Advances in Unmanned Aerial Vehicles (UAVs), also known as drones, offer unprecedented opportunities to boost a wide array of large-scale Internet of Things (IoT) applications. Nevertheless, UAV platforms still face important limitations mainly related to autonomy and weight that impact their remote sensing capabilities when capturing and processing the data required for developing autonomous and robust real-time obstacle detection and avoidance systems. In this regard, Deep Learning (DL) techniques have arisen as a promising alternative for improving real-time obstacle detection and collision avoidance for highly autonomous UAVs. This article reviews the most recent developments on DL Unmanned Aerial Systems (UASs) and provides a detailed explanation on the main DL techniques. Moreover, the latest DL-UAV communication architectures are studied and their most common hardware is analyzed. Furthermore, this article enumerates the most relevant open challenges for current DL-UAV solutions, thus allowing future researchers to define a roadmap for devising the new generation affordable autonomous DL-UAV IoT solutions.

The recent human coronavirus disease (COVID-19) is a respiratory infection caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Given the effects of COVID-19 in pulmonary tissues, chest radiography imaging plays an important role in the screening, early detection, and monitoring of the suspected individuals. Hence, as the pandemic of COVID-19 progresses, there will be a greater reliance on the use of portable equipment for the acquisition of chest X-ray images due to its accessibility, widespread availability, and benefits regarding to infection control issues, minimizing the risk of cross-contamination. This work presents novel fully automatic approaches specifically tailored for the classification of chest X-ray images acquired by portable equipment into 3 different clinical categories: normal, pathological, and COVID-19. For this purpose, 3 complementary deep learning approaches based on a densely convolutional network architecture are herein presented. The joint response of all the approaches allows to enhance the differentiation between patients infected with COVID-19, patients with other diseases that manifest characteristics similar to COVID-19 and normal cases. The proposed approaches were validated over a dataset specifically retrieved for this research. Despite the poor quality of the chest X-ray images that is inherent to the nature of the portable equipment, the proposed approaches provided global accuracy values of 79.62%, 90.27% and 79.86%, respectively, allowing a reliable analysis of portable radiographs to facilitate the clinical decision-making process.

This study was designed to examine diurnal variations in tear film break-up time (BUT) and maximum blink interval (MBI) and to assess two different ways of calculating these variables on video recordings of the BUT test interpreted with the help of especially designed software. The repeatability of interpreting BUT video recordings was also addressed. Twenty-six healthy young adults were enrolled after ruling out dry eye according to a battery of tests (ocular surface disease index, McMonnies questionnaire, Schirmer test, phenol red test and corneal staining). BUT and maximum blink interval were determined on video-recordings of the BUT test conducted over a day in four sessions (9.30 am, 12.30 pm, 3.30 pm and 6.30 pm). In each session, the test was repeated three times to give three videos in which three BUT and MBI values were obtained by a masked observer. BUT and MBI were determined by averaging the three measurements and by averaging only the two closest measurements. Finally, two further experienced observers re-examined the videos to assess the repeatability of the BUT measurements made. No diurnal variation in BUT was observed regardless of whether three or two video measurements were averaged. Significant correlation was detected between BUT and MBI. Inter-observer repeatability was better when BUT times were no longer than 15 seconds. Tear film BUT was not influenced by the time of day and moderate to strong correlation with MBI was observed in all four sessions. The software-assisted method proved useful and identified the need to clarify the BUT end-point and to limit the test to 15 seconds to improve observer repeatability.

The retinal vascular tortuosity can be a potential indicator of relevant vascular and non-vascular diseases. However, the lack of a precise and standard guide for the tortuosity evaluation hinders its use for diagnostic and treatment purposes. This work aims to advance in the standardization of the retinal vascular tortuosity as a clinical biomarker with diagnostic potential, allowing, thereby, the validation of objective computational measurements on the basis of the entire spectrum of the expert knowledge. This paper describes a multi-expert validation process of the computational vascular tortuosity measurements of reference. A group of five experts, covering the different clinical profiles of an ophthalmological service, and a four-grade scale from non-tortuous to severe tortuosity as well as non-tortuous / tortuous and asymptomatic / symptomatic binary classifications are considered for the analysis of the the multi-expert validation procedure. The specialists rating process comprises two rounds involving all the experts and a joint round to establish consensual rates. The expert agreement is analyzed throughout the rating procedure and, then, the consensual rates are set as the reference to validate the prognostic performance of four computational tortuosity metrics of reference. The Kappa indexes for the intra-rater agreement analysis were obtained between 0.35 and 0.83 whereas for the inter-rater agreement in the asymptomatic / symptomatic classification were between 0.22 and 0.76. The Area Under the Curve (AUC) for each expert against the consensual rates were placed between 0.61 and 0.83 whereas the prognostic performance of the best objective tortuosity metric was 0.80. There is a high inter and intra-rater variability, especially for the case of the four grade scale. The prognostic performance of the tortuosity measurements is close to the experts’ performance, especially for Grisan measurement. However, there is a gap between the automatic effectiveness and the expert perception given the lack of clinical criteria in the computational measurements.

The retinal vascular tortuosity presents a valuable potential as a clinical biomarker of many relevant vascular and systemic diseases. Commonly, the existent approaches face the tortuosity quantification by means of fully mathematical representations of the vessel segments. However, the specialists, based on their diagnostic experience, commonly analyze additional domain-related information that is not represented in these mathematical metrics of reference. In this work, we propose a novel computational tortuosity metric that outperforms the mathematical metrics of reference also incorporating anatomical properties of the fundus image such as the distinction between arteries and veins, the distance to the optic disc, the distance to the fovea, and the vessel caliber. The evaluation of its prognostic performance shows that the integration of the anatomical factors provides an accurate tortuosity assessment that is more adjusted to the specialists’ perception.