| 2. |
- Ewerlöf, Maria, 1987-
(författare)
-
Multispectral imaging of hemoglobin oxygen saturation in skin microcirculation
- 2022
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- The ability to measure microcirculatory parameters such as hemoglobin oxygen saturation is important since it mirrors the microcirculatory state of the body. The microcirculation delivers oxygen and nutrients to the cells of the body and, if impaired, may be a sign of circulatory failure. Human skin microcirculation can be accessed non-invasively with bio-optical technologies, where skin acts as a diagnostic window. Diffuse reflectance spectroscopy (DRS) is a technique that access skin microcirculatory parameters, especially hemoglobin oxygen saturation. Basic systems are fiber optic probebased and measure in one point, often in firm contact with the skin. Multispectral diffuse reflectance imaging (MSI) enables spatially resolved DRS, imaging skin optical parameters from spectrally resolved backscattered intensities. Spectral information detected by MSI systems contain information on, e.g., hemoglobin oxygen saturation and optical properties of the tissue. Both spatial and temporal resolved information of hemoglobin oxygen saturation is beneficial for better diagnostics in most clinical applications, e.g., to monitor progression of wound healing processes, or other microcirculatory diseases reflected in hemoglobin spectral changes. Analysis of acquired MSI multispectral data cubes to access information on tissue parameters with high contrast to these variations can be performed in several ways using models and simulations. Time resolved continuous measurements that are spectrally and spatially resolved generate large amounts of data, requiring both storage space and fast analysis. Reducing the number of wavelengths is one way to limit the amount of data, if it does not reduce the quality of interpreted results. Therefore, in my work, I investigated theoretically how to reduce the number of wavelengths, and later implemented my findings using a snapshot MSI camera. Monte Carlo (MC) simulations were used to estimate hemoglobin oxygen saturation from captured MSI data. I also performed temporally resolved in vivo measurements on healthy test subjects during vascular occlusion provocations with a 16-channel snapshot MSI system. The acquired data were analyzed using two different methods: inverse MC and trained artificial neural networks (ANNs). For inverse MC, the acquired spectrum was iteratively compared to simulated spectra, where different optical properties were used for the simulation, trying to find the best fit. ANNs were trained to intensity data measured with the MSI system, using concurrently measured hemoglobin oxygen saturation values from a validated probe-based system as target data. The results and outcome of this thesis indicate good possibility to accurately estimate hemoglobin oxygen saturation with as few as four wavelengths. Estimated hemoglobin oxygen saturation values from analysis of in vivo measurements from the 16-channel snapshot MSI camera show high conformance to values measured by the validated probe-based system. Using the ANN-approach reduces time for analysis of a 512 × 270-pixel image to 0.056 s, compared to 1 h 58 min required by the inverse MC algorithm to analyze the same data. The method enables real-time analysis, and is, consequently, preferable in many clinical situations.
|
|
| 3. |
- Ewerlöf, Maria, 1987-, et al.
(författare)
-
Multispectral snapshot imaging of skin microcirculatory hemoglobin oxygen saturation using artificial neural networks trained on in vivo data
- 2022
-
Ingår i: Journal of Biomedical Optics. - Bellingham, WA, United States : SPIE - International Society for Optical Engineering. - 1083-3668 .- 1560-2281. ; 27:3
-
Tidskriftsartikel (refereegranskat)abstract
- Significance: Developing algorithms for estimating blood oxygenation from snapshot multispectral imaging (MSI) data is challenging due to the complexity of sensor characteristics and photon transport modeling in tissue. We circumvent this using a method where artificial neural networks (ANNs) are trained on in vivo MSI data with target values from a point-measuring reference method.Aim: To develop and evaluate a methodology where a snapshot filter mosaic camera is utilized for imaging skin hemoglobin oxygen saturation (SO2), using ANNs.Approach: MSI data were acquired during occlusion provocations. ANNs were trained to estimate SO2 with MSI data as input, targeting data from a validated probe-based reference system. Performance of ANNs with different properties and training data sets was compared.Results: The method enables spatially resolved estimation of skin tissue SO2. Results are comparable to those acquired using a Monte-Carlo-based approach when relevant training data are used.Conclusions: Training an ANN on in vivo MSI data covering a wide range of target values acquired during an occlusion protocol enable real-time estimation of SO2 maps. Data from the probe-based reference system can be used as target despite differences in sampling depth and measurement position.
|
|
