Biological signals are stochastic in nature and finding anomalies in them is not straightforward. In this undergraduate signal processing class, we worked with real data to detect when "bubble" treatment occurred using mouse heart beat data and wavelet decomposition.
Myocardial hypertrophy is a common heart muscle disease that occurs in 1 out of 500 people. Individuals with myocardial hypertrophy have enlarged heart muscles due to uncontrolled growth of the myocardial tissues. Currently, this hypertrophy is treated with invasive methods such as surgery and catheter for injecting alcohol into enlarged muscle. It has been proposed that we could treat myocardial hypertrophy with a non-invasive procedure that involves acoustic cavitation. This non-invasive procedure involves injecting the blood with small air bubbles, and collapsing these bubbles at a known site and time. A cavitation event is when one of these small air bubbles collapses and thereby launches a tiny but powerful water jet to an adjacent heart cell.This could help with decreasing the size of the enlarged muscle by eliminating unwanted tissues. This cavitation event outputs an abnormal ECG readout. Our task is to identify whether this treatment event occurred in mice by looking at their ECG waveform patterns. Detecting a cavitation event in heart response is crucial in identifying whether the treatment is currently affecting cells or not.