Search

The increasing demand for unmanned aerial vehicles has spurred efforts to safely integrate the routine operation of UAVs and the need to detect potential mid-air collision threats. One of the key barriers with advancement in detecting aircraft is the lack of available data due to the challenging and expensive nature of the problem. In this paper, we present a new approach for vision based aircraft detection which uses deep learning for both the spatial and temporal behaviour of the potential collision threat. We train our network on synthetic image sequences of aircraft on collision course encounters. We then evaluate the performance of our proposed approach on real flight data, where we are able to successfully detect aircraft at ranges approaching those of human pilots.

Aims Acute kidney injury (AKI) during acute heart failure (AHF) is common and associated with increased morbidity and mortality. The underlying pathophysiological mechanism appears to have prognostic relevance; however, the differentiation of true, structural AKI from hemodynamic pseudo‐AKI remains a clinical challenge. Methods and results The Basics in Acute Shortness of Breath Evaluation Study (NCT01831115) prospectively enrolled adult patients presenting with AHF to the emergency department. Mortality of patients was prospectively assessed. Haemoconcentration, transglomerular pressure gradient (n = 231) and tubular injury patterns (n = 253) were evaluated to investigate pathophysiological mechanisms underlying AKI timing (existing at presentation vs. developing during in‐hospital period). Of 1643 AHF patients, 755 patients (46%) experienced an episode of AKI; 310 patients (19%; 41% of AKI patients) presented with community‐acquired AKI (CA‐AKI), 445 patients (27%; 59% of AKI patients) developed in‐hospital AKI. CA‐AKI but not in‐hospital AKI was associated with higher mortality compared with no‐AKI (adjusted hazard ratio 1.32 [95%‐CI 1.01–1.74]; P = 0.04). Independent of AKI timing, haemoconcentration was associated with a lower two‐year mortality. Transglomerular pressure gradient at presentation was significantly lower in CA‐AKI compared to in‐hospital AKI and no‐AKI (P < 0.01). Urinary NGAL ratio concentrations were significantly higher in CA‐AKI compared to in‐hospital AKI (P < 0.01) or no‐AKI (P < 0.01). Conclusions CA‐AKI but not in‐hospital AKI is associated with increased long‐term mortality and marked by decreased transglomerular pressure gradient and tubular injury, probably reflecting prolonged tubular ischemia due to reno‐venous congestion. Adequate decongestion, as assessed by haemoconcentration, is associated with lower long‐term mortality independent of AKI timing.