scholarly journals Learning hidden elasticity with deep neural networks

2021 ◽  
Vol 118 (31) ◽  
pp. e2102721118
Author(s):  
Chun-Teh Chen ◽  
Grace X. Gu
Keyword(s):  
Inverse Problem ◽  
Cancer Diagnosis ◽  
Deep Neural Networks ◽  
De Novo ◽  
Imaging Technique ◽  
Theory Of Elasticity ◽  
Learning Approach ◽  
Missing Measurements ◽  
Heterogeneous Objects ◽  
Ill Posed

Elastography is an imaging technique to reconstruct elasticity distributions of heterogeneous objects. Since cancerous tissues are stiffer than healthy ones, for decades, elastography has been applied to medical imaging for noninvasive cancer diagnosis. Although the conventional strain-based elastography has been deployed on ultrasound diagnostic-imaging devices, the results are prone to inaccuracies. Model-based elastography, which reconstructs elasticity distributions by solving an inverse problem in elasticity, may provide more accurate results but is often unreliable in practice due to the ill-posed nature of the inverse problem. We introduce ElastNet, a de novo elastography method combining the theory of elasticity with a deep-learning approach. With prior knowledge from the laws of physics, ElastNet can escape the performance ceiling imposed by labeled data. ElastNet uses backpropagation to learn the hidden elasticity of objects, resulting in rapid and accurate predictions. We show that ElastNet is robust when dealing with noisy or missing measurements. Moreover, it can learn probable elasticity distributions for areas even without measurements and generate elasticity images of arbitrary resolution. When both strain and elasticity distributions are given, the hidden physics in elasticity—the conditions for equilibrium—can be learned by ElastNet.

scholarly journals Harnessing deep neural networks to solve inverse problems in quantum dynamics: machine-learned predictions of time-dependent optimal control fields

2020 ◽  
Vol 22 (40) ◽  
pp. 22889-22899
Author(s):  
Xian Wang ◽  
Anshuman Kumar ◽  
Christian R. Shelton ◽  
Bryan M. Wong
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Inverse Problem ◽  
Quantum Dynamics ◽  
Deep Neural Networks ◽  
Cost Effective ◽  
Quantum Systems ◽  
Time Dependent ◽  
Learning Approach ◽  
Machine Learning Approach

Deep neural networks are a cost-effective machine-learning approach for solving the inverse problem of constructing electromagnetic fields that enable desired transitions in quantum systems.

scholarly journals Bayesian Reconstruction through Adaptive Image Notion

Proceedings ◽  
2019 ◽  
Vol 33 (1) ◽  
pp. 21
Author(s):  
Fabrizia Guglielmetti ◽  
Eric Villard ◽  
Ed Fomalont
Keyword(s):  
Inverse Problem ◽  
Image Plane ◽  
Source Distribution ◽  
Bayesian Probability ◽  
Model Parameters ◽  
Uncertain Information ◽  
Source Detection ◽  
Component Mixture ◽  
Ill Posed ◽  
Image Planes

A stable and unique solution to the ill-posed inverse problem in radio synthesis image analysis is sought employing Bayesian probability theory combined with a probabilistic two-component mixture model. The solution of the ill-posed inverse problem is given by inferring the values of model parameters defined to describe completely the physical system arised by the data. The analysed data are calibrated visibilities, Fourier transformed from the ( u , v ) to image planes. Adaptive splines are explored to model the cumbersome background model corrupted by the largely varying dirty beam in the image plane. The de-convolution process of the dirty image from the dirty beam is tackled in probability space. Probability maps in source detection at several resolution values quantify the acquired knowledge on the celestial source distribution from a given state of information. The information available are data constrains, prior knowledge and uncertain information. The novel algorithm has the aim to provide an alternative imaging task for the use of the Atacama Large Millimeter/Submillimeter Array (ALMA) in support of the widely used Common Astronomy Software Applications (CASA) enhancing the capabilities in source detection.

ON THE WAY TOWARDS INCREASING THE PREDICTIVE POWER OF THE NUCLEAR MEAN FIELD THEORIES: EVALUATION OF TWO-BODY MATRIX ELEMENTS

2012 ◽  
Vol 21 (05) ◽  
pp. 1250037
Author(s):  
HERVÉ MOLIQUE ◽  
JERZY DUDEK
Keyword(s):  
Inverse Problem ◽  
Predictive Power ◽  
Mean Field ◽  
Field Theories ◽  
Matrix Elements ◽  
The Matrix ◽  
Ill Posed ◽  
Technical Issues ◽  
Numerical Precision ◽  
Mean Field Theories

In this paper we collect a number of technical issues that arise when constructing the matrix representation of the most general nuclear mean field Hamiltonian within which "all terms allowed by general symmetries are considered not only in principle but also in practice". Such a general posing of the problem is necessary when investigating the predictive power of the mean field theories by means of the well-posed inverse problem. [J. Dudek et al., Int. J. Mod. Phys. E21 (2012) 1250053]. To our knowledge quite often ill-posed mean field inverse problems arise in practical realizations what makes reliable extrapolations into the unknown areas of nuclei impossible. The conceptual and technical issues related to the inverse problem have been discussed in the above-mentioned topic whereas here we focus on "how to calculate the matrix elements, fast and with high numerical precision when solving the inverse problem" [For space-limitation reasons we illustrate the principal techniques on the example of the central interactions].

scholarly journals Geometric potentials from deep learning improve prediction of CDR H3 loop structures

2020 ◽  
Vol 36 (Supplement_1) ◽  
pp. i268-i275 ◽  
Author(s):  
Jeffrey A Ruffolo ◽  
Carlos Guerra ◽  
Sai Pooja Mahajan ◽  
Jeremias Sulam ◽  
Jeffrey J Gray
Keyword(s):  
Deep Neural Networks ◽  
De Novo ◽  
Probability Distributions ◽  
Supplementary Information ◽  
Model Parameters ◽  
Loop Structure ◽  
Variable Loops ◽  
Chain Sequence ◽  
Antibody Structure ◽  
Complementarity Determining Region

Abstract Motivation Antibody structure is largely conserved, except for a complementarity-determining region featuring six variable loops. Five of these loops adopt canonical folds which can typically be predicted with existing methods, while the remaining loop (CDR H3) remains a challenge due to its highly diverse set of observed conformations. In recent years, deep neural networks have proven to be effective at capturing the complex patterns of protein structure. This work proposes DeepH3, a deep residual neural network that learns to predict inter-residue distances and orientations from antibody heavy and light chain sequence. The output of DeepH3 is a set of probability distributions over distances and orientation angles between pairs of residues. These distributions are converted to geometric potentials and used to discriminate between decoy structures produced by RosettaAntibody and predict new CDR H3 loop structures de novo. Results When evaluated on the Rosetta antibody benchmark dataset of 49 targets, DeepH3-predicted potentials identified better, same and worse structures [measured by root-mean-squared distance (RMSD) from the experimental CDR H3 loop structure] than the standard Rosetta energy function for 33, 6 and 10 targets, respectively, and improved the average RMSD of predictions by 32.1% (1.4 Å). Analysis of individual geometric potentials revealed that inter-residue orientations were more effective than inter-residue distances for discriminating near-native CDR H3 loops. When applied to de novo prediction of CDR H3 loop structures, DeepH3 achieves an average RMSD of 2.2 ± 1.1 Å on the Rosetta antibody benchmark. Availability and Implementation DeepH3 source code and pre-trained model parameters are freely available at https://github.com/Graylab/deepH3-distances-orientations. Supplementary information Supplementary data are available at Bioinformatics online.

scholarly journals Spending for Advanced Cancer Diagnoses: Comparing Recurrent Versus De Novo Stage IV Disease

2019 ◽  
Vol 15 (7) ◽  
pp. e616-e627 ◽  
Author(s):  
Michael J. Hassett ◽  
Matthew Banegas ◽  
Hajime Uno ◽  
Shicheng Weng ◽  
Angel M. Cronin ◽  
...  
Keyword(s):  
Advanced Cancer ◽  
Cancer Diagnosis ◽  
De Novo ◽  
Stage Iv ◽  
Absolute Difference ◽  
Medicare Data ◽  
Stage Iv Disease ◽  
De Novo Cancer ◽  
Payment Models ◽  
Diagnosis Age

PURPOSE: Spending for patients with advanced cancer is substantial. Past efforts to characterize this spending usually have not included patients with recurrence (who may differ from those with de novo stage IV disease) or described which services drive spending. METHODS: Using SEER-Medicare data from 2008 to 2013, we identified patients with breast, colorectal, and lung cancer with either de novo stage IV or recurrent advanced cancer. Mean spending/patient/month (2012 US dollars) was estimated from 12 months before to 11 months after diagnosis for all services and by the type of service. We describe the absolute difference in mean monthly spending for de novo versus recurrent patients, and we estimate differences after controlling for type of advanced cancer, year of diagnosis, age, sex, comorbidity, and other factors. RESULTS: We identified 54,982 patients with advanced cancer. Before diagnosis, mean monthly spending was higher for recurrent patients (absolute difference: breast, $1,412; colorectal, $3,002; lung, $2,805; all P < .001), whereas after the diagnosis, it was higher for de novo patients (absolute difference: breast, $2,443; colorectal, $4,844; lung, $2,356; all P < .001). Spending differences were driven by inpatient, physician, and hospice services. Across the 2-year period around the advanced cancer diagnosis, adjusted mean monthly spending was higher for de novo versus recurrent patients (spending ratio: breast, 2.39 [95% CI, 2.05 to 2.77]; colorectal, 2.64 [95% CI, 2.31 to 3.01]; lung, 1.46 [95% CI, 1.30 to 1.65]). CONCLUSION: Spending for de novo cancer was greater than spending for recurrent advanced cancer. Understanding the patterns and drivers of spending is necessary to design alternative payment models and to improve value.

Sign in / Sign up

Export Citation Format

Share Document