Stabilisation and analysis of haemoglobin video imaging sequences

Haemoglobin video imaging (HVI) is a technique for the examination of blood flow in microcirculations: the smallest and most delicate blood vessels in the body that bring blood cells and solutes into contact with the tissues they serve. Filtered light is used to image red blood cells (erythrocytes) flowing through the transparent conjunctiva. The method can resolve individual cells and is easily undertaken on the eyes of normal and diseased human subjects.

Conjunctiva is illuminated by light of a wavelength that is restricted to coincide with the long wavelength absorption peak of haemoglobin (approximately 520-600nm). This is reflected off the underlying sclera and imaged using a high-resolution monochrome digital video camera. Erythrocytes, which absorb the light, appear dark against a bright background and their flow through the microcirculation can be analysed.

The aim of analysis is to establish the metrics that characterise normal blood flow and then quantify changes that arise during different systemic and ocular diseases, and their treatments.

Clinical observations indicate that useful metrics fall into three main groups, based on:

  • Changes in the organisation of the microcirculation, and the diameters of its constituent vessels.
  • Variations in flow velocity (mean absolute velocity, pulse wave-form, periodicity of perfusion, flow reversal),
  • The interaction of cellular elements in flowing blood between each other and the vessel walls (erythrocyte aggregation, the size of aggregates and their resistance to re-modelling, interaction of cells with vessel walls, laminar flow and shear rate, gaps in the blood column arising from white cells and plasma).

The first requirement for such analyses is rapid, accurate stabilisation of video sequences. Microscopic observations on human subjects are always confounded by movement artefact. Furthermore, when a subject attempts to fix his/her gaze on a target, their eyes are not immobile, but make continual small excursions which are known as fixation microsaccades. Most of these are horizontal, but there is also a need to correct rotation and warping.

In order to compare results from different patients and different time-points in the same subject, it is first necessary to establish, for each vessel segment, its position in the hierarchy of the microcirculation. A simple mathematical model has been developed, describing the branching structure of the microcirculation; however, this depends on the automatic creation of an accurate map for each angiogram.

Measurement of the different parameters of blood flow in this well-defined sample microcirculation will provide a powerful tool for understanding normal blood flow, diagnosing microvascular diseases and quantifying changes resulting from pathology and its response to treatment. It will also permit the comparison of blood flow between groups of individuals, which is fundamental for pharmacological research.

For building a reliable and objective analysis pipeline for HVI the development of novel mathematical methods is needed that are able to tackle all the challenging factors in this dataset, and in particular will involve nonlinear image registration, motion estimation and image segmentation. Methodology development in this project could be nicely linked to projects P1 and P3, and will appeal to candidates with interests in applied and computational analysis and statistics.

Papers, Publications & Software