Monitoring Reservoirs from Subsea Wells: Ultra-Long Step out VSP Acquisition Using Enhanced Back-Scattering Fiber

While DAS VSP has become relatively standard in dry-tree applications, acquiring data in subsea wells has remained a technical challenge as umbilical can be tens of kilometers long, thereby reducing the overall quantity of backscattered light to the topside interrogator. This adds to the attenuation due to connectors at the wellhead and along the optical path. Yet, the need for subsea DAS interrogation is high, particularly with the onset of complex, deep-water projects that will require on-demand monitoring capabilities. In this article, we report on the successful acquisition and subsequent processing of a zero-offset VSP in an ultra-long step-out context. We simulated a subsea well with 69km worth of lead-in fiber to the wellhead, including attenuation at the wellhead mimicking the connectors. The attenuation was tackled by using an active, subsea amplifier (that would normally sit at the wellhead), and an in-house developed engineered fiber that provides a significant uplift in backscattered energy. We acquired this ZVSP both on fiber and with a standard wireline tool string for comparison. The approach presented here combines hardware and processing strategies to tackle the long step-out challenge. We demonstrate the ability to record seismic data even at very large step-out, a requirement for subsea well monitoring.

Circular-Linear Copulae for Animal Movement Data

1. Animal movement is often modeled in discrete time, formulated in terms of steps taken between successive locations at regular time intervals. Steps are characterized by the distance between successive locations (step-lengths) and changes in direction (turn angles). Animals commonly exhibit a mix of directed movements with large step lengths and turn angles near 0 when traveling between habitat patches and more wandering movements with small step lengths and uniform turn angles when foraging. Thus, step-lengths and turn angles will typically be cross-correlated. 2. Most models of animal movement assume that step-lengths and turn angles are independent, likely due to a lack of available alternatives. Here, we show how the method of copulae can be used to fit multivariate distributions that allow for correlated step lengths and turn angles. 3. We describe several newly developed copulae appropriate for modeling animal movement data and fit these distributions to data collected on fishers (Pekania pennanti). The copulae are able to capture the inherent correlation in the data and provide a better fit than a model that assumes independence. Further, we demonstrate via simulation that this correlation can impact movement patterns (e.g. rates of dispersion overtime). 4. We see many opportunities to extend this framework (e.g. to consider autocorrelation in step attributes) and to integrate it into existing frameworks for modeling animal movement and habitat selection. For example, copula could be used to more accurately sample available locations when conducting habitat-selection analyses.

Artificially engineered antiferromagnetic nanoprobes for ultra-sensitive histopathological level magnetic resonance imaging

Histopathological level imaging in a non-invasive manner is important for clinical diagnosis, which has been a tremendous challenge for current imaging modalities. Recent development of ultra-high-field (UHF) magnetic resonance imaging (MRI) represents a large step toward this goal. Nevertheless, there is a lack of proper contrast agents that can provide superior imaging sensitivity at UHF for disease detection, because conventional contrast agents generally induce T2 decaying effects that are too strong and thus limit the imaging performance. Herein, by rationally engineering the size, spin alignment, and magnetic moment of the nanoparticles, we develop an UHF MRI-tailored ultra-sensitive antiferromagnetic nanoparticle probe (AFNP), which possesses exceptionally small magnetisation to minimize T2 decaying effect. Under the applied magnetic field of 9 T with mice dedicated hardware, the nanoprobe exhibits the ultralow r2/r1 value (~1.93), enabling the sensitive detection of microscopic primary tumours (<0.60 mm) and micrometastases (down to 0.20 mm) in mice. The sensitivity and accuracy of AFNP-enhanced UHF MRI are comparable to those of the histopathological examination, enabling the development of non-invasive visualization of previously undetectable biological entities critical to medical diagnosis and therapy.

Imaging radar for automated driving functions

This work presents the implementation of a synthetic aperture radar (SAR) at 77 GHz, for automotive applications. This implementation is unique in the sense that it is a radar-only solution for most use-cases. The set-up consists of two radar sensors, one to calculate the ego trajectory and the second for SAR measurements. Thus the need for expensive GNSS-based dead reckoning systems, which are in any case not accurate enough to fulfill the requirements for SAR, is eliminated. The results presented here have been obtained from a SAR implementation which is able to deliver processed images in a matter of seconds from the point where the targets were measured. This has been accomplished using radar sensors which will be commercially available in the near future. Hence the results are easily reproducible since the deployed radars are not special research prototypes. The successful widespread use of SAR in the automotive industry will be a large step forward toward developing automated parking functions which will be far superior to today's systems based on ultrasound sensors and radar (short range) beam-forming algorithms. The same short-range radar can be used for SAR, and the ultrasound sensors can thus be completely omitted from the vehicle.

Step size self-adaptation for SGD

<p>Convergence and generalization are two crucial aspects of performance in neural networks. When analyzed separately, these properties may lead to contradictory results. Optimizing a convergence rate yields fast training, but does not guarantee the best generalization error. To avoid the conflict, recent studies suggest adopting a moderately large step size for optimizers, but the added value on the performance remains unclear. We propose the LIGHT function with the four configurations which regulate explicitly an improvement in convergence and generalization on testing. This contribution allows to: 1) improve both convergence and generalization of neural networks with no need to guarantee their stability; 2) build more reliable and explainable network architectures with no need for overparameterization. We refer to it as step size self-adaptation.</p>

Step size self-adaptation for SGD

<p>Convergence and generalization are two crucial aspects of performance in neural networks. When analyzed separately, these properties may lead to contradictory results. Optimizing a convergence rate yields fast training, but does not guarantee the best generalization error. To avoid the conflict, recent studies suggest adopting a moderately large step size for optimizers, but the added value on the performance remains unclear. We propose the LIGHT function with the four configurations which regulate explicitly an improvement in convergence and generalization on testing. This contribution allows to: 1) improve both convergence and generalization of neural networks with no need to guarantee their stability; 2) build more reliable and explainable network architectures with no need for overparameterization. We refer to it as step size self-adaptation.</p>

Deep learning image analysis for in vivo shortwave infrared imaging

In vivo optical imaging is a fast growing field that offers great perspectives for biomedical applications. In particular, imaging in the shortwave infrared window (SWIR: 1000-1700 nm) represents major improvement compared to the NIR-I region (700-900 nm) in terms of temporal and spatial resolutions in depths down to 4 mm. SWIR is a fast and cheap alternative to more precise methods such as X-ray and opto-acoustic imaging. Main obstacles in SWIR imaging are the noise and scattering from tissues and skin that reduce the precision of the method. We demonstrate the combination of SWIR in vivo imaging in the NIR-IIb region (1500-1700 nm) with advanced deep learning image analysis allows to overcome these obstacles and making a large step forward to high resolution imaging: it allows to precisely segment vessels from tissues and noise, provides morphological structure of the vessels network, with learned pseudo-3D shape, their relative position, dynamic information of blood vascularization in depth in small animals and distinguish the vessels types: artieries and veins. For demonstration we use neural the network IterNet that exploits structural redundancy of the blood vessels (L. Li, et.al., The IEEE WACV, 2020), which provides a useful analysis tool for raw SWIR images.