Model based analysis of multimodal data sets for quantitative molecular and microcirculatory imaging

A research project supported by CENIIT and Perimed AB

Project manager: Marcus Larsson, Assistant Professor at the Department of Biomedical Engineering

Motivation and Aim

The well-being of all biological organisms relies on a sufficient supply of nutrition and oxygen. In the human tissue, the most minute vessels of the microcirculatory network fulfill this demand. Consequently, there is a clinical demand for objective methods that rapidly and accurately characterizes the microcirculatory perfusion, blood oxygenation and the mitochondrial oxygen uptake.

The overall aim for this project is to devise new algorithms for quantitative determination of the tissue chromophores and the microcirculatory blood flow using hyper-spectral imaging and laser speckle imaging. To compensate for tissue scattering, these techniques will be combined using an analysis algorithm that is based on extensive computer simulations that accurately predict how photons propagate thru tissue. This includes computational demanding numerical Monte Carlo simulations, multi-dimensional optimization algorithms and parallel processing using graphics processing units for a massive computational speed-up.

Background

Laser Speckle Imaging (LSI) constitutes a number of non-invasive techniques that produce images reflecting the microcirculatory activity. The principle for these techniques is to analyze the speckle pattern formed by photons backscattered from a laser illuminated tissue (fig 1). Spatial and temporal variations in this pattern can be linked to photons that have been Doppler shifted when scattered by moving red blood cells (RBC´s). To quantify the RBC flow velocity, two different analysis approaches can be used: speckle contrast analysis (spatial variations, fig2), and frequency content analysis (temporal variations). In common for all implementations using either approach is that they do not take into account how photons propagate thru tissue. This implies that a change in tissue optical properties (i.e. scattering and absorption), affecting the photon path length, will result in a changed LSI measure. Consequently, it has not yet been possible to use LSI for absolute blood flow quantification, as the tissue optical properties (i.e. absorption and scattering) vary between and within individuals. The speckle contrast analysis of LSI images, from here on refereed to as Laser Speckle Contrast Analysis (LASCA), utilizes speckle images acquired during relatively long integration times (up to about 10-20 ms). Almost all previously suggested LASCA algorithms are based on singles exposure images, where the blurring of the speckle pattern is quantified by calculating the local contrast over small sub-frames (typically 5x5 or 7x7 pixels). This approach, where only a single contrast value is used for characterizing the blood flow, is inadequate, as variations in the true blood concentration and velocity can not be separately determined. A first step towards a refined LASCA algorithm, using a series of multi-exposure images, have recently been presented. Still, LASCA is not capable of producing accurate absolute blood flow measures. This project, therefore, propose a new approach where multi-exposure LASCA is combined with theoretical light scattering models that links the speckle contrast to the actual concentration and velocity of moving RBC, using Monte Carlo simulated Doppler distributions.

Fig 1. Unprocessed speckel image captured from a laser illuminated finger during an increased blood flow (post-occlusive hyperemic reaction).

Fig 2. LSI processed image of a finger displaying an increased blood flowed (post-occlusive hyperemic reaction).

This approach requires additional data on the tissue optical properties. Therefore, the LSI is integrated with a hyper spectral imager (HSI). A similar probe-based technique have proven successful in estimating the tissue optical properties. The HSI system does not only provide means for quantitative perfusion images. It also enables the estimation of tissue chromophores (i.e. absorbing compounds) that can reveal the status of the tissue. We have recently demonstrated that a white light spectroscopic instrument has the potential of estimating not only the amount and oxygenation of blood but also the oxidation of cytochrome aa3, a chromophore that reflects the mitochondrial oxygen uptake and the production of ATP.

Impact and Industrial Motive

LASCA have struggled to become a clinically accepted technique for evaluating the tissue microcirculation. A factor that strongly contributes to this is the lack of an absolute perfusion measure where velocity and concentration can be separately studied. This project description outlines a possible route to overcome this. In addition, the proposed HSI system will be capable of maping the spatial distribution of blood and its oxygenation. Clinically, such a combined system could serve as a useful tool for: Diabetic micro-angiopathy assessment and treatment evaluation; Wound healing assessment; Detection of hypoxic tissue and ischemia.

The collaborating company, Perimed AB, will greatly benefit from this project as they currently produce a laser speckle imager that is based on a primitive algorithm with several limitations and deficiencies. The addition of an HSI instrument would also widen their current line of products.

Project vision

The long-term vision of this project is to demonstrate a fully functional imager that is capable of presenting real-time video sequences that reflects the microcirculatory activity. This Imager will be capable of measuring the amount of blood, the blood oxygenation and the speed of the blood flow. The applicability of such a devise will be evaluated in clinical studies. If successful, our collaborating company, Perimed AB, will present a new product that is based on our results.

Research Environment and Collaboration

The research is carried out at the Department of Biomedical Engineering (IMT), Linköping University. This environment allows for support from technicians and other researchers within the field of biomedical optics. It also facilitates the collaboration with medical researchers located at the University Hospital.

Industrial partner

Perimed AB, established in 1981 and headquartered just outside Stockholm, is the world leader in developing, manufacturing and marketing state-of-the-art equipment for microvascular diagnosis. With customers in more than 80 countries, Perimed also actively participates in a wide range of research projects together with leading universities to deepen their understanding of diseases related to blood perfusion and microvascular functions (http://www.perimed-instruments.com).

LiU contacts

• Marcus Larsson, Assistant professor
• Hanna Karlsson, PhD student

Previous related pulications

1. H. Karlsson, A. Petterson, M. Larsson, and T. Stromberg. Can a one-layer optical skin model including melanin and inhomogeneously distributed blood explain spatially resolved diffuse reflectance spectra. in Optical Tomography and Spectroscopy of Tissue IX. 2011: SPIE.
2. T. Lindbergh, M. Larsson, Z. Szabo, H. Casimir-Ahn, and T. Strömberg, "Intramyocardial oxygen transport by quantitative diffuse reflectance spectroscopy in calves." J Biomed Opt, vol 15(2), pp. 027009, 2010.
3. T. Lindbergh, E. Häggblad, H. Ahn, G.E. Salerud, M. Larsson, and T. Strömberg, "Improved model for myocardial diffuse reflectance spectra by including mitochondrial cytochrome aa3, methemoglobin, and inhomogenously distributed RBC." J Biophotonics, 2010.
4. I. Fredriksson, M. Larsson, and T. Stromberg, "Model-based quantitative laser Doppler flowmetry in skin." Journal of Biomedical Optics, vol 15(5), pp. 057002, 2010.
5. M.J. Draijer, E. Hondebrink, M. Larsson, T.G. van Leeuwen, and W. Steenbergen, "Relation between the contrast in time integrated dynamic speckle patterns and the power spectral density of their temporal intensity fluctuations." Optics Express, vol 18(21), pp. 21883-21891, 2010.
6. I. Fredriksson, M. Larsson, and T. Stromberg, "Forced detection Monte Carlo algorithms for accelerated blood vessel image simulations." J Biophotonics, vol 2(3), pp. 178-84, 2009.
7. I. Fredriksson, M. Larsson, and T. Stromberg, "Optical microcirculatory skin model: assessed by Monte Carlo simulations paired with in vivo laser Doppler flowmetry." Journal of Biomedical Optics, vol 13(1), pp. 014015, 2008.