Hellerkeating5197

The representations are implicitly embedded as weights of the fc layer, such that the cosine distances can be computed efficiently via forward propagation. selleck compound In our experiments on brain MRI and abdominal CT datasets, the proposed framework achieves superior performances for low-shot segmentation towards standard DNN-based (3D U-Net) and classical registration-based (ANTs) methods, e.g., achieving mean Dice coefficients of 81%/69% for brain tissue/abdominal multi-organ segmentation using a single training sample, as compared to 52%/31% and 72%/35% by the U-Net and ANTs, respectively.We address the problem of semantic nighttime image segmentation and improve the state-of-the-art, by adapting daytime models to nighttime without using nighttime annotations. Moreover, we design a new evaluation framework to address the substantial uncertainty of semantics in nighttime images. Our central contributions are 1) a curriculum framework to gradually adapt semantic segmentation models from day to night through progressively darker times of day, exploiting cross-time-of-day correspondences between daytime images from a reference map and dark images to guide the label inference in the dark domains; 2) a novel uncertainty-aware annotation and evaluation framework and metric for semantic segmentation, including image regions beyond human recognition capability in the evaluation in a principled fashion; 3) the Dark Zurich dataset, comprising 2416 unlabeled nighttime and 2920 unlabeled twilight images with correspondences to their daytime counterparts plus a set of 201 nighttime images with fine pixel-level annotations created with our protocol, which serves as a first benchmark for our novel evaluation. Experiments show that our map-guided curriculum adaptation significantly outperforms state-of-the-art methods on nighttime sets both for standard metrics and our uncertainty-aware metric. Furthermore, our uncertainty-aware evaluation reveals that selective invalidation of predictions can improve results on data with ambiguous content such as our benchmark and profit safety-oriented applications involving invalid inputs.Objective measures of image quality generally operate by making local comparisons of pixels of a "degraded" image to those of the original. Relative to human observers, these measures are overly sensitive to resampling of texture regions (e.g., replacing one patch of grass with another). Here we develop the first full-reference image quality model with explicit tolerance to texture resampling. Using a convolutional neural network, we construct an injective and differentiable function that transforms images to a multi-scale overcomplete representation. We empirically show that the spatial averages of the feature maps in this representation capture texture appearance, in that they provide a set of sufficient statistical constraints to synthesize a wide variety of texture patterns. We then describe an image quality method that combines correlation of these spatial averages ("texture similarity) with correlation of the feature maps ("structure similarity). The parameters of the proposed measure are jointly optimized to match human ratings of image quality, while minimizing the reported distances between subimages cropped from the same texture images. Experiments show that the optimized method explains human perceptual scores, both on conventional image quality databases and texture databases. The measure also offers competitive performance on texture classification and retrieval, and show the robustness to geometric transformations. Code is available at https//github.com/dingkeyan93/DISTS.

Cerebral edema characterized as an abnormal accumulation of interstitial fluid has not been treated effectively. We propose a novel edema treatment approach to drive edematous fluid out of the brain by direct current utilizing brain tissue's electroosmotic property.

A finite element (FE) head model is developed and employed to assess the feasibility of the approach. First, the capacity of the model for electric field prediction is validated against human experiments. Second, two electrode configurations (S and D-montage) are designed to evaluate the distribution of the electric field, electroosmotic flow (EOF), current density, and temperature across the brain under an applied direct current.

The S-montage is shown to induce an average EOF velocity of 7e-4 mm/s underneath the anode by a voltage of 15 V, and the D-montage induces a velocity of 9e-4 mm/s by a voltage of 5 V. Meanwhile, the brain temperature in both configurations is below 38 °C, which is within the safety range. Further, the magnitude of EOF is proportional to the electric field, and the EOF direction follows the current flow from anode to cathode. The EOF velocity in the white matter is significantly higher than that in the gray matter under the anode where the fluid is to be drawn out.

The proposed electroosmosis based approach allows alleviating brain edema within the critical time window by direct current.

The approach may be further developed as a new treatment solely or as a complement to existing conventional treatments of edema.

The approach may be further developed as a new treatment solely or as a complement to existing conventional treatments of edema.

We present a novel pipeline that consists of various algorithms for the estimation of the cardiac output (CO) during ventricular assist devices (VADs) support using a single pump inlet pressure (PIP) sensor as well as pump intrinsic signals.

A machine learning (ML) model was constructed for the prediction of the aortic valve opening status. When a closed aortic valve is detected, the estimated CO equals the estimated pump flow. Otherwise, the estimated CO equals the sum of the estimated pump flow and the aortic valve flow, estimated via a Kalman-filter approach. Both the pathophysiological conditions and the pump speed of an in-vitro test bench were adjusted in various combinations to evaluate the performance of the pipeline, as well as the individual estimators.

The ML model yielded a Matthews correlation coefficient of 0.771, a sensitivity of 0.913 and a specificity of 0.871. An overall CO root mean square error (RMSE) of 0.69 L/min was achieved. Replacing the pump flow and aortic pressure estimators with sensors would decrease the RMSE below 0.