Deep cross-modal feature learning applied to predict acutely decompensated heart failure using in-home collected electrocardiography and transthoracic bioimpedance.

BACKGROUND: Deep learning has been successfully applied to ECG data to aid in the accurate and more rapid diagnosis of acutely decompensated heart failure (ADHF). Previous applications focused primarily on classifying known ECG patterns in well-controlled clinical settings. However, this approach does not fully capitalize on the potential of deep learning, which directly learns important features without relying on a priori knowledge. In addition, deep learning applications to ECG data obtained from wearable devices have not been well studied, especially in the field of ADHF prediction. METHODS: We used ECG and transthoracic bioimpedance data from the SENTINEL-HF study, which enrolled patients (≥21 years) who were hospitalized with a primary diagnosis of heart failure or with ADHF symptoms. To build an ECG-based prediction model of ADHF, we developed a deep cross-modal feature learning pipeline, termed ECGX-Net, that utilizes raw ECG time series and transthoracic bioimpedance data from wearable devices. To extract rich features from ECG time series data, we first adopted a transfer learning approach in which ECG time series were transformed into 2D images, followed by feature extraction using ImageNet-pretrained DenseNet121/VGG19 models. After data filtering, we applied cross-modal feature learning in which a regressor was trained with ECG and transthoracic bioimpedance. Then, we concatenated the DenseNet121/VGG19 features with the regression features and used them to train a support vector machine (SVM) without bioimpedance information. RESULTS: The high-precision classifier using ECGX-Net predicted ADHF with a precision of 94 %, a recall of 79 %, and an F1-score of 0.85. The high-recall classifier with only DenseNet121 had a precision of 80 %, a recall of 98 %, and an F1-score of 0.88. We found that ECGX-Net was effective for high-precision classification, while DenseNet121 was effective for high-recall classification. CONCLUSION: We show the potential for predicting ADHF from single-channel ECG recordings obtained from outpatients, enabling timely warning signs of heart failure. Our cross-modal feature learning pipeline is expected to improve ECG-based heart failure prediction by handling the unique requirements of medical scenarios and resource limitations.

Evaluation of peripheral nerve acute crush injury in rabbits: comparison among diffusion kurtosis imaging, diffusion tensor imaging and electromyography.

OBJECTIVE: Diffusion kurtosis imaging (DKI) has been proven to provide additional value for assessing many central nervous system diseases compared with conventional diffusion tensor imaging (DTI); however, whether it has the same value in peripheral nerve injury is unclear. This study aimed to investigate the performance of DKI, DTI, and electromyography (EMG) in evaluating peripheral nerve crush injury (PNCI) in rabbits. MATERIALS AND METHODS: A total of 27 New Zealand white rabbits were selected to establish a PNCI model. Longitudinal DTI, DKI, and EMG were evaluated before surgery and 1 day, 3 days, 1 week, 2 weeks, 4 weeks, 6 weeks, and 8 weeks after surgery. At each time point, two rabbits were randomly selected for pathological examination. RESULTS: The results showed that fractional anisotropy (FA) derived from both DKI and DTI demonstrated a significant difference between injured and control nerves at all time points (all P < 0.005) mean kurtosis of the injured nerve was lower than that on the control side after 2-8 weeks (all P < 0.05). No statistically significant difference was found in radial kurtosis, axial kurtosis, and apparent diffusion coefficient at almost every time point. The difference in compound muscle action potential (CMAP) of the bilateral gastrocnemius at each time point was statistically significant (all P < 0.001). CONCLUSIONS: CMAP was a sensitive and reliable method to assess acute PNCI without being affected by perineural edema. DKI may not be superior to DTI in evaluating peripheral nerves, DTI with a shorter scanning time was preferred as an effective choice for evaluating acute peripheral nerve traumatic injury.

A deep learning-based segmentation pipeline for profiling cellular morphodynamics using multiple types of live cell microscopy.

MOTIVATION: Quantitative studies of cellular morphodynamics rely on extracting leading-edge velocity time series based on accurate cell segmentation from live cell imaging. However, live cell imaging has numerous challenging issues regarding accurate edge localization. Fluorescence live cell imaging produces noisy and low-contrast images due to phototoxicity and photobleaching. While phase contrast microscopy is gentle to live cells, it suffers from the halo and shade-off artifacts that cannot be handled by conventional segmentation algorithms. Here, we present a deep learning-based pipeline, termed MARS-Net (Multiple-microscopy-type-based Accurate and Robust Segmentation Network), that utilizes transfer learning and data from multiple types of microscopy to localize cell edges with high accuracy, allowing quantitative profiling of cellular morphodynamics. SUMMARY: To accurately segment cell edges and quantify cellular morphodynamics from live-cell imaging data, we developed a deep learning-based pipeline termed MARS-Net (multiple-microscopy-type-based accurate and robust segmentation network). MARS-Net utilizes transfer learning and data from multiple types of microscopy to localize cell edges with high accuracy. For effective training on distinct types of live-cell microscopy, MARS-Net comprises a pretrained VGG19 encoder with U-Net decoder and dropout layers. We trained MARS-Net on movies from phase-contrast, spinning-disk confocal, and total internal reflection fluorescence microscopes. MARS-Net produced more accurate edge localization than the neural network models trained with single-microscopy-type datasets. We expect that MARS-Net can accelerate the studies of cellular morphodynamics by providing accurate pixel-level segmentation of complex live-cell datasets.

A machine learning pipeline revealing heterogeneous responses to drug perturbations on vascular smooth muscle cell spheroid morphology and formation.

Machine learning approaches have shown great promise in biology and medicine discovering hidden information to further understand complex biological and pathological processes. In this study, we developed a deep learning-based machine learning algorithm to meaningfully process image data and facilitate studies in vascular biology and pathology. Vascular injury and atherosclerosis are characterized by neointima formation caused by the aberrant accumulation and proliferation of vascular smooth muscle cells (VSMCs) within the vessel wall. Understanding how to control VSMC behaviors would promote the development of therapeutic targets to treat vascular diseases. However, the response to drug treatments among VSMCs with the same diseased vascular condition is often heterogeneous. Here, to identify the heterogeneous responses of drug treatments, we created an in vitro experimental model system using VSMC spheroids and developed a machine learning-based computational method called HETEROID (heterogeneous spheroid). First, we established a VSMC spheroid model that mimics neointima-like formation and the structure of arteries. Then, to identify the morphological subpopulations of drug-treated VSMC spheroids, we used a machine learning framework that combines deep learning-based spheroid segmentation and morphological clustering analysis. Our machine learning approach successfully showed that FAK, Rac, Rho, and Cdc42 inhibitors differentially affect spheroid morphology, suggesting that multiple drug responses of VSMC spheroid formation exist. Overall, our HETEROID pipeline enables detailed quantitative drug characterization of morphological changes in neointima formation, that occurs in vivo, by single-spheroid analysis.

Epithelioid hemangioendothelioma of right innominate vein mimics a teratoma: A case report.

RATIONAL: Epithelioid hemangioendothelioma (EHE) is a rare neoplasm commonly known to arise from the soft tissue, lung, and liver. EHE arising from right innominate vein (RIV) has scarcely been reported in English literature. PATIENT CONCERNS: Herein, we present a rare case of EHE of RIV in a 51-year-old woman with right-lower chest pain for 4 days. Computed tomography of the chest revealed a spherical mass with calcification and fatty foci located in the anterior mediastinum, thus a presumptive diagnosis of teratoma was made. DIAGNOSES, INTERVENTIONS, AND OUTCOMES: Video-assisted thoracoscopic explorations and resection of mediastinal tumor were then performed. The pathological examination showed that the tumor was EHE. Postoperative radiotherapy was delivered to the patient. Pulmonary metastases were found by chest CT a year after surgery. LESSONS: A diagnosis of EHE might be considered, when a mediastinal tumor closely related to veins showing intratumoral calcification and obvious enhancement, despite the presence of a clear boundary and visible fat content.

Intravoxel incoherent motion diffusion-weighted MR imaging in assessing and characterizing solitary pulmonary lesions.

This study aimed to investigate the potential of intravoxel incoherent motion (IVIM) diffusion-weighted MR imaging in assessing solitary pulmonary lesions (SPLs). Sixty-two patients with pathologically confirmed SPLs, including 51 and 11 cases of malignant and benign lesions, respectively, were assessed. Diffusion weighted imaging (DWI) with 13 b values was used to derive apparent diffusion coefficient (ADC) and IVIM parameters, including true diffusion coefficient (D), pseudo-diffusion coefficient (D*), and perfusion fraction (f). Our results showed that, there was an excellent inter-observer agreement on the measurements of D and ADC between observers (inter-class correlation coefficient, ICC = 0.902 and 0.884, respectively). Meanwhile, f and D* showed good and substantial reproducibility (ICC = 0.787 and 0.623, respectively). D and ADC of malignant lesions were significantly lower than those of benign lesions (both P ≤ 0.001), while similar values were obtained in both groups for D* and f (both P > 0.05). In receiver operating characteristic (ROC) analysis, D showed the highest area under curve (AUC) for distinguishing malignant from benign lesions, followed by ADC. Accompanying signs of SPLs have specific features on IVIM maps. In conclusion, IVIM provides functional information in characterizing SPLs which is helpful to differential diagnosis. D and ADC have a significantly higher diagnostic value than f and D*.