Any comprehensive archival review of the literature on this subject results in abundant citations to this problem and references to this problem go back to the early 1970s when computer-based signal processing tools and techniques first gained some acceptance in this area. However, a more focused review of the literature, dating from only 1995, showed the first use of wavelet transform techniques as applied to this problem. While the use of wavelet transform has improved the level of success-and continues to show further promise-to date, the medical community has not accepted any currently available computer-based detection algorithm for VT.

Ventricular tachycardia is an extremely unstable cardiac rhythm. Ventricular tachycardia is important because of its unpredictability and potential to cause sudden death. This arrhythmia originates in the ventricles of the human heart and results in a rapid heartbeat. It is identified clinically when a patient's electrocardiograph (ECG) exhibits three or more premature ventricular contractions (PVCs) in direct succession and the ventricular pumping rate exceeds 100 beats/minute. Because the duration of diastole (the relaxation phase of the heart-pumping cycle) is reduced at high heart rates, patients with VT exhibit a markedly diminished ventricle filling time with a corresponding reduction in filling volume-blood available for cardiac pumping. As a result of this drop in cardiac output-a net reduction in cardiac mechanical efficiency-the patient's physical condition can quickly deteriorate. Furthermore, at these accelerated heart rates (100 to 200 beats / minute) patients with VT often require emergency medical treatment since their cardiac output is almost always critically low. This arrhythmia often comes before ventricular fibrillation and sudden cardiac death and it is especially dangerous for patients who are not hospitalized at the time of its onset. The recommended treatment for patients diagnosed with sustainable VT is an implanted defibrillator.

Currently the most reliable method of determining whether ventricular tachycardia is both inducible and sustainable in a post-MI patient is to use an invasive procedure that must be undertaken by medical personnel in an operating room setting or Electro- physiology (EP) laboratory. Briefly, this medical procedure begins by inserting a catheter into one of the patient's major blood vessels leading to the heart. A pair of millimeter diameter wires is then inserted into the catheter and located on the heart. An electrical current sufficient to cause an arrhythmia is then applied to the heart in an attempt to initiate a sustained VT event. The results are measurable on a standard twelve-lead surface electrocardiograph (ECG) of the patient. Clearly this invasive procedure is uncomfortable to the patient and presents-although minor-some risk. This procedure is referred to as the 'gold standard'. Thus, one can see why a reliable noninvasive technique is constantly being sought to pre-test for the presence of inducible ventricular tachycardia in post-MI patients.

It should be noted here that one is not able to detect the potential for this arrhythmia by simply looking at the surface ECG signal in the time domain. However, in recordings of the human ECG, late potentials-i.e. patients with ventricular tachycardia-are usually defined as small signals initially below 40 V with a frequency content above 40 Hz. They can occur during the whole QRS complex of the ECG, but usually are found in the terminal part of the QRS complex and in the beginning of the ST segment. Thus, the primary focus of this research was to develop a signal processing technique-centered on the use of noninvasive signals available from a patient's 12-lead surface electrocardiograph (ECG)-to detect for the presence of inducible ventricular tachycardia.

Our work required unprocessed or raw patient data from a standard 12-lead surface electrocardiograph (ECG) system to be divided into two groups. The first group was known to have ventricular tachycardia and was identified as inducible for VT. This first group comprised patients in which VT was inducible, i.e. an unwanted reentry circuit could be established in the damaged tissue of the heart and the resulting VT was sustainable. Cardiac reentry circuits are believed to be the root path by which ventricular tachycardia is inducible in both humans and animals. The second group of patients was recognized as non-inducible for VT and was identified as the control. This second group included patients for whom this arrhythmia could not be identified-i.e. patients in whom VT is non-inducible or non-sustainable. Data from both groups was then processed using standard signal-processing technique-signal averaging and filtering-decomposed with an orthogonal Daubechies wavelet transform, and the wavelet coefficients in the expected area analyzed.

Our results to date seem promising. We can now detect for the presence of VT and differentiate between groups of patients with and without this condition. However, our current data set is too small to present any statistical analysis of our results. Our next step is to increase the size of our data set and continue the work. To that end we are seeking government funding under a new initiative at Wentworth Institute of Technology called the WEAR Program. WEAR-a summer research program at the Institute-seeks to fund faculty and students interested in this type of applied research.