A Recent Study on Hardware Accelerated Monte Carlo Modeling of Light Propagation in Biological Tissues

Měsíček, Jakub; Krejcar, Ondřej; Selamat, Ali; Kuča, Kamil

doi:10.1007/978-3-319-42007-3_43

Cited by 2 publications

(2 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The highly parallelized and Graphics Processing Unit (GPU)-optimized software was used to simulate the photon propagation in tissue. Although the simulation results are very accurate when facilitating 5 × 10 8 photons [ 28 , 29 ], the stochastic simulation process is computationally costly. Note that currently MCX only uses a single value for the scattering anisotropy instead of two-dimensional parameter maps.…”

Section: Methodsmentioning

confidence: 99%

Efficient Photoacoustic Image Synthesis with Deep Learning

Rix

Dreher

Nölke

et al. 2023

Sensors

View full text Add to dashboard Cite

Photoacoustic imaging potentially allows for the real-time visualization of functional human tissue parameters such as oxygenation but is subject to a challenging underlying quantification problem. While in silico studies have revealed the great potential of deep learning (DL) methodology in solving this problem, the inherent lack of an efficient gold standard method for model training and validation remains a grand challenge. This work investigates whether DL can be leveraged to accurately and efficiently simulate photon propagation in biological tissue, enabling photoacoustic image synthesis. Our approach is based on estimating the initial pressure distribution of the photoacoustic waves from the underlying optical properties using a back-propagatable neural network trained on synthetic data. In proof-of-concept studies, we validated the performance of two complementary neural network architectures, namely a conventional U-Net-like model and a Fourier Neural Operator (FNO) network. Our in silico validation on multispectral human forearm images shows that DL methods can speed up image generation by a factor of 100 when compared to Monte Carlo simulations with 5×108 photons. While the FNO is slightly more accurate than the U-Net, when compared to Monte Carlo simulations performed with a reduced number of photons (5×106), both neural network architectures achieve equivalent accuracy. In contrast to Monte Carlo simulations, the proposed DL models can be used as inherently differentiable surrogate models in the photoacoustic image synthesis pipeline, allowing for back-propagation of the synthesis error and gradient-based optimization over the entire pipeline. Due to their efficiency, they have the potential to enable large-scale training data generation that can expedite the clinical application of photoacoustic imaging.

show abstract

Section: Methodsmentioning

confidence: 99%

Efficient Photoacoustic Image Synthesis with Deep Learning

Rix

Dreher

Nölke

et al. 2023

Sensors

View full text Add to dashboard Cite

show abstract

“…Deoxidized hemoglobin found in the veins has the tendency to absorb maximum radiation with a wavelength range between 7.6 × 10 −4 mm [41]. IR (Infra Red) radiation absorption increases with large veins as compared to tissues adjacent to it [42,43]. As such, the contrast between the object under investigation (large veins) and surrounding objects (tissue) is easily accomplished.…”

Section: Problem Definitionmentioning

confidence: 99%

Multi-Biometric System Based on Cutting-Edge Equipment for Experimental Contactless Verification

Kolda

Krejcar

Selamat

et al. 2019

Sensors

Self Cite

View full text Add to dashboard Cite

Biometric verification methods have gained significant popularity in recent times, which has brought about their extensive usage. In light of theoretical evidence surrounding the development of biometric verification, we proposed an experimental multi-biometric system for laboratory testing. First, the proposed system was designed such that it was able to identify and verify a user through the hand contour, and blood flow (blood stream) at the upper part of the hand. Next, we detailed the hard and software solutions for the system. A total of 40 subjects agreed to be a part of data generation team, which produced 280 hand images. The core of this paper lies in evaluating individual metrics, which are functions of frequency comparison of the double type faults with the EER (Equal Error Rate) values. The lowest value was measured for the case of the modified Hausdorff distance metric - Maximally Helicity Violating (MHV). Furthermore, for the verified biometric characteristics (Hamming distance and MHV), appropriate and suitable metrics have been proposed and experimented to optimize system precision. Thus, the EER value for the designed multi-biometric system in the context of this work was found to be 5%, which proves that metrics consolidation increases the precision of the multi-biometric system. Algorithms used for the proposed multi-biometric device shows that the individual metrics exhibit significant accuracy but perform better on consolidation, with a few shortcomings.

show abstract

A Recent Study on Hardware Accelerated Monte Carlo Modeling of Light Propagation in Biological Tissues

Cited by 2 publications

References 30 publications

Efficient Photoacoustic Image Synthesis with Deep Learning

Efficient Photoacoustic Image Synthesis with Deep Learning

Multi-Biometric System Based on Cutting-Edge Equipment for Experimental Contactless Verification

Contact Info

Product

Resources

About