OpenCL framework for fast estimation of optical properties from spatial frequency domain images

Naglič, Peter; Zelinskyi, Yevhen; Likar, Boštjan; Pernuš, Franjo; Bürmen, Miran

doi:10.1117/12.2509986

Cited by 3 publications

(6 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“… 2 , 6 , 26 When run on simulated versions of these same phantom spectra, the model performed with an accuracy of

for

and

for

, on par with the accuracy that Naglič et al.’s model achieved over a narrower range of

values. 36 …”

Section: Resultsmentioning

confidence: 99%

“…demonstrates that sub-diffuse optical property heatmaps can be accurately rendered with ANNs using as few as five spatial frequencies: (0.0, 0.1, 0.2, 0.4, and

). 36 …”

Section: Discussionmentioning

confidence: 99%

“…The model was a multi-layer perceptron with two hidden layers and an output layer, 37 implemented with Keras and configured to work on OpenCL. 36 The model, which has several weights that must be trained by fitting to observed data, ultimately behaves as an empirically trained analytical equation which maps the

and

of a pixel to an optical property value. As such, it computes very quickly.…”

Section: Methodsmentioning

confidence: 99%

“…Please contact Dr. Naglič directly for inquiries to using this code. 36 Phantom data and simulation has been made publicly available through the Texas Data Repository at https://doi.org/10.18738/T8/BCKNMU , free to use with citation of this publication. Clinical data used in this work may be made available for access to specific parties through the Texas Data Repository under a Material Transfer Agreement at https://doi.org/10.18738/T8/BCKNMU .…”

Section: Code Data and Materials Availabilitymentioning

confidence: 99%

“…demonstrated a deep learning method for rendering sub-diffuse optical property heatmaps from sd-SFDI images that is capable of real-time speeds. 36 , 37 However, while this method was shown to be successful for simulated data, this method has never been calibrated for and applied to physical experimental data. Moreover, Naglič et al.’s model was trained on data simulated using the Gegenbauer kernel (GK) phase function approximation, whose domain 38 also does not cover the full range of optical properties seen in biological tissue.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Imaging sub-diffuse optical properties of cancerous and normal skin tissue using machine learning-aided spatial frequency domain imaging

et al. 2021

View full text Add to dashboard Cite

. Significance: Sub-diffuse optical properties may serve as useful cancer biomarkers, and wide-field heatmaps of these properties could aid physicians in identifying cancerous tissue. Sub-diffuse spatial frequency domain imaging (sd-SFDI) can reveal such wide-field maps, but the current time cost of experimentally validated methods for rendering these heatmaps precludes this technology from potential real-time applications. Aim : Our study renders heatmaps of sub-diffuse optical properties from experimental sd-SFDI images in real time and reports these properties for cancerous and normal skin tissue subtypes. Approach: A phase function sampling method was used to simulate sd-SFDI spectra over a wide range of optical properties. A machine learning model trained on these simulations and tested on tissue phantoms was used to render sub-diffuse optical property heatmaps from sd-SFDI images of cancerous and normal skin tissue. Results: The model accurately rendered heatmaps from experimental sd-SFDI images in real time. In addition, heatmaps of a small number of tissue samples are presented to inform hypotheses on sub-diffuse optical property differences across skin tissue subtypes. Conclusion: These results bring the overall process of sd-SFDI a fundamental step closer to real-time speeds and set a foundation for future real-time medical applications of sd-SFDI such as image guided surgery.

show abstract

“… 2 , 6 , 26 When run on simulated versions of these same phantom spectra, the model performed with an accuracy of

for

and

for

, on par with the accuracy that Naglič et al.’s model achieved over a narrower range of

values. 36 …”

Section: Resultsmentioning

confidence: 99%

“…demonstrates that sub-diffuse optical property heatmaps can be accurately rendered with ANNs using as few as five spatial frequencies: (0.0, 0.1, 0.2, 0.4, and

). 36 …”

Section: Discussionmentioning

confidence: 99%

and

of a pixel to an optical property value. As such, it computes very quickly.…”

Section: Methodsmentioning

confidence: 99%

Section: Code Data and Materials Availabilitymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Imaging sub-diffuse optical properties of cancerous and normal skin tissue using machine learning-aided spatial frequency domain imaging

et al. 2021

View full text Add to dashboard Cite

show abstract

A Machine Learning Approach to Determining Sub-Diffuse Optical Properties

Stier

Goth

Zhang

et al. 2020

Biophotonics Congress: Biomedical Optics 2020 (Translational, Microscopy, OCT, OTS, BRAIN)

View full text Add to dashboard Cite

show abstract

Generic and Model-Based Calibration Method for Spatial Frequency Domain Imaging with Parameterized Frequency and Intensity Correction

Lohner,

Nothelfer,

Kienle

2023

Sensors

View full text Add to dashboard Cite

Spatial frequency domain imaging (SFDI) is well established in biology and medicine for non-contact, wide-field imaging of optical properties and 3D topography. Especially for turbid media with displaced, tilted or irregularly shaped surfaces, the reliable quantitative measurement of diffuse reflectance requires efficient calibration and correction methods. In this work, we present the implementation of a generic and hardware independent calibration routine for SFDI setups based on the so-called pinhole camera model for both projection and detection. Using a two-step geometric and intensity calibration, we obtain an imaging model that efficiently and accurately determines 3D topography and diffuse reflectance for subsequently measured samples, taking into account their relative distance and orientation to the camera and projector, as well as the distortions of the optical system. Derived correction procedures for position- and orientation-dependent changes in spatial frequency and intensity allow the determination of the effective scattering coefficient μs′ and the absorption coefficient μa when measuring a spherical optical phantom at three different measurement positions and at nine wavelengths with an average error of 5% and 12%, respectively. Model-based calibration allows the characterization of the imaging properties of the entire SFDI system without prior knowledge, enabling the future development of a digital twin for synthetic data generation or more robust evaluation methods.

show abstract

OpenCL framework for fast estimation of optical properties from spatial frequency domain images

Cited by 3 publications

References 13 publications

Imaging sub-diffuse optical properties of cancerous and normal skin tissue using machine learning-aided spatial frequency domain imaging

Imaging sub-diffuse optical properties of cancerous and normal skin tissue using machine learning-aided spatial frequency domain imaging

A Machine Learning Approach to Determining Sub-Diffuse Optical Properties

Generic and Model-Based Calibration Method for Spatial Frequency Domain Imaging with Parameterized Frequency and Intensity Correction

Contact Info

Product

Resources

About