More Like this

1. Estimating Noise Propagation of Neural Network Based Image Reconstruction Using Automated Differentiation

   Wang, X ; Ludwig, D. ; Rawson, M. ; Balan, R.V. ; Ernst, T. ( July 2022 , ISMRM)

   Image reconstructions involving neural networks (NNs) are generally non-iterative and computationally efficient. However, without analytical expression describing the reconstruction process, the computation of noise propagation becomes difficult. Automated differentiation allows rapid computation of derivatives without an analytical expression. In this work, the feasibility of computing noise propagation with automated differentiation was investigated. The noise propagation of image reconstruction by End-to-end variational-neural-network was estimated using automated differentiation and compared with Monte-Carlo simulation. The root-mean-square error (RMSE) map showed great agreement between automated differentiation and Monte-Carlo simulation over a wide range of SNRs. 

   more » « less
2. On the Effect of Imperfect Reference Signal Phase Recovery on Performance of PSK System Influenced by TWDP Fading

   https://doi.org/10.3390/e25091341  

   Djordjevic, Goran T. ; Milic, Dejan N. ; Vasic, Bata ; Makal, Jarosław ; Vasic, Bane ( July 2023 , Entropy)

   We examine the effects of imperfect phase estimation of a reference signal on the bit error rate and mutual information over a communication channel influenced by fading and thermal noise. The Two-Wave Diffuse-Power (TWDP) model is utilized for statistical characterization of propagation environment where there are two dominant line-of-sight components together with diffuse ones. We derive novel analytical expression of the Fourier series for probability density function arising from the composite received signal phase. Further, the expression for the bit error rate is presented and numerically evaluated. We develop efficient analytical, numerical and simulation methods for estimating the value of the error floor and identifying the range of acceptable signal-to-noise ratio (SNR) values in cases when the floor is present during the detection of multilevel phase-shift keying (PSK) signals. In addition, we use Monte Carlo simulations in order to evaluate the mutual information for modulation orders two, four and eight, and identify its dependence on receiver hardware imperfections under the given channel conditions. Our results expose direct correspondence between bit error rate and mutual information value on one side, and the parameters of TWDP channel, SNR and phase noise standard deviation on the other side. The results illustrate that the error floor values are strongly influenced by the phase noise when signals propagate over a TWDP channel. In addition, the phase noise considerably affects the mutual information.

    

   more » « less
3. Reconstructing Translucent Objects using Differentiable Rendering

   https://doi.org/10.1145/3528233.3530714  

   Deng, Xi ; Luan, Fujun ; Walter, Bruce ; Bala, Kavita ; Marschner, Steve ( July 2022 , SIGGRAPH '22: ACM SIGGRAPH 2022 Conference Proceedings)

   Inverse rendering is a powerful approach to modeling objects from photographs, and we extend previous techniques to handle translucent materials that exhibit subsurface scattering. Representing translucency using a heterogeneous bidirectional scattering-surface reflectance distribution function (BSSRDF), we extend the framework of path-space differentiable rendering to accommodate both surface and subsurface reflection. This introduces new types of paths requiring new methods for sampling moving discontinuities in material space that arise from visibility and moving geometry. We use this differentiable rendering method in an end-to-end approach that jointly recovers heterogeneous translucent materials (represented by a BSSRDF) and detailed geometry of an object (represented by a mesh) from a sparse set of measured 2D images in a coarse-to-fine framework incorporating Laplacian preconditioning for the geometry. To efficiently optimize our models in the presence of the Monte Carlo noise introduced by the BSSRDF integral, we introduce a dual-buffer method for evaluating the L2 image loss. This efficiently avoids potential bias in gradient estimation due to the correlation of estimates for image pixels and their derivatives and enables correct convergence of the optimizer even when using low sample counts in the renderer. We validate our derivatives by comparing against finite differences and demonstrate the effectiveness of our technique by comparing inverse-rendering performance with previous methods. We show superior reconstruction quality on a set of synthetic and real-world translucent objects as compared to previous methods that model only surface reflection. 

   more » « less
4. Learning Optical Map in Liquid Xenon Detector with Poisson Likelihood Loss

   Shixiao Liang and Christopher Tunnell ( December 2023 , NeurIPS Machine Learning in the Physical Sciences)

   Dual-phase liquid xenon time projection chambers (LXeTPC) have been successfully applied in rare event searches in astroparticle physics because of their ability to reach low backgrounds and detect small scintillation signals with photosensors. Accurate modeling of optical properties is essential for reconstructing particle interactions within these detectors as well as for developing data selection criteria. This is commonly achieved with discretized maps derived from Monte Carlo simulation or approximated with empirical analytical models. In this work, we employ a novel approach to this using a neural network trained with a Poisson log-likelihood ratio loss to model the mapping from light source location to the expected light intensity for each photosensor. We demonstrate its effectiveness by integrating it into a likelihood fitter for position reconstruction, simultaneously providing insights into the uncertainty associated with the reconstructed position. 

   more » « less
5. Investigation of the source‐detector separation in near infrared spectroscopy for healthy and clinical applications

   https://doi.org/10.1002/jbio.201900175  

   Wang, Lei ; Ayaz, Hasan ; Izzetoglu, Meltem ( November 2019 , Journal of Biophotonics)

   Understanding near infrared light propagation in tissue is vital for designing next generation optical brain imaging devices. Monte Carlo (MC) simulations provide a controlled mechanism to characterize and evaluate contributions of diverse near infrared spectroscopy (NIRS) sensor configurations and parameters. In this study, we developed a multilayer adult digital head model under both healthy and clinical settings and assessed light‐tissue interaction through MC simulations in terms of partial differential pathlength, mean total optical pathlength, diffuse reflectance, detector light intensity and spatial sensitivity profile of optical measurements. The model incorporated four layers: scalp, skull, cerebrospinal‐fluid and cerebral cortex with and without a customizable lesion for modeling hematoma of different sizes and depths. The effect of source‐detector separation (SDS) on optical measurements' sensitivity to brain tissue was investigated. Results from 1330 separate simulations [(4 lesion volumes × 4 lesion depths for clinical +3 healthy settings) × 7 SDS × 10 simulation = 1330)] each with 100 million photons indicated that selection of SDS is critical to acquire optimal measurements from the brain and recommended SDS to be 25 to 35 mm depending on the wavelengths to obtain optical monitoring of the adult brain function. The findings here can guide the design of future NIRS probes for functional neuroimaging and clinical diagnostic systems.

    

   more » « less