Abstract. We present an algorithm to synthetically increase the resolution of a solitary depth image using only a generic database of local patches. Modern range sensors measure depths with non-Gaussian noise and at lower starting resolutions than typical visible-light cameras. While patch based approaches for upsampling intensity images continue to improve, this is the first exploration of patching for depth images. We match against the height field of each low resolution input depth patch, and search our database for a list of appropriate high resolution candidate patches. Selecting the right candidate at each location in the depth image is then posed as a Markov random field labeling problem. Our experiments also show how important further depth-specific processing, such as noise removal and correct patch normalization, dramatically improves our results. Perhaps surprisingly, even better results are achieved on a variety of real test scenes by providing our algorithm with only synthetic training depth data.

The Deep Noise Suppression (DNS) challenge is designed to foster innovation in the area of noise suppression to achieve superior perceptual speech quality. We recently organized a DNS challenge special session at INTERSPEECH and ICASSP 2020. We opensourced training and test datasets for the wideband scenario. We also open-sourced a subjective evaluation framework based on ITU-T standard P.808, which was also used to evaluate participants of the challenge. Many researchers from academia and industry made significant contributions to push the field forward, yet even the best noise suppressor was far from achieving superior speech quality in challenging scenarios. In this version of the challenge organized at INTERSPEECH 2021, we are expanding both our training and test datasets to accommodate full band scenarios. The two tracks in this challenge will focus on real-time denoising for (i) wide band, and (ii) full band scenarios. We are also making available a reliable nonintrusive objective speech quality metric for wide band called DNS-MOS for the participants to use during their development phase.

Single plane wave transmissions are promising for automated imaging tasks requiring high ultrasound frame rates over an extended field of view. However, a single plane wave insonification typically produces suboptimal image quality. To address this limitation, we are exploring the use of deep neural networks (DNNs) as an alternative to delay-and-sum beamforming. The objectives of this work are to obtain information directly from raw channel data and to simultaneously generate both a segmentation map for automated ultrasound tasks and a corresponding ultrasound B-mode image for interpretable supervision of the automation. We focus on visualizing and segmenting anechoic targets surrounded by tissue and ignoring or de-emphasizing less important surrounding structures. DNNs trained with Field II simulations were tested with simulated, experimental phantom, and in vivo datasets that were not included during training. With unfocused input channel data (i.e., prior to the application of receive time delays), simulated, experimental phantom, and in vivo test datasets achieved mean ± standard deviation Dice similarity coefficients of 0.92 ± 0.13, 0.92±0.03, and 0.77±0.07, respectively, and generalized contrastto-noise ratios (gCNR) of 0.95±0.08, 0.93±0.08, and 0.75±0.14, respectively. With subaperture beamformed channel data and a modification to the input layer of the DNN architecture to accept these data, the fidelity of image reconstruction increased (e.g., mean gCNR of multiple acquisitions of two in vivo breast cysts ranged 0.89-0.96), but DNN display frame rates were reduced from 395 Hz to 287 Hz. Overall, the DNNs successfully translated feature representations learned from simulated data to phantom and in vivo data, which is promising for this novel approach to simultaneous ultrasound image formation and segmentation.

No abstract

Short-lag spatial coherence (SLSC) imaging displays the spatial coherence between backscattered ultrasound echoes instead of their signal amplitudes and is more robust to noise and clutter artifacts when compared with traditional delay-and-sum (DAS) B-mode imaging. However, SLSC imaging does not consider the content of images formed with different lags, and thus does not exploit the differences in tissue texture at each short-lag value. Our proposed method improves SLSC imaging by weighting the addition of lag values (i.e., M-weighting) and by applying robust principal component analysis (RPCA) to search for a low-dimensional subspace for projecting coherence images created with different lag values. The RPCA-based projections are considered to be denoised versions of the originals that are then weighted and added across lags to yield a final robust SLSC (R-SLSC) image. Our approach was tested on simulation, phantom, and in vivo liver data. Relative to DAS B-mode images, the mean contrast, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) improvements with R-SLSC images are 21.22 dB, 2.54, and 2.36, respectively, when averaged over simulated, phantom, and in vivo data and over all lags considered, which corresponds to mean improvements of 96.4%, 121.2%, and 120.5%, respectively. When compared with SLSC images, the corresponding mean improvements with R-SLSC images were 7.38 dB, 1.52, and 1.30, respectively (i.e., mean improvements of 14.5%, 50.5%, and 43.2%, respectively). Results show great promise for smoothing out the tissue texture of SLSC images and enhancing anechoic or hypoechoic target visibility at higher lag values, which could be useful in clinical tasks such as breast cyst visualization, liver vessel tracking, and obese patient imaging.