Exploring nonlinear feature space dimension reduction and data representation in breast Cadx with Laplacian eigenmaps and t-SNE

CT Angiography Radiomics Combining Traditional Risk Factors to Predict Brain Arteriovenous Malformation Rupture: a Machine Learning, Multicenter Study

This study aimed to develop a machine learning model for predicting brain arteriovenous malformation (bAVM) rupture using a combination of traditional risk factors and radiomics features. This multicenter retrospective study enrolled 586 patients with unruptured bAVMs from 2010 to 2020. All patients were grouped into the hemorrhage (n = 368) and non-hemorrhage (n = 218) groups. The bAVM nidus were segmented on CT angiography images using Slicer software, and radiomic features were extracted using Pyradiomics. The dataset included a training set and an independent testing set. The machine learning model was developed on the training set and validated on the testing set by merging numerous base estimators and a final estimator based on the stacking method. The area under the receiver operating characteristic (ROC) curve, precision, and the f1 score were evaluated to determine the performance of the model. A total of 1790 radiomics features and 8 traditional risk factors were contained in the original dataset, and 241 features remained for model training after L1 regularization filtering. The base estimator of the ensemble model was Logistic Regression, whereas the final estimator was Random Forest. In the training set, the area under the ROC curve of the model was 0.982 (0.967-0.996) and 0.893 (0.826-0.960) in the testing set. This study indicated that radiomics features are a valuable addition to traditional risk factors for predicting bAVM rupture. In the meantime, ensemble learning can effectively improve the performance of a prediction model.

A generalizable and easy-to-use COVID-19 stratification model for the next pandemic via immune-phenotyping and machine learning

Introduction

The coronavirus disease 2019 (COVID-19) pandemic has affected billions of people worldwide, and the lessons learned need to be concluded to get better prepared for the next pandemic. Early identification of high-risk patients is important for appropriate treatment and distribution of medical resources. A generalizable and easy-to-use COVID-19 severity stratification model is vital and may provide references for clinicians.

Methods

Three COVID-19 cohorts (one discovery cohort and two validation cohorts) were included. Longitudinal peripheral blood mononuclear cells were collected from the discovery cohort (n = 39, mild = 15, critical = 24). The immune characteristics of COVID-19 and critical COVID-19 were analyzed by comparison with those of healthy volunteers (n = 16) and patients with mild COVID-19 using mass cytometry by time of flight (CyTOF). Subsequently, machine learning models were developed based on immune signatures and the most valuable laboratory parameters that performed well in distinguishing mild from critical cases. Finally, single-cell RNA sequencing data from a published study (n = 43) and electronic health records from a prospective cohort study (n = 840) were used to verify the role of crucial clinical laboratory and immune signature parameters in the stratification of COVID-19 severity.

Results

Patients with COVID-19 were determined with disturbed glucose and tryptophan metabolism in two major innate immune clusters. Critical patients were further characterized by significant depletion of classical dendritic cells (cDCs), regulatory T cells (Tregs), and CD4+ central memory T cells (Tcm), along with increased systemic interleukin-6 (IL-6), interleukin-12 (IL-12), and lactate dehydrogenase (LDH). The machine learning models based on the level of cDCs and LDH showed great potential for predicting critical cases. The model performances in severity stratification were validated in two cohorts (AUC = 0.77 and 0.88, respectively) infected with different strains in different periods. The reference limits of cDCs and LDH as biomarkers for predicting critical COVID-19 were 1.2% and 270.5 U/L, respectively.

Conclusion

Overall, we developed and validated a generalizable and easy-to-use COVID-19 severity stratification model using machine learning algorithms. The level of cDCs and LDH will assist clinicians in making quick decisions during future pandemics.

CD57-positive CD8 + T cells define the response to anti-programmed cell death protein-1 immunotherapy in patients with advanced non-small cell lung cancer

Immune checkpoint inhibitors have transformed the treatment landscape of non-small cell lung cancer (NSCLC). However, accurately identifying patients who will benefit from immunotherapy remains a challenge. This study aimed to discover potential biomarkers for predicting immunotherapy response in NSCLC patients. Single-cell mass cytometry (CyTOF) was utilized to analyze immune cell subsets in peripheral blood mononuclear cells (PBMCs) obtained from NSCLC patients before and 12 weeks after single-agent immunotherapy. The CyTOF findings were subsequently validated using flow cytometry and multiplex immunohistochemistry/immunofluorescence in PBMCs and tumor tissues, respectively. RNA sequencing (RNA-seq) was performed to elucidate the underlying mechanisms. In the CyTOF cohort (n = 20), a high frequency of CD57+CD8+ T cells in PBMCs was associated with durable clinical benefit from immunotherapy in NSCLC patients (p = 0.034). This association was further confirmed in an independent cohort using flow cytometry (n = 27; p < 0.001), with a determined cutoff value of 12.85%. The cutoff value was subsequently validated in another independent cohort (AUC = 0.733). We also confirmed the CyTOF findings in pre-treatment formalin-fixed and paraffin-embedded tissues (n = 90; p < 0.001). RNA-seq analysis revealed 475 differentially expressed genes (DEGs) between CD57+CD8+ T cells and CD57-CD8+ T cells, with functional analysis identifying DEGs significantly enriched in immune-related signaling pathways. This study highlights CD57+CD8+ T cells as a promising biomarker for predicting immunotherapy success in NSCLC patients.

The Hybrid Deep Learning Model for Identification of Attention-Deficit/Hyperactivity Disorder Using EEG

Common misbehavior among children that prevents them from paying attention to tasks and interacting with their surroundings appropriately is attention-deficit/hyperactivity disorder (ADHD). Studies of children's behavior presently face a significant problem in the early and timely diagnosis of this disease. To diagnose this disease, doctors often use the patient's description and questionnaires, psychological tests, and the patient's behavior in which reliability is questionable. Convolutional neural network (CNN) is one deep learning technique that has been used for the diagnosis of ADHD. CNN, however, does not account for how signals change over time, which leads to low classification performances and ambiguous findings. In this study, the authors designed a hybrid deep learning model that combines long-short-term memory (LSTM) and CNN to simultaneously extract and learn the spatial features and long-term dependencies of the electroencephalography (EEG) data. The effectiveness of the proposed hybrid deep learning model was assessed using 2 publicly available EEG datasets. The suggested model achieves a classification accuracy of 98.86% on the ADHD dataset and 98.28% on the FOCUS dataset, respectively. The experimental findings show that the proposed hybrid CNN-LSTM model outperforms the state-of-the-art methods to diagnose ADHD using EEG. Hence, the proposed hybrid CNN-LSTM model could therefore be utilized to help with the clinical diagnosis of ADHD patients.

Impaired Efferocytosis Enables Apoptotic Osteoblasts to Escape Osteoimmune Surveillance During Aging

Macrophage efferocytosis of apoptotic osteoblasts (apoOBs) is a key osteoimmune process for bone homeostasis. However, apoOBs frequently accumulate in aged bone marrow, where they may mount proinflammatory responses and progressive bone loss. The reason why apoOBs are not cleared during aging remains unclear. In this study, it is demonstrated that aged apoOBs upregulate the immune checkpoint molecule CD47, which is controlled by SIRT6-regulated transcriptional pausing, to evade clearance by macrophages. Using osteoblast- and myeloid-specific gene knockout mice, SIRT6 is further revealed to be a critical modulator for apoOBs clearance via targeting CD47-SIRPα checkpoint. Moreover, apoOBs activate SIRT6-mediated chemotaxis to recruit macrophages by releasing apoptotic vesicles. Two targeting delivery strategies are developed to enhance SIRT6 activity, resulting in rejuvenated apoOBs clearance and delayed age-related bone loss. Collectively, the findings reveal a previously unknown linkage between immune surveillance and bone homeostasis and targeting the SIRT6-regulated mechanism can be a promising therapeutic strategy for age-related bone diseases.

Artificial intelligence in cancer diagnosis and therapy: Current status and future perspective

Artificial intelligence (AI) in healthcare plays a pivotal role in combating many fatal diseases, such as skin, breast, and lung cancer. AI is an advanced form of technology that uses mathematical-based algorithmic principles similar to those of the human mind for cognizing complex challenges of the healthcare unit. Cancer is a lethal disease with many etiologies, including numerous genetic and epigenetic mutations. Cancer being a multifactorial disease is difficult to be diagnosed at an early stage. Therefore, genetic variations and other leading factors could be identified in due time through AI and machine learning (ML). AI is the synergetic approach for mining the drug targets, their mechanism of action, and drug-organism interaction from massive raw data. This synergetic approach is also facing several challenges in data mining but computational algorithms from different scientific communities for multi-target drug discovery are highly helpful to overcome the bottlenecks in AI for drug-target discovery. AI and ML could be the epicenter in the medical world for the diagnosis, treatment, and evaluation of almost any disease in the near future. In this comprehensive review, we explore the immense potential of AI and ML when integrated with the biological sciences, specifically in the context of cancer research. Our goal is to illuminate the many ways in which AI and ML are being applied to the study of cancer, from diagnosis to individualized treatment. We highlight the prospective role of AI in supporting oncologists and other medical professionals in making informed decisions and improving patient outcomes by examining the intersection of AI and cancer control. Although AI-based medical therapies show great potential, many challenges must be overcome before they can be implemented in clinical practice. We critically assess the current hurdles and provide insights into the future directions of AI-driven approaches, aiming to pave the way for enhanced cancer interventions and improved patient care.

First-line durvalumab and tremelimumab with chemotherapy in RAS-mutated metastatic colorectal cancer: a phase 1b/2 trial

Although patients with microsatellite instable metastatic colorectal cancer (CRC) benefit from immune checkpoint blockade, chemotherapy with targeted therapies remains the only therapeutic option for microsatellite stable (MSS) tumors. The single-arm, phase 1b/2 MEDITREME trial evaluated the safety and efficacy of durvalumab plus tremelimumab combined with mFOLFOX6 chemotherapy in first line, in 57 patients with RAS-mutant unresectable metastatic CRC. Safety was the primary objective of phase Ib; no safety issue was observed. The phase 2 primary objective of efficacy in terms of 3-month progression-free survival (PFS) in patients with MSS tumors was met, with 3-month PFS of 90.7% (95% confidence interval (CI): 79.2-96%). For secondary objectives, response rate was 64.5%; median PFS was 8.2 months (95% CI: 5.9-8.6); and overall survival was not reached in patients with MSS tumors. We observed higher tumor mutational burden and lower genomic instability in responders. Integrated transcriptomic analysis underlined that high immune signature and low epithelial-mesenchymal transition were associated with better outcome. Immunomonitoring showed induction of neoantigen and NY-ESO1 and TERT blood tumor-specific T cell response associated with better PFS. The combination of durvalumab-tremelimumab with mFOLFOX6 was tolerable with promising clinical activity in MSS mCRC. Clinicaltrials.gov identifier: NCT03202758 .

Statistical shape model-based tibiofibular assessment of syndesmotic ankle lesions using weight-bearing CT

Forced external rotation is hypothesized as the key mechanism of syndesmotic ankle injuries, inducing a three-dimensional deviation from the normal distal tibiofibular joint (DTFJ) alignment. However, current diagnostic imaging modalities are impeded by a two-dimensional assessment, without considering ligamentous stabilizers. Therefore, our aim is threefold: (1) to construct an articulated statistical shape model of the normal DTFJ with the inclusion of ligamentous morphometry, (2) to investigate the effect of weight-bearing on the DTFJ alignment, and (3) to detect differences in predicted syndesmotic ligament length of patients with syndesmotic lesions with respect to normative data. Training data comprised non-weight-bearing CT scans from asymptomatic controls (N = 76), weight-bearing CT scans from patients with syndesmotic ankle injury (N = 13), and their weight-bearing healthy contralateral side (N = 13). Path and length of the syndesmotic ligaments were predicted using a discrete element model, wrapped around bony contours. Statistical shape model evaluation was based on accuracy, generalization, and compactness. The predicted ligament length in patients with syndesmotic lesions was compared with healthy controls. With respect to the first aim, our presented skeletal shape model described the training data with an accuracy of 0.23 ± 0.028 mm. Mean prediction accuracy of ligament insertions was 0.53 ± 0.12 mm. In accordance with the second aim, our results showed an increased tibiofibular diastasis in healthy ankles after weight-bearing. Concerning our third aim, a statistically significant difference in anterior syndesmotic ligament length was found between ankles with syndesmotic lesions and healthy controls (p = 0.017). There was a significant correlation between the presence of syndesmotic injury and the positional alignment between the distal tibia and fibula (r = 0.873, p < 0,001). Clinical Significance: Statistical shape modeling combined with patient-specific ligament wrapping techniques can facilitate the diagnostic workup of syndesmosic ankle lesions under weight-bearing conditions. In doing so, an increased anterior tibiofibular distance was detected, corresponding to an "anterior open-book injury" of the ankle syndesmosis as a result of anterior inferior tibiofibular ligament elongation/rupture.

Single-cell RNA sequencing reveals different signatures of mesenchymal stromal cell pluripotent-like and multipotent populations

Somatic stem cells are advantageous research targets for understanding the properties required to maintain stemness. Human bone marrow-mesenchymal stromal cells (BM-MSCs) were separated into pluripotent-like SSEA-3(+) Muse cells (Muse-MSCs) and multipotent SSEA-3(−) MSCs (MSCs) and were subjected to single-cell RNA sequencing analysis. Compared with MSCs, Muse-MSCs exhibited higher expression levels of the p53 repressor MDM2; signal acceptance-related genes EGF, VEGF, PDGF, WNT, TGFB, INHB, and CSF; ribosomal protein; and glycolysis and oxidative phosphorylation. Conversely, MSCs had higher expression levels of FGF and ANGPT; Rho family and caveola-related genes; amino acid and cofactor metabolism; MHC class I/II, and lysosomal enzyme genes than Muse-MSCs. Unsupervised clustering further divided Muse-MSCs into two clusters stratified by the expression of cell cycle-related genes, and MSCs into three clusters stratified by the expression of cell cycle-, cytoskeleton-, and extracellular matrix-related genes. This study evaluating the differentiation ability of BM-MSC subpopulations provides intriguing insights for understanding stemness.

•

MSCs were separated into pluripotent-like Muse-MSCs and multipotent MSCs

•

Gene expressions of Muse-MSCs and MSCs were analyzed by single-cell RNA sequencing

•

p53 suppressor, ribosomal protein, and energy metabolism were higher in Muse-MSCs

•

Genes related to the cytoskeleton, amino acid metabolism, and MHC were higher in MSCs

Naïve pluripotent-like characteristics of non-tumorigenic Muse cells isolated from human amniotic membrane

Multilineage-differentiating stress-enduring (Muse) cells are non-tumorigenic pluripotent-like stem cells that exhibit triploblastic differentiation and self-renewability at the single-cell level, and are collectable as pluripotent surface marker SSEA-3(+) from the bone marrow (BM), peripheral blood, and organ connective tissues. SSEA-3(+) cells from human amniotic membrane mesenchymal stem cells (hAMSCs) were compared with hBM-Muse cells. Similar to hBM-Muse cells, hAMSC-SSEA-3(+) cells expressed pluripotency genes (OCT3/4, NANOG, and SOX2), differentiated into triploblastic cells from a single cell, self-renewed, and exhibited non-tumorigenicity. Notably, however, they exhibited unique characteristics not seen in hBM-Muse cells, including higher expression of genes related to germline- and extraembryonic cell-lineages compared with those in hBM-Muse cells in single-cell RNA-sequencing; and enhanced expression of markers relevant to germline- (PRDM14, TFAP2C, and NANOS3) and extraembryonic cell- (CDX2, GCM1, and ID2) lineages when induced by cytokine subsets, suggesting a broader differentiation potential similar to naïve pluripotent stem cells. t-SNE dimensionality reduction and Gene ontology analysis visualized hAMSC-SSEA-3(+) cells comprised a large undifferentiated subpopulation between epithelial- and mesenchymal-cell states and a small mesenchymal subpopulation expressing genes relevant to the placental formation. The AM is easily accessible by noninvasive approaches. These unique cells are a potentially interesting target naïve pluripotent stem cell-like resource without tumorigenicity.

Phagocytosing differentiated cell-fragments is a novel mechanism for controlling somatic stem cell differentiation within a short time frame

Stem cells undergo cytokine-driven differentiation, but this process often takes longer than several weeks to complete. A novel mechanism for somatic stem cell differentiation via phagocytosing ‘model cells’ (apoptotic differentiated cells) was found to require only a short time frame. Pluripotent-like Muse cells, multipotent mesenchymal stem cells (MSCs), and neural stem cells (NSCs) phagocytosed apoptotic differentiated cells via different phagocytic receptor subsets than macrophages. The phagocytosed-differentiated cell-derived contents (e.g., transcription factors) were quickly released into the cytoplasm, translocated into the nucleus, and bound to promoter regions of the stem cell genomes. Within 24 ~ 36 h, the cells expressed lineage-specific markers corresponding to the phagocytosed-differentiated cells, both in vitro and in vivo. At 1 week, the gene expression profiles were similar to those of the authentic differentiated cells and expressed functional markers. Differentiation was limited to the inherent potential of each cell line: triploblastic-, adipogenic-/chondrogenic-, and neural-lineages in Muse cells, MSCs, and NSCs, respectively. Disruption of phagocytosis, either by phagocytic receptor inhibition via small interfering RNA or annexin V treatment, impeded differentiation in vitro and in vivo. Together, our findings uncovered a simple mechanism by which differentiation-directing factors are directly transferred to somatic stem cells by phagocytosing apoptotic differentiated cells to trigger their rapid differentiation into the target cell lineage.

Single-Nucleus RNA-seq of Normal-Appearing Brain Regions in Relapsing-Remitting vs. Secondary Progressive Multiple Sclerosis: Implications for the Efficacy of Fingolimod

Multiple sclerosis (MS) is an immune-mediated demyelinating disease that alters central nervous system (CNS) functions. Relapsing-remitting MS (RRMS) is the most common form, which can transform into secondary-progressive MS (SPMS) that is associated with progressive neurodegeneration. Single-nucleus RNA sequencing (snRNA-seq) of MS lesions identified disease-related transcriptomic alterations; however, their relationship to non-lesioned MS brain regions has not been reported and which could identify prodromal or other disease susceptibility signatures. Here, snRNA-seq was used to generate high-quality RRMS vs. SPMS datasets of 33,197 nuclei from 8 normal-appearing MS brains, which revealed divergent cell type-specific changes. Notably, SPMS brains downregulated astrocytic sphingosine kinases (SPHK1/2) - the enzymes required to phosphorylate and activate the MS drug, fingolimod. This reduction was modeled with astrocyte-specific Sphk1/2 null mice in which fingolimod lost activity, supporting functionality of observed transcriptomic changes. These data provide an initial resource for studies of single cells from non-lesioned RRMS and SPMS brains.

Identification and classification of cis-regulatory elements in the amphipod crustacean Parhyale hawaiensis

Emerging research organisms enable the study of biology that cannot be addressed using classical 'model' organisms. New data resources can accelerate research in such animals. Here, we present new functional genomic resources for the amphipod crustacean Parhyale hawaiensis, facilitating the exploration of gene regulatory evolution using this emerging research organism. We use Omni-ATAC-seq to identify accessible chromatin genome-wide across a broad time course of Parhyale embryonic development. This time course encompasses many major morphological events, including segmentation, body regionalization, gut morphogenesis and limb development. In addition, we use short- and long-read RNA-seq to generate an improved Parhyale genome annotation, enabling deeper classification of identified regulatory elements. We discover differential accessibility, predict nucleosome positioning, infer transcription factor binding, cluster peaks based on accessibility dynamics, classify biological functions and correlate gene expression with accessibility. Using a Minos transposase reporter system, we demonstrate the potential to identify novel regulatory elements using this approach. This work provides a platform for the identification of novel developmental regulatory elements in Parhyale, and offers a framework for performing such experiments in other emerging research organisms.

Roles of plasma leptin and resistin in novel subgroups of type 2 diabetes driven by cluster analysis

Background

A novel classification has been introduced to promote precision medicine in diabetes. The current study aimed to investigate the relationship between leptin and resistin levels with novel refined subgroups in patients with type 2 diabetes mellitus (T2DM).

Methods

The k-means analysis was conducted to cluster 541 T2DM patients into the following four subgroups: mild obesity-related diabetes (MOD), severe insulin-deficient diabetes (SIDD), severe insulin-resistant diabetes (SIRD) and mild age-related diabetes (MARD). Individuals meeting the exclusion criteria were eliminated, the data for 285 patients were analyzed. Characteristics were determined using various clinical parameters. Both the leptin and resistin levels were determined using enzyme-linked immunosorbent assay.

Results

The highest levels of plasma leptin were in the MOD group with relatively lower levels in the SIDD and SIRD groups (P < 0.001). The SIRD group had a higher resistin concentration than the MARD group (P = 0.024) while no statistical significance in resistin levels was found between the SIDD and MOD groups. Logistic regression demonstrated that plasma resistin was associated with a higher risk of diabetic nephropathy (odds ratios (OR) = 2.255, P = 0.001). According to receiver operating characteristic (ROC) curves, the area under the curve (AUC) of resistin (0.748, 95% CI 0.610–0.887) was significantly greater than that of HOMA2-IR (0.447, 95% CI 0.280–0.614) (P < 0.05) for diabetic nephropathy in the SIRD group.

Conclusions

Leptin levels were different in four subgroups of T2DM and were highest in the MOD group. Resistin was elevated in the SIRD group and was closely related to diabetic nephropathy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12944-022-01623-z.

Lung Cancer Models Reveal Severe Acute Respiratory Syndrome Coronavirus 2-Induced Epithelial-to-Mesenchymal Transition Contributes to Coronavirus Disease 2019 Pathophysiology

INTRODUCTION

Coronavirus disease 2019 is an infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which enters host cells through the cell surface proteins ACE2 and TMPRSS2.

METHODS

Using a variety of normal and malignant models and tissues from the aerodigestive and respiratory tracts, we investigated the expression and regulation of ACE2 and TMPRSS2.

RESULTS

We find that ACE2 expression is restricted to a select population of epithelial cells. Notably, infection with SARS-CoV-2 in cancer cell lines, bronchial organoids, and patient nasal epithelium induces metabolic and transcriptional changes consistent with epithelial-to-mesenchymal transition (EMT), including up-regulation of ZEB1 and AXL, resulting in an increased EMT score. In addition, a transcriptional loss of genes associated with tight junction function occurs with SARS-CoV-2 infection. The SARS-CoV-2 receptor, ACE2, is repressed by EMT through the transforming growth factor-β, ZEB1 overexpression, and onset of EGFR tyrosine kinase inhibitor resistance. This suggests a novel model of SARS-CoV-2 pathogenesis in which infected cells shift toward an increasingly mesenchymal state, associated with a loss of tight junction components with acute respiratory distress syndrome-protective effects. AXL inhibition and ZEB1 reduction, as with bemcentinib, offer a potential strategy to reverse this effect.

CONCLUSIONS

These observations highlight the use of aerodigestive and, especially, lung cancer model systems in exploring the pathogenesis of SARS-CoV-2 and other respiratory viruses and offer important insights into the potential mechanisms underlying the morbidity and mortality of coronavirus disease 2019 in healthy patients and patients with cancer alike.

Unique Habitual Food Intakes in the Gut Microbiota Cluster Associated with Type 2 Diabetes Mellitus

This cross-sectional study aimed to clarify the characteristic gut microbiota of Japanese patients with type 2 diabetes (T2DM) using t-distributed stochastic neighbor embedding analysis and the k-means method and to clarify the relationship with background data, including dietary habits. The gut microbiota data of 383 patients with T2DM and 114 individuals without T2DM were classified into red, blue, green, and yellow groups. The proportions of patients with T2DM in the red, blue, green, and yellow groups was 86.8% (112/129), 69.8% (81/116), 76.3% (90/118), and 74.6% (100/134), respectively; the red group had the highest prevalence of T2DM. There were no intergroup differences in sex, age, or body mass index. The red group had higher percentages of the Bifidobacterium and Lactobacillus genera and lower percentages of the Blautia and Phascolarctobacterium genera. Higher proportions of patients with T2DM in the red group used α-glucosidase inhibitors and glinide medications and had a low intake of fermented soybean foods, including miso soup, than those in the other groups. The gut microbiota pattern of the red group may indicate characteristic changes in the gut microbiota associated with T2DM in Japan. These results also suggest that certain diabetes drugs and fermented foods may be involved in this change. Further studies are needed to confirm the relationships among traditional dietary habits, the gut microbiota, and T2DM in Japan.

Proteomics analysis identifies PEA-15 as an endosomal phosphoprotein that regulates α5β1 integrin endocytosis

Endosomal trafficking of cell surface receptors is essential to their function. Integrins are transmembrane receptors that integrate adhesion to the extracellular matrix with engagement of the cytoskeleton. Ligated integrins mediate diverse signals that regulate matrix assembly, cell survival, cell morphology, and cell motility. Endosomal trafficking of integrins modulates these signals and contributes to cell motility and is required for cancer cell invasion. The phosphoprotein PEA-15 modulates integrin activation and ERK MAP Kinase signaling. To elucidate novel PEA-15 functions we utilized an unbiased proteomics approach. We identified several binding partners for PEA-15 in the endosome including clathrin and AP-2 as well as integrin β1 and other focal adhesion complex proteins. We confirmed these interactions using proximity ligation analysis, immunofluorescence imaging, pull-down and co-immunoprecipitation. We further found that PEA-15 is enriched in endosomes and was required for efficient endosomal internalization of α5β1 integrin and cellular migration. Importantly, PEA-15 promotion of migration was dependent on PEA-15 phosphorylation at serines 104 and 116. These data support a novel endosomal role for PEA-15 in control of endosomal trafficking of integrins through an association with the β1 integrin and clathrin complexes, and thereby regulation of cell motility.

Multi-Stage Harmonization for Robust AI across Breast MR Databases

Simple Summary

Batch harmonization of radiomic features extracted from magnetic resonance images of breast lesions from two databases was applied to an artificial intelligence/machine learning classification workflow. Training and independent test sets from the two databases, as well as the combination of them, were used in pre-harmonization and post-harmonization forms to investigate the generalizability of performance in the task of distinguishing between malignant and benign lesions. Most training and independent test scenarios were statistically equivalent, demonstrating that batch harmonization with feature selection harmonization can potentially develop generalizable classification models.

Abstract

Radiomic features extracted from medical images may demonstrate a batch effect when cases come from different sources. We investigated classification performance using training and independent test sets drawn from two sources using both pre-harmonization and post-harmonization features. In this retrospective study, a database of thirty-two radiomic features, extracted from DCE-MR images of breast lesions after fuzzy c-means segmentation, was collected. There were 944 unique lesions in Database A (208 benign lesions, 736 cancers) and 1986 unique lesions in Database B (481 benign lesions, 1505 cancers). The lesions from each database were divided by year of image acquisition into training and independent test sets, separately by database and in combination. ComBat batch harmonization was conducted on the combined training set to minimize the batch effect on eligible features by database. The empirical Bayes estimates from the feature harmonization were applied to the eligible features of the combined independent test set. The training sets (A, B, and combined) were then used in training linear discriminant analysis classifiers after stepwise feature selection. The classifiers were then run on the A, B, and combined independent test sets. Classification performance was compared using pre-harmonization features to post-harmonization features, including their corresponding feature selection, evaluated using the area under the receiver operating characteristic curve (AUC) as the figure of merit. Four out of five training and independent test scenarios demonstrated statistically equivalent classification performance when compared pre- and post-harmonization. These results demonstrate that translation of machine learning techniques with batch data harmonization can potentially yield generalizable models that maintain classification performance.

Effect of topical applications of sunflower seed oil on systemic fatty acid levels in under-two children under rehabilitation for severe acute malnutrition in Bangladesh: a randomized controlled trial

Background

Children with severe acute malnutrition (SAM) have inadequate levels of fatty acids (FAs) and limited capacity for enteral nutritional rehabilitation. We hypothesized that topical high-linoleate sunflower seed oil (SSO) would be effective adjunctive treatment for children with SAM.

Methods

This study tested a prespecified secondary endpoint of a randomized, controlled, unblinded clinical trial with 212 children with SAM aged 2 to 24 months in two strata (2 to < 6 months, 6 to 24 months in a 1:2 ratio) at Dhaka Hospital of icddr,b, Bangladesh between January 2016 and December 2017. All children received standard-of-care management of SAM. Children randomized to the emollient group also received whole-body applications of 3 g/kg SSO three times daily for 10 days. We applied difference-in-difference analysis and unsupervised clustering analysis using t-distributed stochastic neighbor embedding (t-SNE) to visualize changes in FA levels in blood from day 0 to day 10 of children with SAM treated with emollient compared to no-emollient.

Results

Emollient therapy led to systematically higher increases in 26 of 29 FAs over time compared to the control. These effects were driven primarily by changes in younger subjects (27 of 29 FAs). Several FAs, especially those most abundant in SSO showed high-magnitude but non-significant incremental increases from day 0 to day 10 in the emollient group vs. the no-emollient group; for linoleic acid, a 237 μg/mL increase was attributable to enteral feeding and an incremental 98 μg/mL increase (41%) was due to emollient therapy. Behenic acid (22:0), gamma-linolenic acid (18:3n6), and eicosapentaenoic acid (20:5n3) were significantly increased in the younger age stratum; minimal changes were seen in the older children.

Conclusions

SSO therapy for SAM augmented the impact of enteral feeding in increasing levels of several FAs in young children. Further research is warranted into optimizing this novel approach for nutritional rehabilitation of children with SAM, especially those < 6 months.

Trial registration

ClinicalTrials.gov: NCT02616289.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12937-021-00707-3.

RNA sequencing reveals the expression profiles of circRNA and identifies a four-circRNA signature acts as a prognostic marker in esophageal squamous cell carcinoma

Background

CircRNAs with tissue-specific expression and stable structure may be good tumor prognostic markers. However, the expression of circRNAs in esophageal squamous cell carcinoma (ESCC) remain unknown. We aim to identify prognostic circRNAs and construct a circRNA-related signature in ESCC.

Methods

RNA sequencing was used to test the circRNA expression profiles of 73 paired ESCC tumor and normal tissues after RNase R enrichment. Bioinformatics methods, such as principal component analysis (PCA), t-distributed Stochastic Neighbor Embedding (t-SNE) algorithm, unsupervised clustering and hierarchical clustering were performed to analyze the circRNA expression characteristics. Univariate cox regression analysis, random survival forests-variable hunting (RSFVH), Kaplan–Meier analysis, multivariable Cox regression and ROC (receiver operating characteristic) curve analysis were used to screen the prognostic circRNA signature. Real-time quantitative PCR (qPCR) and fluorescence in situ hybridization(FISH) in 125 ESCC tissues were performed.

Results

Compared with normal tissues, there were 11651 differentially expressed circRNAs in cancer tissues. A total of 1202 circRNAs associated with ESCC prognosis (P < 0.05) were identified. Through bioinformatics analysis, we screened a circRNA signature including four circRNAs (hsa_circ_0000005, hsa_circ_0007541, hsa_circ_0008199, hsa_circ_0077536) which can classify the ESCC patients into two groups with significantly different survival (log rank P < 0.001), and found its predictive performance was better than that of the TNM stage(0.84 vs. 0.66; 0.65 vs. 0.62). Through qPCR and FISH experiment, we validated the existence of the screened circRNAs and the predictive power of the circRNA signature.

Conclusion

The prognostic four-circRNA signature could be a new prognostic biomarker for ESCC, which has high clinical application value.

Lung cancer models reveal SARS-CoV-2-induced EMT contributes to COVID-19 pathophysiology

COVID-19 is an infectious disease caused by SARS-CoV-2, which enters host cells via the cell surface proteins ACE2 and TMPRSS2. Using a variety of normal and malignant models and tissues from the aerodigestive and respiratory tracts, we investigated the expression and regulation of ACE2 and TMPRSS2. We find that ACE2 expression is restricted to a select population of highly epithelial cells. Notably, infection with SARS-CoV-2 in cancer cell lines, bronchial organoids, and patient nasal epithelium, induces metabolic and transcriptional changes consistent with epithelial to mesenchymal transition (EMT), including upregulation of ZEB1 and AXL, resulting in an increased EMT score. Additionally, a transcriptional loss of genes associated with tight junction function occurs with SARS-CoV-2 infection. The SARS-CoV-2 receptor, ACE2, is repressed by EMT via TGFbeta, ZEB1 overexpression and onset of EGFR TKI inhibitor resistance. This suggests a novel model of SARS-CoV-2 pathogenesis in which infected cells shift toward an increasingly mesenchymal state, associated with a loss of tight junction components with acute respiratory distress syndrome-protective effects. AXL-inhibition and ZEB1-reduction, as with bemcentinib, offers a potential strategy to reverse this effect. These observations highlight the utility of aerodigestive and, especially, lung cancer model systems in exploring the pathogenesis of SARS-CoV-2 and other respiratory viruses, and offer important insights into the potential mechanisms underlying the morbidity and mortality of COVID-19 in healthy patients and cancer patients alike.

Exercise modifies glutamate and other metabolic biomarkers in cerebrospinal fluid from Gulf War Illness and Myalgic encephalomyelitis / Chronic Fatigue Syndrome

Myalgic encephalomyelitis / Chronic Fatigue Syndrome (ME/CFS) and Gulf War Illness (GWI) share many symptoms of fatigue, pain, and cognitive dysfunction that are not relieved by rest. Patterns of serum metabolites in ME/CFS and GWI are different from control groups and suggest potential dysfunction of energy and lipid metabolism. The metabolomics of cerebrospinal fluid was contrasted between ME/CFS, GWI and sedentary controls in 2 sets of subjects who had lumbar punctures after either (a) rest or (b) submaximal exercise stress tests. Postexercise GWI and control subjects were subdivided according to acquired transient postexertional postural tachycardia. Banked cerebrospinal fluid specimens were assayed using Biocrates AbsoluteIDQ® p180 kits for quantitative targeted metabolomics studies of amino acids, amines, acylcarnitines, sphingolipids, lysophospholipids, alkyl and ether phosphocholines. Glutamate was significantly higher in the subgroup of postexercise GWI subjects who did not develop postural tachycardia after exercise compared to nonexercise and other postexercise groups. The only difference between nonexercise groups was higher lysoPC a C28:0 in GWI than ME/CFS suggesting this biochemical or phospholipase activities may have potential as a biomarker to distinguish between the 2 diseases. Exercise effects were suggested by elevation of short chain acylcarnitine C5-OH (C3-DC-M) in postexercise controls compared to nonexercise ME/CFS. Limitations include small subgroup sample sizes and absence of postexercise ME/CFS specimens. Mechanisms of glutamate neuroexcitotoxicity may contribute to neuropathology and “neuroinflammation” in the GWI subset who did not develop postural tachycardia after exercise. Dysfunctional lipid metabolism may distinguish the predominantly female ME/CFS group from predominantly male GWI subjects.

The Association between Antidiabetic Agents and Leukocyte Telomere Length in the Novel Classification of Type 2 Diabetes Mellitus

AIMS

This study aimed to explore the new role of telomere length (TL) in the novel classification of type 2 diabetes mellitus (T2DM) patients driven by cluster analysis.

MATERIALS AND METHODS

A total of 541 T2DM patients were divided into 4 subgroups by k-means analysis: mild obesity-related diabetes (MOD), severe insulin-deficient diabetes (SIDD), severe insulin-resistant diabetes (SIRD), and mild age-related diabetes (MARD). After patients with insufficient data were excluded, further analysis was conducted on 246 T2DM patients. The TL was detected using telomere restriction fragment, and the related diabetic indexes were also measured by clinical standard procedures.

RESULTS

The MARD group had significantly shorter TLs than the MOD and SIDD groups. Then, we subdivided all T2DM patients into the MARD and NONMARD groups, which included the MOD, SIDD, and SIRD groups. The TLs of the MARD group, associated with age, were discovered to be significantly shorter than those of the NONMARD group (p = 0.0012), and this difference in TL disappeared after metformin (p = 0.880) and acarbose treatment (p = 0.058). The linear analysis showed that metformin can more obviously reduce telomere shortening in the MARD group (r = 0.030, 95% CI 0.010-0.051, p = 0.004), and acarbose can more apparently promote telomere attrition in the SIRD group (r = -0.069, 95% CI -0.100 to -0.039, p< 0.001) compared with other T2DM patients after adjusting for age and gender.

CONCLUSIONS

The MARD group was found to have shorter TLs and benefit more from the antiaging effect of metformin than other T2DM. Shorter TLs were observed in the SIRD group after acarbose use.

Age-associated changes in the transcriptomes of non-cultured adipose-derived stem cells from young and old mice assessed via single-cell transcriptome analysis

Adipose-derived stem cells (ASCs) exhibit self-renewal and pluripotency. The differentiation potency of ASCs has been reported to deteriorate with aging; however, relevant studies used ASCs that were isolated and subcultured several times. It is still unclear whether subcultured ASCs accurately reflect the in vivo state. To address this question, we used freshly isolated stromal vascular fractions (SVFs) and performed comprehensive single-cell transcriptome analysis. In this study, we identified three cell populations as putative ASC candidates in SVFs and three novel ASC-related genes: Adamts7, Snai2, and Tgfbr1, that are reported to be negative regulators of cell differentiation. Moreover, we identified age-associated high gene expression levels of Adamts7, Egfr, and Igfbp4 in the earliest differentiation stage of ASCs. These results suggest that aging may make it impossible to maintain the stringency of the regulation of the expression of some genes related to ASC differentiation.

Isolation and transcriptional characterization of mouse perivascular astrocytes

In the post-natal mammalian brain perivascular astrocytes (PAs) ensheath blood vessels to regulate their unique permeability properties known as the blood-brain barrier (BBB). Very little is known about PA-expressed genes and signaling pathways that mediate contact and communication with endothelial cells (ECs) to regulate BBB physiology. This is due, in part, to lack of suitable models to distinguish PAs from other astrocyte sub-populations in the brain. To decipher the unique biology of PAs, we used in vivo gene knock-in technology to fluorescently label these cells in the adult mouse brain followed by fractionation and quantitative single cell RNA sequencing. In addition, PAs and non-PAs were also distinguished with transgenic fluorescent reporters followed by gene expression comparisons using bulk RNA sequencing. These efforts have identified several genes and pathways in PAs with potential roles in contact and communication with brain ECs. These genes encode various extracellular matrix (ECM) proteins and adhesion receptors, secreted growth factors, and intracellular signaling enzymes. Collectively, our experimental data reveal a set of genes that are expressed in PAs with putative roles in BBB physiology.

Single-cell analyses reveal increased intratumoral heterogeneity after the onset of therapy resistance in small-cell lung cancer

The natural history of small cell lung cancer (SCLC) includes rapid evolution from chemosensitivity to chemoresistance, although mechanisms underlying this evolution remain obscure due to scarcity of post-relapse tissue samples. We generated circulating tumor cell (CTC)-derived xenografts (CDXs) from SCLC patients to study intratumoral heterogeneity (ITH) via single-cell RNAseq of chemo-sensitive and -resistant CDXs and patient CTCs. We found globally increased ITH including heterogeneous expression of therapeutic targets and potential resistance pathways, such as EMT, between cellular subpopulations following treatment-resistance. Similarly, serial profiling of patient CTCs directly from blood confirmed increased ITH post-relapse. These data suggest that treatment-resistance in SCLC is characterized by coexisting subpopulations of cells with heterogeneous gene expression leading to multiple, concurrent resistance mechanisms. These findings emphasize the need for clinical efforts to focus on rational combination therapies for treatment-naïve SCLC tumors to maximize initial responses and counteract the emergence of ITH and diverse resistance mechanisms.

Multiplexed single-cell morphometry for hematopathology diagnostics

The diagnosis of lymphomas and leukemias requires hematopathologists to integrate microscopically visible cellular morphology with antibody-identified cell surface molecule expression. To merge these into one high-throughput, highly multiplexed, single-cell assay, we quantify cell morphological features by their underlying, antibody-measurable molecular components, which empowers mass cytometers to 'see' like pathologists. When applied to 71 diverse clinical samples, single-cell morphometric profiling reveals robust and distinct patterns of 'morphometric' markers for each major cell type. Individually, lamin B1 highlights acute leukemias, lamin A/C helps distinguish normal from neoplastic mature T cells, and VAMP-7 recapitulates light-cytometric side scatter. Combined with machine learning, morphometric markers form intuitive visualizations of normal and neoplastic cellular distribution and differentiation. When recalibrated for myelomonocytic blast enumeration, this approach is superior to flow cytometry and comparable to expert microscopy, bypassing years of specialized training. The contextualization of traditional surface markers on independent morphometric frameworks permits more sensitive and automated diagnosis of complex hematopoietic diseases.

Accelerating Medicines Partnership: Organizational Structure and Preliminary Data From the Phase 1 Studies of Lupus Nephritis

The Accelerating Medicines Partnership (AMP) Lupus Network was established as a partnership between the National Institutes of Health, pharmaceutical companies, nonprofit stakeholders, and lupus investigators across multiple academic centers to apply high-throughput technologies to the analysis of renal tissue, urine, and blood from patients with lupus nephritis (LN). The AMP network provides publicly accessible data to the community with the goal of generating new scientific hypotheses and improving diagnostic and therapeutic tools so as to improve disease outcomes. We present here a description of the structure of the AMP Lupus Network and a summary of the preliminary results from the phase 1 studies. The successful completion of phase 1 sets the stage for analysis of a large cohort of LN samples in phase 2 and provides a model for establishing similar discovery cohorts.

Artificial Intelligence: reshaping the practice of radiological sciences in the 21st century

Advances in computing hardware and software platforms have led to the recent resurgence in artificial intelligence (AI) touching almost every aspect of our daily lives by its capability for automating complex tasks or providing superior predictive analytics. AI applications are currently spanning many diverse fields from economics to entertainment, to manufacturing, as well as medicine. Since modern AI's inception decades ago, practitioners in radiological sciences have been pioneering its development and implementation in medicine, particularly in areas related to diagnostic imaging and therapy. In this anniversary article, we embark on a journey to reflect on the learned lessons from past AI's chequered history. We further summarize the current status of AI in radiological sciences, highlighting, with examples, its impressive achievements and effect on re-shaping the practice of medical imaging and radiotherapy in the areas of computer-aided detection, diagnosis, prognosis, and decision support. Moving beyond the commercial hype of AI into reality, we discuss the current challenges to overcome, for AI to achieve its promised hope of providing better precision healthcare for each patient while reducing cost burden on their families and the society at large.

Predicting structure/property relationships in multi-dimensional nanoparticle data using t-distributed stochastic neighbour embedding and machine learning

Combining researchers' domain expertise and advanced dimension reduction methods we demonstrate how visually comparing the distribution of nanoparticles mapped from multiple dimensions to a two dimensional plane can rapidly identify possible single-structure/property relationships and to a lesser extent multi-structure/property relationships. These relationships can be further investigated and confirmed with machine learning, using genetic programming to inform the choice of property-specific models and their hyper-parameters. In the case of our nanodiamond case study, we visually identify and confirm a strong relationship between the size and the probability of observation (stability) and a more complicated (and visually ambiguous) relationship between the ionisation potential and band gaps with a range of different structural, chemical and statistical surface features, making it more difficult to engineer in practice.

A novel automated image analysis system using deep convolutional neural networks can assist to differentiate MDS and AA

Detection of dysmorphic cells in peripheral blood (PB) smears is essential in diagnostic screening of hematological diseases. Myelodysplastic syndromes (MDS) are hematopoietic neoplasms characterized by dysplastic and ineffective hematopoiesis, which diagnosis is mainly based on morphological findings of PB and bone marrow. We developed an automated diagnostic support system of MDS by combining an automated blood cell image-recognition system using a deep learning system (DLS) powered by convolutional neural networks (CNNs) with a decision-making system using extreme gradient boosting (XGBoost). The DLS of blood cell image-recognition has been trained using datasets consisting of 695,030 blood cell images taken from 3,261 PB smears including hematopoietic malignancies. The DLS simultaneously classified 17 blood cell types and 97 morphological features of such cells with >93.5% sensitivity and >96.0% specificity. The automated MDS diagnostic system successfully differentiated MDS from aplastic anemia (AA) with high accuracy; 96.2% of sensitivity and 100% of specificity (AUC 0.990). This is the first CNN-based automated initial diagnostic system for MDS using PB smears, which is applicable to develop new automated diagnostic systems for various hematological disorders.

Finding Meanings in Low Dimensional Structures: Stochastic Neighbor Embedding Applied to the Analysis of Indri indri Vocal Repertoire

Simple Summary

The description of the vocal repertoire represents a critical step before deepening other aspects of animal behaviour. Repertoires may contain both discrete vocalizations—acoustically distinct and distinguishable from each other—or graded ones, with a less rigid acoustic structure. The gradation level is one of the causes that make repertoires challenging to be objectively quantified. Indeed, the higher the level of gradation in a system, the higher the complexity in grouping its components. A large sample of Indri indri calls was divided into ten putative categories from the acoustic similarity among them. We extracted frequency and duration parameters and then performed two different analyses that were able to group the calls accordingly to the a priori categories, indicating the presence of ten robust vocal classes. The analyses also showed a neat grouping of discrete vocalizations and a weaker classification of graded ones.

Abstract

Although there is a growing number of researches focusing on acoustic communication, the lack of shared analytic approaches leads to inconsistency among studies. Here, we introduced a computational method used to examine 3360 calls recorded from wild indris (Indri indri) from 2005–2018. We split each sound into ten portions of equal length and, from each portion we extracted spectral coefficients, considering frequency values up to 15,000 Hz. We submitted the set of acoustic features first to a t-distributed stochastic neighbor embedding algorithm, then to a hard-clustering procedure using a k-means algorithm. The t-distributed stochastic neighbor embedding (t-SNE) mapping indicated the presence of eight different groups, consistent with the acoustic structure of the a priori identification of calls, while the cluster analysis revealed that an overlay between distinct call types might exist. Our results indicated that the t-distributed stochastic neighbor embedding (t-SNE), successfully been employed in several studies, showed a good performance also in the analysis of indris’ repertoire and may open new perspectives towards the achievement of shared methodical techniques for the comparison of animal vocal repertoires.

Single-cell RNA sequencing identifies inflammatory tissue T cells in eosinophilic esophagitis

T cell heterogeneity is highly relevant to allergic disorders. We resolved the heterogeneity of human tissue CD3+ T cells during allergic inflammation, focusing on a tissue-specific allergic disease, eosinophilic esophagitis (EoE). We investigated 1088 single T cells derived from patients with a spectrum of disease activity. Eight disparate tissue T cell subtypes (designated T1-T8) were identified, with T7 and T8 enriched in the diseased tissue. The phenotypes of T7 and T8 resemble putative Treg (FOXP3+) and effector Th2-like (GATA3+) cells, respectively. Prodigious levels of IL-5 and IL-13 were confined to HPGDS+ CRTH2+IL-17RB+FFAR3+CD4+ T8 effector Th2 cells. EoE severity closely paralleled a lipid/fatty acid-induced activation node highlighted by the expression of the short-chain fatty acid receptor FFAR3. Ligands for FFAR3 induced Th2 cytokine production from human and murine T cells, including in an in vivo allergy model. Therefore, we elucidated the defining characteristics of tissue-residing CD3+ T cells in EoE, a specific enrichment of CD4+ Treg and effector Th2 cells, confinement of type 2 cytokine production to the CD4+ effector population, a highly likely role for FFAR3 in amplifying local Th2 responses in EoE, and a resource to further dissect tissue lymphocytes and allergic responses.

High-Dimensional Single-Cell Cartography Reveals Novel Skeletal Muscle-Resident Cell Populations

Adult tissue repair and regeneration require stem-progenitor cells that can self-renew and generate differentiated progeny. Skeletal muscle regenerative capacity relies on muscle satellite cells (MuSCs) and their interplay with different cell types within the niche. However, our understanding of skeletal muscle tissue cellular composition is limited. Here, using a combined approach of single-cell RNA sequencing and mass cytometry, we precisely mapped 10 different mononuclear cell types in adult mouse muscle. We also characterized gene signatures and determined key discriminating markers of each cell type. We identified two previously understudied cell populations in the interstitial compartment. One expresses the transcription factor scleraxis and generated tenocytes in vitro. The second expresses markers of smooth muscle and mesenchymal cells (SMMCs) and, while distinct from MuSCs, exhibited myogenic potential and promoted MuSC engraftment following transplantation. The blueprint presented here yields crucial insights into muscle-resident cell-type identities and can be exploited to study muscle diseases.

Artificial intelligence in cancer imaging: Clinical challenges and applications

Judgement, as one of the core tenets of medicine, relies upon the integration of multilayered data with nuanced decision making. Cancer offers a unique context for medical decisions given not only its variegated forms with evolution of disease but also the need to take into account the individual condition of patients, their ability to receive treatment, and their responses to treatment. Challenges remain in the accurate detection, characterization, and monitoring of cancers despite improved technologies. Radiographic assessment of disease most commonly relies upon visual evaluations, the interpretations of which may be augmented by advanced computational analyses. In particular, artificial intelligence (AI) promises to make great strides in the qualitative interpretation of cancer imaging by expert clinicians, including volumetric delineation of tumors over time, extrapolation of the tumor genotype and biological course from its radiographic phenotype, prediction of clinical outcome, and assessment of the impact of disease and treatment on adjacent organs. AI may automate processes in the initial interpretation of images and shift the clinical workflow of radiographic detection, management decisions on whether or not to administer an intervention, and subsequent observation to a yet to be envisioned paradigm. Here, the authors review the current state of AI as applied to medical imaging of cancer and describe advances in 4 tumor types (lung, brain, breast, and prostate) to illustrate how common clinical problems are being addressed. Although most studies evaluating AI applications in oncology to date have not been vigorously validated for reproducibility and generalizability, the results do highlight increasingly concerted efforts in pushing AI technology to clinical use and to impact future directions in cancer care.

Visualization of High-Dimensional Data by Pairwise Fusion Matrices Using t-SNE

We applied t-distributed stochastic neighbor embedding (t-SNE) to visualize Urdu handwritten numerals (or digits). The data set used consists of 28 × 28 images of handwritten Urdu numerals. The data set was created by inviting authors from different categories of native Urdu speakers. One of the challenging and critical issues for the correct visualization of Urdu numerals is shape similarity between some of the digits. This issue was resolved using t-SNE, by exploiting local and global structures of the large data set at different scales. The global structure consists of geometrical features and local structure is the pixel-based information for each class of Urdu digits. We introduce a novel approach that allows the fusion of these two independent spaces using Euclidean pairwise distances in a highly organized and principled way. The fusion matrix embedded with t-SNE helps to locate each data point in a two (or three-) dimensional map in a very different way. Furthermore, our proposed approach focuses on preserving the local structure of the high-dimensional data while mapping to a low-dimensional plane. The visualizations produced by t-SNE outperformed other classical techniques like principal component analysis (PCA) and auto-encoders (AE) on our handwritten Urdu numeral dataset.

Nonlinear expression and visualization of nonmetric relationships in genetic diseases and microbiome data

Background

The traditional methods of visualizing high-dimensional data objects in low-dimensional metric spaces are subject to the basic limitations of metric space. These limitations result in multidimensional scaling that fails to faithfully represent non-metric similarity data.

Results

Multiple maps t-SNE (mm-tSNE) has drawn much attention due to the construction of multiple mappings in low-dimensional space to visualize the non-metric pairwise similarity to eliminate the limitations of a single metric map. mm-tSNE regularization combines the intrinsic geometry between data points in a high-dimensional space. The weight of data points on each map is used as the regularization parameter of the manifold, so the weights of similar data points on the same map are also as close as possible. However, these methods use standard momentum methods to calculate parameters of gradient at each iteration, which may lead to erroneous gradient search directions so that the target loss function fails to achieve a better local minimum. In this article, we use a Nesterov momentum method to learn the target loss function and correct each gradient update by looking back at the previous gradient in the candidate search direction.

By using indirect second-order information, the algorithm obtains faster convergence than the original algorithm. To further evaluate our approach from a comparative perspective, we conducted experiments on several datasets including social network data, phenotype similarity data, and microbiomic data.

Conclusions

The experimental results show that the proposed method achieves better results than several versions of mm-tSNE based on three evaluation indicators including the neighborhood preservation ratio (NPR), error rate and time complexity.

A graph-based lesion characterization and deep embedding approach for improved computer-aided diagnosis of nonmass breast MRI lesions

Nonmass-like enhancements are a common but diagnostically challenging finding in breast MRI. Nonmass-like lesions can be described as clusters of spatially and temporally inter-connected regions of enhancements, so they can be modeled as networks and their properties characterized via network-based connectivity. In this work, we represented nonmass lesions as graphs using a link formation energy model that favors linkages between regions of similar enhancement and closer spatial proximity. However, adding graph features to an existing computer-aided diagnosis (CAD) pipeline incurs an increase of feature space dimensionality, which poses additional challenges to traditional supervised machine learning techniques due to the inability to increase accordingly the number of training datasets. We propose the combination of unsupervised dimensionality reduction and embedded space clustering followed by a supervised classifier to improve the performance of a CAD system for nonmass-like lesions in breast MRI. Our work extends a previoulsy proposed framework for deep embedded unsupervised clustering (DEC) to embedding space classification, with the joint optimization of objective functions for DEC and supervised multi-layered perceptron (MLP) classification. The strength of the method lies in the ability to learn and further optimize an embedded feature representation of lower dimensionality that maximizes the diagnostic accuracy of a CAD lesion classifier to discriminate between benign and malignant lesions. We identified 792 nonmass-like enhancements (267 benign, 110 malignant and 415 unknown) in 411 patients undergoing breast MRI at our institution. The diagnostic performance of the proposed method was evaluated and compared to the performance of a conventional supervised MLP classifier in original feature space. A statistically significant increase in diagnostic area under the ROC curve (AUC) was achieved. Generalization AUC increased from 0.67 ± 0.08 to 0.81 ± 0.10 (21% increase, p-value=4.2×10^-8) with the proposed graph-based lesion characterization and deep embedding framework.

Global transcriptional activity dynamics reveal functional enhancer RNAs

Active enhancers of the human genome generate long noncoding transcripts known as enhancer RNAs (eRNAs). How dynamic transcriptional changes of eRNAs are physically and functionally linked with target gene transcription remains unclear. To investigate the dynamic functional relationships among eRNAs and target promoters, we obtained a dense time series of GRO-seq and ChIP-seq data to generate a time-resolved enhancer activity map of a cell undergoing an innate antiviral immune response. Dynamic changes in eRNA and pre-mRNA transcription activities suggest distinct regulatory roles of enhancers. Using a criterion based on proximity and transcriptional inducibility, we identified 123 highly confident pairs of virus-inducible enhancers and their target genes. These enhancers interact with their target promoters transiently and concurrently at the peak of gene activation. Accordingly, their physical disassociation from the promoters is likely involved in post-induction repression. Functional assessments further establish that these eRNAs are necessary for full induction of the target genes and that a complement of inducible eRNAs functions together to achieve full activation. Lastly, we demonstrate the potential for eRNA-targeted transcriptional reprogramming through targeted reduction of eRNAs for a clinically relevant gene, TNFSF10, resulting in a selective control of interferon-induced apoptosis.

Subgroup-specific immune and stromal microenvironment in medulloblastoma

Knowledge on immune and stromal cells in medulloblastoma microenvironment is still limited as previous work was frequently restricted by low sample size and the lack of molecular subgroup information. We characterized 10 microenvironment cell populations as well as PD-L1 from gene expression in 1422 brain tumors and 763 medulloblastomas. All in all, medulloblastomas showed low expression of immune markers. Still, there were substantial differences with a clustering of medulloblastoma subgroups according to their microenvironment profile. Specifically, SHH medulloblastomas displayed strong signatures of fibroblasts, T cells and macrophages, while markers of cytotoxic lymphocytes were enriched in Group 4 tumors. PD-L1 gene expression appeared to be relatively high in single SHH and WNT cases but was undetectable by immunohistochemistry. In addition, two diverse immuno-stromal patterns were identified, indicating distinct types of local tumor immunosuppression, which were primarily controlled by either macrophage and regulatory T cell-mediated mechanisms or immunosuppressive cytokines and checkpoints, respectively. None of the immune cell signatures had an independent prognostic value in the present dataset after multiple testing correction. These results suggest a mild, but subgroup-specific infiltration of immune cells in medulloblastoma.

Visualizing histopathologic deep learning classification and anomaly detection using nonlinear feature space dimensionality reduction

BACKGROUND

There is growing interest in utilizing artificial intelligence, and particularly deep learning, for computer vision in histopathology. While accumulating studies highlight expert-level performance of convolutional neural networks (CNNs) on focused classification tasks, most studies rely on probability distribution scores with empirically defined cutoff values based on post-hoc analysis. More generalizable tools that allow humans to visualize histology-based deep learning inferences and decision making are scarce.

RESULTS

Here, we leverage t-distributed Stochastic Neighbor Embedding (t-SNE) to reduce dimensionality and depict how CNNs organize histomorphologic information. Unique to our workflow, we develop a quantitative and transparent approach to visualizing classification decisions prior to softmax compression. By discretizing the relationships between classes on the t-SNE plot, we show we can super-impose randomly sampled regions of test images and use their distribution to render statistically-driven classifications. Therefore, in addition to providing intuitive outputs for human review, this visual approach can carry out automated and objective multi-class classifications similar to more traditional and less-transparent categorical probability distribution scores. Importantly, this novel classification approach is driven by a priori statistically defined cutoffs. It therefore serves as a generalizable classification and anomaly detection tool less reliant on post-hoc tuning.

CONCLUSION

Routine incorporation of this convenient approach for quantitative visualization and error reduction in histopathology aims to accelerate early adoption of CNNs into generalized real-world applications where unanticipated and previously untrained classes are often encountered.

Machine learning for medical ultrasound: status, methods, and future opportunities

Ultrasound (US) imaging is the most commonly performed cross-sectional diagnostic imaging modality in the practice of medicine. It is low-cost, non-ionizing, portable, and capable of real-time image acquisition and display. US is a rapidly evolving technology with significant challenges and opportunities. Challenges include high inter- and intra-operator variability and limited image quality control. Tremendous opportunities have arisen in the last decade as a result of exponential growth in available computational power coupled with progressive miniaturization of US devices. As US devices become smaller, enhanced computational capability can contribute significantly to decreasing variability through advanced image processing. In this paper, we review leading machine learning (ML) approaches and research directions in US, with an emphasis on recent ML advances. We also present our outlook on future opportunities for ML techniques to further improve clinical workflow and US-based disease diagnosis and characterization.

Deep transfer learning for characterizing chondrocyte patterns in phase contrast X-Ray computed tomography images of the human patellar cartilage

Phase contrast X-ray computed tomography (PCI-CT) has been demonstrated to be effective for visualization of the human cartilage matrix at micrometer resolution, thereby capturing osteoarthritis induced changes to chondrocyte organization. This study aims to systematically assess the efficacy of deep transfer learning methods for classifying between healthy and diseased tissue patterns. We extracted features from two different convolutional neural network architectures, CaffeNet and Inception-v3 for characterizing such patterns. These features were quantitatively evaluated in a classification task measured by the area (AUC) under the Receiver Operating Characteristic (ROC) curve as well as qualitative visualization through a dimension reduction approach t-Distributed Stochastic Neighbor Embedding (t-SNE). The best classification performance, for CaffeNet, was observed when using features from the last convolutional layer and the last fully connected layer (AUCs >0.91). Meanwhile, off-the-shelf features from Inception-v3 produced similar classification performance (AUC >0.95). Visualization of features from these layers further confirmed adequate characterization of chondrocyte patterns for reliably distinguishing between healthy and osteoarthritic tissue classes. Such techniques, can be potentially used for detecting the presence of osteoarthritis related changes in the human patellar cartilage.

T cell developmental arrest in former premature infants increases risk of respiratory morbidity later in infancy

The inverse relationship between gestational age at birth and postviral respiratory morbidity suggests that infants born preterm (PT) may miss a critical developmental window of T cell maturation. Despite a continued increase in younger PT survivors with respiratory complications, we have limited understanding of normal human fetal T cell maturation, how ex utero development in premature infants may interrupt normal T cell development, and whether T cell development has an effect on infant outcomes. In our longitudinal cohort of 157 infants born between 23 and 42 weeks of gestation, we identified differences in T cells present at birth that were dependent on gestational age and differences in postnatal T cell development that predicted respiratory outcome at 1 year of age. We show that naive CD4+ T cells shift from a CD31-TNF-α+ bias in mid gestation to a CD31+IL-8+ predominance by term gestation. Former PT infants discharged with CD31+IL8+CD4+ T cells below a range similar to that of full-term born infants were at an over 3.5-fold higher risk for respiratory complications after NICU discharge. This study is the first to our knowledge to identify a pattern of normal functional T cell development in later gestation and to associate abnormal T cell development with health outcomes in infants.