Neural network forward model and transfer learning calibration from Monte Carlo to diffuse reflectance spectroscopy

Lan, Qing; McClarren, Ryan G.; Vishwanath, Karthik

doi:10.1117/12.2581556

Cited by 5 publications

(9 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Mean squared error (MSE) was used to optimize the weights in the neural network. Earlier study comparing the time efficiency of the NNFM to Monte Carlo was showed in our paper [26].…”

Section: Forward Modelmentioning

confidence: 80%

“…Once trained, unlike the traditional MCLUT that requires a MC simulation for each point of the grid, the NNFM is able to make predictions for grid points fast, significantly saving time and computational cost. In our previous work [10], we have demonstrated the NN model is faster than MC simulations by an order of magnitude of 10 2 − 10 4 . The NNFM is able to generate tables with higher speed and more flexibility as making predictions for any grid takes almost the same computational cost, giving a potential faster tool for clinical diagnosis.…”

Section: Discussionmentioning

confidence: 92%

“…where N is the number of instances in the data, y is true data and ŷ is the predicted value. The reason of using NMAE is because relative error can not be directly calculated due to presence of very small values in diffuse reflectance, especially at large SDDs [10].…”

Section: Performance Of Nnfmmentioning

confidence: 99%

“…With the advent of machine learning (ML), studies have been carried out to solve problems in biophotonics with ML models. Neural networks (NN), as one of the most powerful ML methods, have been employed to improve the forward modeling from geometry and optical parameters to diffuse optical spectra [9][10][11][12][13]. Further, NN-based models have been developed to estimate optical parameters from optical spectra or images [14][15][16][17][18][19][20].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Neural Network-based Inverse Model for Diffuse Reflectance Spectroscopy

Lan¹,

McClarren²,

Vishwanath³

2023

Preprint

View full text Add to dashboard Cite

In diffuse optical spectroscopy, the retrieval of the optical properties of a target requires the inversion of a measured reflectance spectrum. This is typically achieved through the use of forward models such as diffusion theory or Monte Carlo simulations, which are iteratively applied to optimize the reconstruction of the optical parameters. In this paper, we propose a novel neural network-based approach for solving this inverse problem, and establish its performance using experimentally measured diffuse reflectance data from a previously reported phantom study. Our inverse model was developed from a neural network forward model that was pre-trained with data from Monte Carlo simulations. The neural network forward model then creates a lookup table to invert the diffuse reflectance to the optical coefficients. We describe the construction of the neural network-based inverse model and test its ability to accurately retrieve optical properties from experimentally acquired diffuse reflectance data in liquid optical phantoms. Our results indicate that the developed neural network-based model achieves comparable accuracy to traditional Monte Carlo-based inverse models while offering improved speed and flexibility, potentially providing an alternative for developing faster clinical diagnosis tools. This study highlights the potential of neural networks in solving inverse problems in diffuse optical spectroscopy.

show abstract

“…Mean squared error (MSE) was used to optimize the weights in the neural network. Earlier study comparing the time efficiency of the NNFM to Monte Carlo was showed in our paper [26].…”

Section: Forward Modelmentioning

confidence: 80%

Section: Discussionmentioning

confidence: 92%

Section: Performance Of Nnfmmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Neural Network-based Inverse Model for Diffuse Reflectance Spectroscopy

Lan¹,

McClarren²,

Vishwanath³

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Monte Carlo simulations can be used to create such training data sets if accurate models are used for skin tissue, light propagation, and measuring system properties. [83,85] The system model must account for light source characteristics, spectral sensitivity, noise, and possible drift during measurements. Using measured spectra for training is another alternative if credible target data is available.…”

Section: Selecting Training Datamentioning

confidence: 99%

Multispectral imaging of hemoglobin oxygen saturation in skin microcirculation

Ewerlöf

2022

Linköping Studies in Science and Technology. Dissertations

View full text Add to dashboard Cite

show abstract

Transfer learning as a method to reproduce high-fidelity non-local thermodynamic equilibrium opacities in simulations

2023

Self Cite

View full text Add to dashboard Cite

Simulations of high-energy density physics often need non-local thermodynamic equilibrium opacity data. These data, however, are expensive to produce at relatively low fidelity. It is even more so at high fidelity such that the opacity calculations can contribute 95 % of the total computation time. This proportion can even reach large proportions. Neural networks can be used to replace the standard calculations of low-fidelity data, and the neural networks can be trained to reproduce artificial, high-fidelity opacity spectra. In this work, it is demonstrated that a novel neural network architecture trained to reproduce high-fidelity krypton spectra through transfer learning can be used in simulations. Further, it is demonstrated that this can be done while achieving a relative per cent error of the peak radiative temperature of the hohlraum of approximately 1 % to 4 % while achieving a 19.4 $\times$ speed up.

show abstract

Neural network forward model and transfer learning calibration from Monte Carlo to diffuse reflectance spectroscopy

Cited by 5 publications

References 0 publications

Neural Network-based Inverse Model for Diffuse Reflectance Spectroscopy

Neural Network-based Inverse Model for Diffuse Reflectance Spectroscopy

Multispectral imaging of hemoglobin oxygen saturation in skin microcirculation

Transfer learning as a method to reproduce high-fidelity non-local thermodynamic equilibrium opacities in simulations

Contact Info

Product

Resources

About