| Summary: | People with diabetes suffer from a high risk of developing foot related diseases. It is therefore important to perform a blood pressure measurement on the toe to be able to diagnose and treat in time. Using laser Doppler flowmetry has been proven to be a useful technique for this purpose during a standard blood pressure measurement procedure using a cuff. The laser Doppler probe detects once the blood flow returns which can then be related to the pressure value. However, the algorithm currently used by the company for detection of return of blood flow is in need of improvements. This thesis aims to develop an improved algorithm, which is more robust against artifacts. Furthermore, a warning system for uncertainties in the detection will be developed and integrated with the new algorithm. To create the algorithm an investigation of the signals’ appearances was performed to obtain an understanding of what artifacts and characteristics the algorithm should be able to handle. First three different basic approaches were implemented and tested, namely model curve, threshold and pulsations. These algorithms were then combined into two different more complex algorithms. One of them consisted of the model curve and the pulsation algorithm, the second combined algorithm consisted of the threshold algorithm and the pulsation algorithm. From the result it was found that the second combined algorithm performed best. It had a high accuracy and a well-functioning warning system. However, the algorithm had problems to correctly detect the return of flow when it is characterised by a slow increase of the perfusion. The biggest contribution by this thesis is the newly developed warning system. A false detection can lead to a false diagnose to be given if the operator is not attentive. The warning system is therefore an important feature since it can prevent this from occurring.
|
|---|
