Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

Radiology Radiology

A Fully Automatic Deep Learning System for COVID-19 Diagnostic and Prognostic Analysis.

In The European respiratory journal

Coronavirus disease 2019 (COVID-19) has spread globally, and medical resources become insufficient in many regions. Fast diagnosis of COVID-19, and finding high-risk patients with worse prognosis for early prevention and medical resources optimisation is important. Here, we proposed a fully automatic deep learning system for COVID-19 diagnostic and prognostic analysis by routinely used computed tomography.We retrospectively collected 5372 patients with computed tomography images from 7 cities or provinces. Firstly, 4106 patients with computed tomography images were used to pre-train the DL system, making it learn lung features. Afterwards, 1266 patients (924 with COVID-19, and 471 had follow-up for 5+ days; 342 with other pneumonia) from 6 cities or provinces were enrolled to train and externally validate the performance of the deep learning system.In the 4 external validation sets, the deep learning system achieved good performance in identifying COVID-19 from other pneumonia (AUC=0.87 and 0.88) and viral pneumonia (AUC=0.86). Moreover, the deep learning system succeeded to stratify patients into high-risk and low-risk groups whose hospital-stay time have significant difference (p=0.013 and 0.014). Without human-assistance, the deep learning system automatically focused on abnormal areas that showed consistent characteristics with reported radiological findings.Deep learning provides a convenient tool for fast screening COVID-19 and finding potential high-risk patients, which may be helpful for medical resource optimisation and early prevention before patients show severe symptoms.

Wang Shuo, Zha Yunfei, Li Weimin, Wu Qingxia, Li Xiaohu, Niu Meng, Wang Meiyun, Qiu Xiaoming, Li Hongjun, Yu He, Gong Wei, Bai Yan, Li Li, Zhu Yongbei, Wang Liusu, Tian Jie


Radiology Radiology

Comparison of 3 deep learning neural networks for classifying the relationship between the mandibular third molar and the mandibular canal on panoramic radiographs.

In Oral surgery, oral medicine, oral pathology and oral radiology ; h5-index 33.0

OBJECTIVE : The aim of this study was to compare time and storage space requirements, diagnostic performance, and consistency among 3 image recognition convolutional neural networks (CNNs) in the evaluation of the relationships between the mandibular third molar and the mandibular canal on panoramic radiographs.

STUDY DESIGN : Of 600 panoramic radiographs, 300 each were assigned to noncontact and contact groups based on the relationship between the mandibular third molar and the mandibular canal. The CNNs were trained twice by using cropped image patches with sizes of 70 × 70 pixels and 140 × 140 pixels. Time and storage space were measured for each system. Accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve (AUC) were determined. Intra-CNN and inter-CNN consistency values were calculated.

RESULTS : Time and storage space requirements depended on the depth of CNN layers and number of learned parameters, respectively. The highest AUC values ranged from 0.88 to 0.93 in the CNNs created by 70 × 70 pixel patches, but there were no significant differences in diagnostic performance among any of the models with smaller patches. Intra-CNN and inter-CNN consistency values were good or very good for all CNNs.

CONCLUSIONS : The size of the image patches should be carefully determined to ensure acquisition of high diagnostic performance and consistency.

Fukuda Motoki, Ariji Yoshiko, Kise Yoshitaka, Nozawa Michihito, Kuwada Chiaki, Funakoshi Takuma, Muramatsu Chisako, Fujita Hiroshi, Katsumata Akitoshi, Ariji Eiichiro


General General

Enhancer Predictions and Genome-Wide Regulatory Circuits.

In Annual review of genomics and human genetics

Spatiotemporal control of gene expression during development requires orchestrated activities of numerous enhancers, which are cis-regulatory DNA sequences that, when bound by transcription factors, support selective activation or repression of associated genes. Proper activation of enhancers is critical during embryonic development, adult tissue homeostasis, and regeneration, and inappropriate enhancer activity is often associated with pathological conditions such as cancer. Multiple consortia [e.g., the Encyclopedia of DNA Elements (ENCODE) Consortium and National Institutes of Health Roadmap Epigenomics Mapping Consortium] and independent investigators have mapped putative regulatory regions in a large number of cell types and tissues, but the sequence determinants of cell-specific enhancers are not yet fully understood. Machine learning approaches trained on large sets of these regulatory regions can identify core transcription factor binding sites and generate quantitative predictions of enhancer activity and the impact of sequence variants on activity. Here, we review these computational methods in the context of enhancer prediction and gene regulatory network models specifying cell fate. Expected final online publication date for the Annual Review of Genomics and Human Genetics, Volume 21 is August 31, 2020. Please see for revised estimates.

Beer Michael A, Shigaki Dustin, Huangfu Danwei


Cardiology Cardiology

Automatic Detection of Atrial Fibrillation Based on CNN-LSTM and Shortcut Connection.

In Healthcare (Basel, Switzerland)

Atrial fibrillation (AF) is one of the most common persistent arrhythmias, which has a close connection to a large number of cardiovascular diseases. However, if spotted early, the diagnosis of AF can improve the effectiveness of clinical treatment and effectively prevent serious complications. In this paper, a combination of an 8-layer convolutional neural network (CNN) with a shortcut connection and 1-layer long short-term memory (LSTM), named 8CSL, was proposed for the Electrocardiogram (ECG) classification task. Compared with recurrent neural networks (RNN) and multi-scale convolution neural networks (MCNN), not only can 8CSL extract features skillfully, but also deal with long-term dependency between data. In particular, 8CSL includes eight shortcut connections that can improve the speed of the data transmission and processing as a result of the shortcut connections. The model was evaluated on the base of the test set of the Computing in Cardiology Challenge 2017 dataset with the F1 score. The ECG recordings were cropped or padded to the same length. After 10-fold cross-validation, the average test F1 score was 84.89%, 89.55%, and 85.64% when the segment length was 5, 10, 20 seconds, respectively. The experiment results demonstrate excellent performance with potential practical applications.

Ping Yongjie, Chen Chao, Wu Lu, Wang Yinglong, Shu Minglei


CNN with shortcut connection, atrial fibrillation (AF), long short-term memory (LSTM)

Surgery Surgery

Converging intracortical signatures of two separated processing timescales in human early auditory cortex.

In NeuroImage ; h5-index 117.0

Neural oscillations in auditory cortex are argued to support parsing and representing speech constituents at their corresponding temporal scales. Yet, how incoming sensory information interacts with ongoing spontaneous brain activity, what features of the neuronal microcircuitry underlie spontaneous and stimulus-evoked spectral fingerprints, and what these fingerprints entail for stimulus encoding, remain largely open questions. We used a combination of human invasive electrophysiology, computational modeling and decoding techniques to assess the information encoding properties of brain activity and to relate them to a plausible underlying neuronal microarchitecture. We analyzed intracortical auditory EEG activity from 10 patients while they were listening to short sentences. Pre-stimulus neural activity in early auditory cortical regions often exhibited power spectra with a shoulder in the delta range and a small bump in the beta range. Speech decreased power in the beta range, and increased power in the delta-theta and gamma ranges. Using multivariate machine learning techniques, we assessed the spectral profile of information content for two aspects of speech processing: detection and discrimination. We obtained better phase than power information decoding, and a bimodal spectral profile of information content with better decoding at low (delta-theta) and high (gamma) frequencies than at intermediate (beta) frequencies. These experimental data were reproduced by a simple rate model made of two subnetworks with different timescales, each composed of coupled excitatory and inhibitory units, and connected via a negative feedback loop. Modeling and experimental results were similar in terms of pre-stimulus spectral profile (except for the iEEG beta bump), spectral modulations with speech, and spectral profile of information content. Altogether, we provide converging evidence from both univariate spectral analysis and decoding approaches for a dual timescale processing infrastructure in human auditory cortex, and show that it is consistent with the dynamics of a simple rate model.

Baroni Fabiano, Morillon Benjamin, Trébuchon Agnès, Liégeois-Chauvel Catherine, Olasagasti Itsaso, Giraud Anne-Lise


auditory cortex, brain decoding, computational modeling, iEEG, spectral analysis, speech perception

General General

Predictive regression modeling with MEG/EEG: from source power to signals and cognitive states.

In NeuroImage ; h5-index 117.0

Predicting biomedical outcomes from Magnetoencephalography and Electroencephalography (M/EEG) is central to applications like decoding, brain-computer-interfaces (BCI) or biomarker development and is facilitated by supervised machine learning. Yet most of the literature is concerned with classification of outcomes defined at the event-level. Here, we focus on predicting continuous outcomes from M/EEG signal defined at the subject-level, and analyze about 600 MEG recordings from Cam-CAN dataset and about 1000 EEG recordings from TUH dataset. Considering different generative mechanisms for M/EEG signals and the biomedical outcome, we propose statistically-consistent predictive models that avoid source-reconstruction based on the covariance as representation. Our mathematical analysis and ground truth simulations demonstrated that consistent function approximation can be obtained with supervised spatial filtering or by embedding with Riemannian geometry. Additional simulations revealed that Riemannian methods were more robust to model violations, in particular geometric distortions induced by individual anatomy. To estimate the relative contribution of brain dynamics and anatomy to prediction performance, we propose a novel model inspection procedure based on biophysical forward modeling. Applied to prediction of outcomes at the subject-level, the analysis revealed that the Riemannian model better exploited anatomical information while sensitivity to brain dynamics was similar across methods. We then probed the robustness of the models across different data cleaning options. Environmental denoising was globally important but Riemannian models were strikingly robust and continued performing well even without preprocessing. Our results suggest each method has its niche: supervised spatial filtering is practical for event-level prediction while the Riemannian model may enable simple end-to-end learning.

Sabbagh David, Ablin Pierre, Varoquaux Gaël, Gramfort Alexandre, Engemann Denis A


Covariance, MEG/EEG, Machine Learning, Neuronal oscillations, Riemannian Geometry, Spatial Filters