Abstract
In this paper we study the heart sound segmentation problem using Deep Neural Networks. The impact of available electrocardiogram (ECG) signals in addition to phonocardiogram (PCG) signals is evaluated. To incorporate ECG, two different models considered, which are built upon a 1D U-net - an early fusion one that fuses ECG in an early processing stage, and a late fusion one that averages the probabilities obtained by two networks applied independently on PCG and ECG data. Results show that, in contrast with traditional uses of ECG for PCG gating, early fusion of PCG and ECG information can provide more robust heart sound segmentation. As a proof of concept, we use the publicly available PhysioNet dataset. Validation results provide, on average, a sensitivity of 97.2%, 94.5%, and 95.6% and a Positive Predictive Value of 97.5%, 96.2%, and 96.1% for Early-fusion, Late-fusion, and unimodal (PCG only) models, respectively, showing the advantages of combining both signals at early stages to segment heart sounds.

Abstract
The use of contrast-enhanced computed tomography (CTCA) for detection of coronary artery disease (CAD) exposes patients to the risks of iodine contrast-agents and excessive radiation, increases scanning time and healthcare costs. Deep learning generative models have the potential to artificially create a pseudo-enhanced image from non-contrast computed tomography (CT) scans. In this work, two specific models of generative adversarial networks (GANs) - the Pix2Pix-GAN and the Cycle-GAN - were tested with paired non-contrasted CT and CTCA scans from a private and public dataset. Furthermore, an exploratory analysis of the trade-off of using 2D and 3D inputs and architectures was performed. Using only the Structural Similarity Index Measure (SSIM) and the Peak Signal-to-Noise Ratio (PSNR), it could be concluded that the Pix2Pix-GAN using 2D data reached better results with 0.492 SSIM and 16.375 dB PSNR. However, visual analysis of the output shows significant blur in the generated images, which is not the case for the Cycle-GAN models. This behavior can be captured by the evaluation of the Fr ' echet Inception Distance (FID), that represents a fundamental performance metric that is usually not considered by related works in the literature.

Abstract
An algorithm for blind source separation (BSS) of the second heart sound (S2) into aortic and pulmonary components is proposed. It recovers aortic (A2) and pulmonary (P2) waveforms, as well as their relative delays, by solving an alternating optimization problem on the set of S2 sounds, without the use of auxiliary ECG or respiration phase measurement data. This unsupervised and data-driven approach assumes that the A2 and P2 components maintain the same waveform across heartbeats and that the relative delay between onset of the components varies according to respiration phase. The proposed approach is applied to synthetic heart sounds and to real-world heart sounds from 43 patients. It improves over two state-of-the-art BSS approaches by 10% normalized root mean-squared error in the reconstruction of aortic and pulmonary components using synthetic heart sounds, demonstrates robustness to noise, and recovery of splitting delays. The detection of pulmonary hypertension (PH) in a Brazilian population is demonstrated by training a classifier on three scalar features from the recovered A2 and P2 waveforms, and this yields an auROC of 0.76.

Abstract
Gastric Intestinal Metaplasia (GIM) is a precancerous gastric lesion and its early detection facilitates patient followup, thus lowering significantly the risk of death by gastric cancer. However, effective screening of this condition is a very challenging task, resulting low intra and inter-observer concordance. Computer assisted diagnosis systems leveraging deep neural networks (DNNs) have emerged as a way to mitigate these ailments. Notwithstanding, these approaches typically require large datasets in order to learn invariance to the extreme variations typically present in Esophagogastroduodenoscopy (EGD) still frames, such as perspective, illumination, and scale. Hence, we propose to combine a priori information regarding texture characteristics of GIM with data-driven DNN solutions. In particular, we define two different models that treat pre-trained DNNs as general features extractors, whose pairwise interactions with a collection of highly invariant local texture descriptors grounded on fractal geometry are computed by means of an outer product in the embedding space. Our experiments show that these models outperform a baseline DNN by a significant margin over several metrics (e.g., area under the curve (AUC) 0.792 vs. 0.705) in a dataset comprised of EGD narrow-band images. Our best model measures double the positive likelihood ratio when compared to a baseline GIM detector.

Abstract
An early detection of neurodegenerative diseases, such as Parkinson’s disease, can improve therapy effectiveness and, by consequence, the patient’s quality of life. This paper proposes a new methodology for automatic classification of voice samples regarding the presence of acoustic patterns of Parkinson’s disease, using a deep structured neural network. This is a low cost non-invasive approach that can raise alerts in a pre-clinical stage. Aiming to a higher diagnostic detail, it is also an objective to accurately estimate the stage of evolution of the disease allowing to understand in what extent the symptoms have developed. Therefore, two types of classification problems are explored: binary classification and multiclass classification. For binary classification, a deep structured neural network was developed, capable of correctly diagnosing 93.4% of cases. For the multiclass classification scenario, in addition to the deep neural network, a K-nearest neighbour algorithm was also used to establish a reference for comparison purposes, while using a common database. In both cases the original feature set was optimized using principal component analysis and the results showed that the proposed deep structure neural network was able to provide more accurate estimations about the disease’s stage, reaching a score of 84.7%. The obtained results are promising and create the motivation to further explore the model’s flexibility and to pursue better results. © 2020, Springer Nature Switzerland AG.

Abstract
E-commerce has become a booming market for wildlife trafficking, as online platforms are increasingly more accessible and easier to navigate by sellers, while still lacking adequate supervision. Artificial intelligence models, and specifically deep learning, have been emerging as promising tools for the automated analysis and monitoring of digital online content pertaining to wildlife trade. Here, we used and fine-tuned freely available artificial intelligence models (i.e., convolutional neural networks) to understand the potential of these models to identify instances of wildlife trade. We specifically focused on pangolin species, which are among the most trafficked mammals globally and receiving increasing trade attention since the COVID-19 pandemic. Our convolutional neural networks were trained using online images (available from iNaturalist, Flickr and Google) displaying both traded and non-traded pangolin settings. The trained models showed great performances, being able to identify over 90 % of potential instances of pangolin trade in the considered imagery dataset. These instances included the showcasing of pangolins in popular marketplaces (e.g., wet markets and cages), and the displaying of commonly traded pangolin parts and derivates (e.g., scales) online. Nevertheless, not all instances of pangolin trade could be identified by our models (e.g., in images with dark colours and shaded areas), leaving space for further research developments. The methodological developments and results from this exploratory study represent an advancement in the monitoring of online wildlife trade. Complementing our approach with other forms of online data, such as text, would be a way forward to deliver more robust monitoring tools for online trafficking.