The development of a system that emulates percussion to detect the borders of the liver
Please cite this item using this persistent URLhttp://hdl.handle.net/10019.1/2331
This item appears in the following collection/s
Percussion is a centuries old bedside diagnostic technique that is used to diagnose various conditions of the thorax and abdomen, among these, abnormalities of the liver. The physician taps the patient’s skin in the area of interest to determine the qualities or presence of the underlying tissue or organ, by listening to the generated sound. The research contained in this thesis views percussion as a system identification method which uses an impulse response to identify the underlying system. A design employing an electromagnetic actuator as input pulse generator and accelerometer as impulse response recorder was motivated and built. Tests were performed on volunteers and the recorded signals were analysed to find methods of identifying the presence of the liver from these signals. The analyses matched signals to models or simply extracted signal features and matched these model parameters or signal features to the presence of the liver. Matching was done using statistical pattern recognition methods and the true presence of the liver was established using MR images. Features extracted from test data could not be matched to the presence of the liver with sufficient confidence which led to the conclusion that either the test, apparatus or analysis was flawed. The lack of success compelled a further test on a mock-up of the problem – a silicone model with an anomaly representing the organ under test. Results from these tests showed that signals should be measured further from the actuator and the approach followed during this test could lead to the successful location of the anomaly and discrimination between subtle differences in the consistency thereof. It is concluded that further research should aim to first validate percussion as performed by the physician and increase complexity in a phased manner, validating results and apparatus at each step. The approach followed was perhaps too bold in light of the lack of fundamental understanding of percussion and the underlying mechanisms.