmembrane proteins. They make the catalytic component of Gamma secretase, a multiportion transmembrane protease. Amyloid protein, Notch and beta catenin are among more than 90 substrates of Presenilins. Mutations in Presenilins lead to defects in proteolytic cleavage of its substrate resulting in some of the most devastating pathological conditions including Alzheimer disease (AD), developmental disorders and cancer. In addition to catalytic roles, Presenilin protein is also shown to be involved in many non-catalytic roles i.e. calcium homeostasis, regulation of autophagy and protein trafficking etc. These proteolytic proteins are highly conserved, present in almost all the major eukaryotic groups. Studies on wide variety of organisms ranging from human to unicellular dictyostelium have shown the important catalytic and non-catalytic roles of Presenilins. In the current research project, we aimed to elucidate the phylogenetic history of Presenilins. We showed that Presenilins are the most ancient of the Gamma secretase proteins and might have their origin in last common eukaryotic ancestor (LCEA). We also demonstrated that these proteins have been evolving under strong purifying selection. Through evolutionary trace analysis, we showed that Presenilin protein sites which undergoes mutations in Familial Alzheimer Disease are highly conserved in metazoans. Finally, we discussed the evolutionary, physiological and pathological implication of our findings and proposed that evolutionary profile of Presenilins supports the loss of function hypothesis of AD pathogenesis.

evolutionary biology↗

bioRxiv2018

The pathological hallmarks of Alzheimer’s disease derive from compensatory responses to NMDA receptor insufficiency

Alzheimers disease is characterized by intracellular aggregates of hyperphosphorylated tau protein and extracellular plaques of amyloid {beta} peptide, a product of APP processing. The origin of these pathological hallmarks has remained elusive. Here, we have tested the idea that both alterations, at the onset of the disease, may constitute compensatory responses to the same causative and initial trigger, namely NMDA receptor insufficiency. Treatment of rat cortical neurons with the specific NMDA receptor antagonist AP5 within 4 h caused a significant increase in tau phosphorylation at the AT8 and S404 epitopes as well as an increase in APP expression and A{beta} 40 secretion. Single intraperitoneal injections of the NMDA receptor open channel blocker MK-801 into wild-type mice reproduced all of these changes in a brain region-specific fashion either at latency 4 h or 24 h. Subchronic treatment with MK-801 for 6 weeks induced AT8, S404 and S396 immunoreactivity selectively in female mice. We conclude that the pivotal pathological alterations in Alzheimers disease represent runaway physiological responses to persistently insufficient excitatory neurotransmission. In view of the evidence for excitatory insufficiency in trisomy 21 patients, PS1 mutation carriers and ApoE4 carriers, our data suggest a common pathomechanism behind familial, sporadic, and risk allele-triggered Alzheimers disease. The potential of this mechanism to reconcile previous conflicting observations is discussed.

neuroscience↗

bioRxiv2026

Schizophrenia-like neurodevelopmental pathology reshapes experience-dependent brain network remodeling following adolescent alcohol exposure

Background Alcohol use disorder (AUD) is highly prevalent in schizophrenia, yet the neurobiological basis of this vulnerability remains poorly understood. Neurodevelopmental models suggest that pre-existing brain dysconnectivity may increase vulnerability to AUD. We therefore tested whether schizophrenia-like neurodevelopmental pathology alters how alcohol-related experience is incorporated into large-scale brain networks. Methods Resting-state functional connectivity was assessed in male Sprague-Dawley rats (n = 18-21/group) with neonatal ventral hippocampal lesions (NVHL), a neurodevelopmental model of schizophrenia, and sham-operated controls, with or without voluntary adolescent alcohol exposure. Functional connectivity was assessed using seed-to-voxel and seed-to-seed analyses within a cortico-striato-limbic network. We additionally examined whether individual alcohol intake during adolescence predicted adult functional connectivity according to neurodevelopmental status. Results NVHL and adolescent alcohol exposure independently produced predominantly hypoconnected cortico-striato-limbic networks. However, alcohol exposure did not exacerbate NVHL-associated dysconnectivity but instead induced a distinct network reorganization characterized by functional hyperconnectivity. Although alcohol intake was comparable between groups, dose-dependent relationships between adolescent alcohol consumption and adult functional connectivity were observed in sham animals but were absent or markedly attenuated in NVHL rats. These effects were primarily centered on prelimbic cortex connectivity with the amygdala, hippocampus, and dorsal striatum, highlighting this circuitry as a major locus of altered experience-dependent remodeling. Conclusions These findings suggest that vulnerability to AUD associated with schizophrenia-like neurodevelopment may arise less from additive network dysfunction than from an altered capacity of large-scale brain networks for experience-dependent functional remodeling. Schizophrenia-like neurodevelopmental pathology may therefore change how alcohol-related experience is translated into persistent brain network organization.

neuroscience↗

bioRxiv2017

QuPath: Open source software for digital pathology image analysis

QuPath is new bioimage analysis software designed to meet the growing need for a user-friendly, extensible, open-source solution for digital pathology and whole slide image analysis. In addition to offering a comprehensive panel of tumor identification and high-throughput biomarker evaluation tools, QuPath provides researchers with powerful batch-processing and scripting functionality, and an extensible platform with which to develop and share new algorithms to analyze complex tissue images. Furthermore, QuPaths flexible design make it suitable for a wide range of additional image analysis applications across biomedical research.

pathology↗

bioRxiv2016

Nonsinusoidal oscillations underlie pathological phase-amplitude coupling in the motor cortex in Parkinson’s disease

Parkinsons disease (PD) is associated with abnormal beta oscillations (13-30 Hz) in the basal ganglia and motor cortex (M1). Recent reports show that M1 beta-high gamma (50-200 Hz) phase-amplitude coupling (PAC) is exaggerated in PD and is reduced following acute deep brain stimulation (DBS). Here we analyze invasive M1 electrocorticography recordings in PD patients on and off DBS, and in isolated cervical dystonia patients, and show that M1 beta oscillations are nonsinusoidal, having sharp and asymmetric features. These sharp oscillatory beta features underlie the previously reported PAC, providing an alternative to the standard interpretation of PAC as an interaction between two distinct frequency components. Specifically, the ratio between peak and trough sharpness is nearly perfectly correlated with beta-high gamma PAC (r = 0.96) and predicts PD-related motor deficit. Using a simulation of the local field potential, we demonstrate that sharp oscillatory waves can arise from synchronous synaptic activity. We propose that exaggerated beta-high gamma PAC may actually reflect such synchronous synaptic activity, manifesting as sharp beta oscillations that are \"smoothed out\" with DBS. These results support the \"desynchronization\" hypothesis of DBS wherein DBS counteracts pathological synchronization throughout the basal ganglia-thalamocortical loop. We argue that PAC can be influenced by more than one mechanism. In this case synaptic synchrony, rather than the often assumed spike-field coherence, may underlie exaggerated PAC. These often overlooked temporal features of the oscillatory waveform carry critical physiological information about neural processes and dynamics that may lead to better understanding of underlying neuropathology.

Neuroscience↗

bioRxiv2017

Cellular Interplay and Cytokine Hierarchy Cause Pathological Cardiac Hypertrophy in RAF1-Mutant Noonan Syndrome

Noonan syndrome (NS) is caused by mutations in RAS/ERK pathway genes, and is characterized by craniofacial, growth, cognitive and cardiac defects. NS patients with kinase-activating RAF1 alleles typically develop pathological left ventricular hypertrophy (LVH), which is reproduced in Raf1L613V/+ knock-in mice. Here, using inducible Raf1L613V expression, we show that LVH results from the interplay of cardiac cell types. Cardiomyocyte Raf1L613V enhances Ca2+ sensitivity and cardiac contractility without causing hypertrophy. Raf1L613V expression in cardiomyocytes or activated fibroblasts exacerbates pressure overload-evoked fibrosis. Endothelial/endocardial (EC) Raf1L613V causes cardiac hypertrophy without affecting contractility. Co-culture and neutralizing antibody experiments reveal a cytokine (TNF/IL6) hierarchy in Raf1L613V-expressing ECs that drives cardiomyocyte hypertrophy in vitro. Furthermore, post-natal TNF inhibition normalizes the increased wall thickness and cardiomyocyte hypertrophy in vivo. We conclude that NS cardiomyopathy involves cardiomyocytes, ECs, and fibroblasts, TNF/IL6 signaling components represent potential therapeutic targets, and abnormal EC signaling might contribute to other forms of LVH.

cell biology↗

bioRxiv2017

Tacrolimus Rescues Endothelial ALK1 Loss-Of-Function Signaling And Improves HHT Vascular Pathology

Hereditary hemorrhagic telangiectasia (HHT) is a genetic vascular disorder arising from endothelial cell (EC) proliferation and hypervascularization, for which no cure exists. Because HHT is caused by loss-of-function mutations in BMP9-ALK1-Smad1/5/8 signaling, interventions aimed at activating this pathway are of therapeutic value. By screening FDA-approved drug libraries, we identified tacrolimus (FK-506) as a potent activator of Smad1/5/8 in BMP9-challenged reporter cells. In primary ECs, tacrolimus activated Smad1/5/8 to oppose the pro-angiogenic gene expression signature associated with ALK1 loss-of-function, by notably reducing Dll4 expression. In these cells, tacrolimus also inhibited Akt and p38 stimulation by VEGF. In the BMP9/10-immunodepleted postnatal retina--a mouse model of HHT vascular pathology--tacrolimus activated endothelial Smad1/5/8 and prevented the Dll4 overexpression and hypervascularization associated with this model. Finally, tacrolimus stimulated Smad1/5/8 in cells transfected with BMP9-unresponsive ALK1 HHT mutants and in HHT patient blood outgrowth ECs. We propose that tacrolimus repurposing has therapeutic potential in HHT.

cell biology↗

bioRxiv2017

Reversal of molecular pathology by RNA-targeting Cas9 in a myotonic dystrophy mouse model

The dominantly inherited, multi-systemic disease myotonic dystrophy type I (DM1) is caused by triplet repeat CTG expansions in the DMPK gene and is the most common form of adult-onset muscular dystrophy. Elimination of the toxic, repetitive CUG RNA constitutes a therapeutic for this disease. We report an RNA-targeting Cas9 (RCas9) system that supports efficient reversal of DM1 phenotypes via delivery to adult poly(CUG) DM1 mouse muscle using adeno-associated virus (AAV). We observe elimination of CUG RNA, restoration of CUG foci-associated Mbnl1 protein to wild-type subcellular localization, correction of DM1-type alternative splicing patterns in candidate genes including the voltage-gated chloride channel 1 (Clcn1) responsible for characteristic myotonia, recovery of Clcn1 staining, and reduction in centralized myonuclei. Our results establish RCas9 as a potential long-term in vivo therapeutic for DM1.\n\nOne Sentence SummaryA repurposed CRISPR system termed RNA-targeting Cas9 reverses the molecular pathology associated with the most common type of adult onset muscular dystrophy in adult mouse muscle.

molecular biology↗

bioRxiv2017

Methylation-To-Expression Feature Models of Breast Cancer Accurately Predict Overall Survival, Distant-Recurrence Free Survival, And Pathologic Complete Response in Multiple Cohorts

BackgroundApproaches that capitalize on the benefits of multi-omic data integration in invasive breast carcinoma to define prognostic biomarkers for precision medicine have been slow to emerge. In this work, we examined the efficacy of our methylation-to-expression feature model (M2EFM) approach to combining molecular and clinical predictors as part of a single analysis to create prognostic risk scores for overall survival, distant metastasis, and chemosensitivity.\n\nMethodsGene expression and DNA methylation values as well as clinical variables were integrated via M2EFM to build prognostic models of overall survival using 1028 breast tumor samples and further applied to external validation cohorts of 61 and 327 samples. Data-integrated prognostic models of distant recurrence-free survival and pathologic complete response were built using 306 samples and validated on 182 samples of external validation data. Additionally, we compared the discrimination and calibration of M2EFM models to other approaches.\n\nResultsDespite different populations and assays, M2EFM models validated with good accuracy (C-index or AUC [≥] .7) for all outcomes in all validation data. M2EFM models had the most consistent performance overall and superior calibration, suggesting a greater likelihood of clinical utility. Finally, we demonstrated that M2EFM identifies functionally relevant genes, which could be useful in translating an M2EFM biomarker to the clinic.\n\nConclusionM2EFM uses multiple levels of genomic data to infer disrupted regulatory patterns, thus providing a gene signature that connects loss of regulatory control with cancer prognosis.\n\nFundingThe analyses described in this report were supported by NIH grants R01ES022222, P30CA138292, P30ES019776, and R01DE022772.\n\nConflicts of InterestThe authors declare no potential conflicts of interest.

bioinformatics↗

bioRxiv2017

Deep Convolutional Neural Networks Enable Discrimination of Heterogeneous Digital Pathology Images

Pathological evaluation of tumor tissue is pivotal for diagnosis in cancer patients and automated image analysis approaches have great potential to increase precision of diagnosis and help reduce human error.\n\nIn this study, we utilize various computational methods based on convolutional neural networks (CNN) and build a stand-alone pipeline to effectively classify different histopathology images across different types of cancer. In particular, we demonstrate the utility of our pipeline to discriminate between two subtypes of lung cancer, four biomarkers of bladder cancer, and five biomarkers of breast cancer. In addition, we apply our pipeline to discriminate among four immunohistochemistry (IHC) staining scores of bladder and breast cancers.\n\nOur classification pipeline utilizes a basic architecture of CNN, Googles Inceptions within three training strategies, and an ensemble of two state-of-the-art algorithms, Inception and ResNet. These strategies include training the last layer of Googles Inceptions, training the network from scratch, and fine-tunning the parameters for our data using two pre-trained version of Googles Inception architectures, Inception-V1 and Inception-V3.\n\nWe demonstrate the power of deep learning approaches for identifying cancer subtypes, and the robustness of Googles Inceptions even in presence of extensive tumor heterogeneity. Our pipeline on average achieved accuracies of 100%, 92%, 95%, and 69% for discrimination of various cancer types, subtypes, biomarkers, and scores, respectively. Our pipeline and related documentation is freely available at https://github.com/ih-lab/CNN_Smoothie.

bioinformatics↗