Graphic STI
logo EPFL
Text EPFL
english only
Biomedical Imaging Group
Student Project: Alessandra Griffa
BIG > Teaching > Student Project > Alessandra Griffa

Time-course analysis of perfusion measurements in laser Doppler imaging

Alessandra GriffaMaster project
Politecnico di Torino, ItalyOctober 2007

 

CONTENTS

Abstract

The objective of this work is the study and the characterization of blood perfusion signals during post-occlusion reactive hyperemia tests, acquired by a laser Doppler imaging system. Laser Doppler imaging (LDI) is a non invasive method enabling the monitoring of microvascular blood flow. Even though LDI measurements are non-absolute, the enormous interest in microvascular perfusion has led to many clinical works and a number of studies related to the technique. Specifically, this technique finds applications in diabetology.

Studies on diabetic subjects suffering from microangiopathies have shown a decrease of post-occlusive reaction, particularly evident at the level of the foot and of the forearm. Therefore, arterial occlusion test results to be a useful instrument for the assessment of microcirculatory disease and the features of the post-occlusive hyperemic peak are actually recognized as an indicator of functional abnormalities. LDI perfusion signals relative to occlusion tests, from both healthy and diabetic subjects, are available thanks to an agreement with the H™pitaux Universitaire de Genève for a clinical test program.

We would like to achieve a quantitative description of the signals and assess a set of parameters that result representative of the signals and show a discriminating power between healthy and diabetic subjects. Then our idea is that of characterizing the perfusion signals by a linear model, which particularly should deal with post-occlusive hyperaemic peak variability, in terms of both height and temporal amplitude. We propose a five piecewise regressors linear model, which correlates with an electrical analogue. By the analysis of the available signals and exploiting the designed model, we identify a set of parameters representative of the signals and we propose them has good candidates for healthy and diabetic groups clustering. The model allows us evaluating parameters values with a minor dependency on the noise in the recordings.

Moreover, a short spatial analysis of perfusion maps is performed. We find substantial differences between the nail area and the finger area and we conclude that, with the objective of comparing post-occlusion signals from healthy and diabetic subjects, the investigated area should exclude the nail. Finally, a brief frequency analysis of perfusion signals is performed: the physiological fluctuating components (heartbeat, breathing, vasomotion and neurogenic activities) has been identified.

 


webmaster.big@epfl.ch • 11.12.2007