Visible-light spectroscopic optical coherence tomography: applications and technology
Carlos Cuartas-Vélez is a PhD student in the department Biomedical Photonic Imaging. Promotors are prof.dr.ir. N. Bosschaart and prof.dr.ir. W. Steenbergen from the faculty of Science & Technology.
Optical coherence tomography (OCT) is a non-invasive, medical imaging modality that creates three-dimensional images of tissue, using an interferometer with a broadband source. As the light source in OCT spans multiple wavelengths, spectral analysis of the OCT signal (termed spectroscopic OCT, sOCT) can be employed to retrieve wavelength-dependent properties of tissue. sOCT systems using visible-light sources (vis-sOCT) benefit from shorter wavelengths and higher spectral contrast between the main tissue chromophores than those in the infrared. Vis-sOCT is particularly sensitive to the absorption by hemoglobin since it is more pronounced in the visible wavelength range. Throughout this thesis, we explore multiple applications of vis-sOCT in different multidisciplinary fields, including i) the ex-vivo and in-vivo quantification of total hemoglobin concentrations [tHb] in flowing blood, ii) imaging neovascularization in organ-on-chip models of the retina and the formation of thrombi in a blood vessel-on-chip model, and iii) the Leidenfrost dynamics of solid carbon dioxide discs.
Chapter 1 offers a general introduction to vis-sOCT. In this chapter, we discuss the fundamental concepts behind vis-sOCT, from the interferometer to the final vis-sOCT signal. Relevant applications for vis-sOCT are discussed at the end of the chapter.
Given the high hemoglobin absorption for visible light, vis-sOCT presents a unique opportunity to parameterize blood-related properties. In Chapter 2, we demonstrate the potential of vis-sOCT to quantify [tHb] within the clinical range (6-23 g/dL). We develop a new method to retrieve [tHb] by numerical optimization of the optical density (OD), and present its implementation by analyzing data that was acquired under two different schemes. In both cases, the new approach significantly improves the precision and bias compared to existing models based on a Lambert-Beer fit. We perform experiments with whole blood flowing through a capillary with [tHb] concentrations in the clinical range. With the blood flowing through a capillary glass, we demonstrate that vis-sOCT achieves a precision similar to other non-invasive methods such as smartphone imagery, photoplethysmography, and photoacoustics.
In Chapter 3, we develop a method to quantify [tHb] in-vivo and apply and validate this methodology to data acquired from a group of 27 healthy volunteers. In this method, we extend the OD-based approach from Chapter 2 to data in-vivo in the posterior forearm. With the OD approach, we retrieve a map of the [tHb], termed OCT-hemogram. Quantitative analysis of the OCT-hemogram provides an estimation of the mean [tHb] for each volunteer. Furthermore, we demonstrate that the OCT-hemogram generates vascular maps similar to the information from OCT angiography. We show that our methodology is consistent for all volunteers and compare our [tHb] estimations with measurements from a commercial blood analyzer. From the volunteer group, we analyze the effect of several potential factors of influence, including gender, skin tone, and epidermal thickness.
Since OCT creates label-free, non-invasive, and real-time images, we introduce the application of vis-sOCT in the organ-on-chip field as a relevant read-out. In Chapter 4, we exploit vis-sOCT to analyze an organ-on-chip model of the retinal blood barrier mimicking neovascularization as in wet age-related macular degeneration (AMD). We utilize vis-sOCT to assess the ingrowing vascular structures in a collagen matrix and compare the results with data from AMD, as previously observed in the literature.
In Chapter 5, we follow the dynamics of thrombus formation in a microfluidic blood vessel-on-chip. We develop an approach to monitor thrombus formation by analyzing the correlation of the OCT signal during a blood perfusion assay. We show that forming thrombi exhibit a higher correlation compared to flowing blood. We localize the origin locations of thrombi in the vessel-on-chip and monitor its expansion over time. We analyze thrombi formation in terms of location, shape, and size, and validate our results with microscopy images of fibrin and platelet markers after the blood perfusion assay.
Finally, in Chapter 6, we apply vis-sOCT to measure the vapor layer dynamics of a levitating solid carbon dioxide pellet. We verify a theoretical model to predict the vapor layer thickness as a function of substrate temperature and time. We use a modified vis-sOCT system, a common path interferometer, to measure the vapor layer thickness for different substrate temperatures. We monitor experimentally the behavior of the vapor layer under the dry ice over time and perform multiple measurements to validate the model prediction of the vapor layer thickness, the pellet lifetime, the change in geometry, and the dependency on the substrate temperature.
In conclusion, we present several methods and applications for vis-sOCT. We leverage the unique advantages of vis-sOCT in terms of resolution and spectral contrast and show its versatility in medical, biological, and scientific areas.