Chih-ming James Chiang

Digital Signal Processing Chip Implementation for Detection and Analysis of Intracardiac Electrograms

The adoption of digital signal processing (DSP) microchips for detection and analysis of electrocardiographic signals offers a means for increased computational speed and the opportunity for design of customized architecture to address real‐time requirements. A system using the Motorola 56001 DSP chip has been designed to realize cycle‐by‐cycle detection (triggering) and waveform analysis using a time‐domain template matching technique, correlation waveform analysis (CWA). The system digitally samples an electrocardiographic signal at 1000 Hz, incorporates an adaptive trigger for detection of cardiac events, and classifies each waveform as normal or abnormal. Ten paired sets of single‐chamber bipolar intracardiac electrograms (1–500 Hz) were processed with each pair containing a sinus rhythm (SR) passage and a corresponding arrhythmia segment from the same patient. Four of ten paired sets contained intraatrial electrograms that exhibited retrograde atrial conduction during ventricular pacing; the remaining six paired sets of intraventricular electrograms consisted of either ventricular tachycardia (4) or paced ventricular rhythm (2). Of 2,978 depolarizations in the test set, the adaptive trigger failed to detect 6 (99.8% detection sensitivity) and had 11 false triggers (99.6% specificity). Using patient dependent thresholds for CWA to classify waveforms, the program correctly identified 1,175 of 1,197 (98.2% specificity) sinus rhythm depolarizations and 1,771 of 1.781 (99.4% sensitivity) abnormal depolarizations. From the results, the algorithm appears to hold potential for applications such as realtime monitoring of electrophysiology studies or detection and classification of tachycardias in implantable antitachycardia devices.

Pattern recognition of cardiac arrhythmias using two intracardiac channels

New techniques of arrhythmia detection using morphology of the waveform have shown promise in correctly detecting fatal arrhythmias. Morphologic analysis of two channels, atrial and ventricular, has the potential of improving the ability of the algorithm to distinguish arrhythmias. This study uses correlation waveform analysis of intracardiac electrograms from the right ventricular apex and the high right atrium to define a two dimensional feature space with linear decision boundaries created from a training set by a least squares minimum distance classifier. The total set represented 16 patients. This method correctly discriminated 48/48 cycles of sinus rhythm, 47/48 of ventricular tachycardia (VT), 39/48 of atrial tachycardia and 40/48 of VT with retrograde, and supraventricular tachycardia with aberrant conduction. Performing this technique on two intracardiac electrograms appears to reliably differentiate various arrhythmias and warrants further study.<<ETX>>

The Value of Rate Regularity and Multiplicity Measures to Detect Ventricular Tachycardia in the Presence of Atrial Fibrillation or Flutter

The predominant cause of inappropriate therapy by implantable antitachycardia devices with pacing and nonpacing cardioverter defibrillators, is mistaking a fast ventricular response during atrial fibrillation or flutter with true ventricular tachycardia (VT). The distinction between these arrhythmias is an important consideration in addressing the problem of reducing false‐positives in detection mechanisms for implantable devices. Dual chamber analysis that examines atrial and ventricular event ratios has been proposed as a solution to this problem, but would still fail in distinguishing paroxysmal VT requiring treatment from a fast but otherwise benign ventricular response during atrial fibrillation or flutter. In this study, two methods for discriminating these tachyarrhythmias were evaluated. Method 1 examined ventricular rate and rate regularity as a method for VT detection. Method 2 combined rate and regularity as well as an additional multiplicity criterion for recognition of atrial flutter with a fast ventricular response. In 20 patients. Method 1 had 100% sensitivity of VT detection and 80% specificity for detection of atrial fibrillation or flutter. Method 2 had 90% sensitivity and 90% specificity. These results suggest that use of these algorithms in future implantable devices would result in a decrease in false‐positive device therapies.

An innovative two-channel rate and morphology method for complex arrhythmia diagnosis in impiantatile devices

Impiantatile cardioverter-defìbrillators (ICDs) are highly effective in preventing sudden cardiac death. However, false-positive diagnoses cause inappropriate device shocks estimated to be as high as 40%. To address this problem, the Augmented Two-Channel Arrhythmia Detection [A2CAD] algorithm, a real-time scheme utilizing information from both the atrial and ventricular channels, is introduced. The algorithm uses rate detection as a first stage and augments this with morphological signal analysis in rhythms that confound the rate-only diagnoses. Results from 40 patient cases indicate 100% success rate for A2CAD, validating its potential for implementation in future implantable devices.

Median filtering as a sudden onset criterion to separate sinus tachycardia from ventricular tachycardia

Because the use of absolute rate alone to distinguish sinus tachycardia (ST) from ventricular tachycardia (VT) lacks precision, rate variation algorithms including sudden onset have been proposed to improve the detection of pathological tachycardias by implantable antitachycardia devices. In this study, a median filtering algorithm was designed, tested, and compared to these established algorithms using identical data. Onset of ST during exercise and during VT was analyzed for 50 cases each. Fixed-interval change and percent change of cardiac cycles were compared as criteria for sudden onset. Optimal performance of the five-cycle median filter was achieved using a threshold of 25% change in median ventricular rate. It successfully differentiated ST from paroxysmal VT with a sensitivity of 92% and specificity of 96%. Median filtering appears to yield better performance when compared to previously proposed onset algorithms for VT identification.<<ETX>>

Augmented Two‐Channel Arrhythmia Detection: An Efficient Diagnostic Method for Implantable Devices

ICDs are highly effective in preventing sudden cardiac death. However, inappropriate device shocks caused by false‐positive diagnoses are estimated to happen in 20% of all patients. The need for implantable electrical devices to detect with precision arrhythmias requiring therapy has spawned a variety of proposals for better means of tachycardia identification. To address this problem, the augmented two‐channel arrhythmia detection (A2CAD) algorithm, a real‐time scheme utilizing timing and morphology from both the atrial and ventricular channels, is introduced. The algorithm uses rate detection as a first stage and augments this with morphological signal analysis in rhythms that confound the rate only diagnoses. The software executes in real‐time (online), and has been tested on 60 passages of two‐channel intracardiac signals. The following arrhythmias constituted the test set: 10 AF and/or atrial flutter; 15 SVT; 16 VT; 10 ventricular flutter or VF; 5 sinus tachycardia; and 4 cases of AF concurrent with VF. Results from 60 patient cases indicate 57 (95%) of 60 success rate for A2CAD, validating its potential for implementation in future implantable devices.

Dual-chamber intracardiac arrhythmia analysis ☆

The need for implantable devices to classify potentially pace-terminable rhythms from those requiring immediate defibrillation has become a necessity in newer generation devices with tiered-therapy protocols. Reports of acceleration of slower tachycardias into faster, more lethal forms upon delivery of initial pacing therapy further emphasizes the need for more critical analysis of the originating tachycardia. Single-chamber analysis employing ventricular electrodes is standard, but this unfortunately ignores concomitant atria1 information. This remains the case despite gains in pacing technology over the past two decades, which has led, inevitably, to two-chamber pacemaker implementation. Our hypothesis is that two-channel analysis is requisite for accurate arrhythmia classification. Moreover, our recent studies have demonstrated that morphological evaluation of the signal can provide improved specificity over rate-based systems. We propose here the first comprehensive rate and morphological classification of two-channel (atria1 and

Augmented two-channel arrhythmia detection

A2CAD was tested on recordings of bipolar intraatrial and intraventricular electrograms (1 cm, l-500 Hz) acquired during electrophysiology studies. The 40 patient cases (lo-50 s) analyzed included 6 atria1 flutters, 4 atria1 fibrillations, 6 supraventricular tachycardias, 2 sinus tachycardias, 12 ventricular tachycardias, 7 ventricular flutters, and 3 ventricular fibrillations. AZCAD successfully diagnosed 40 of 40 cases for an overall success rate of 100%. Materials and Methods

Effect of filtering upon morphological methods for arrhythmia detection in implantable devices

Morphological analysis of intracardiac electrograms (ICEs) has been proposed as a complementary method for discrimination of monomorphic ventricular tachycardia from sinus rhythm by implantable pacemaker-cardioverter-defibrillators. This study compared the impact of filtering on two previously proposed time-domain methods: (1) correlation waveform analysis (CWA-independent of both amplitude fluctuation and ICE baseline) and (2) difference of area (DOA-dependent on amplitude fluctuation and ICE baseline). Using a statistical validation method for nonparametric distributions requiring 75% separation with 95% confidence interval, it was found that with CWA a bandwidth of 10-80 Hz gave results equivalent to that in wideband recordings, and in DOA a bandwidth of 20-80 Hz retained equivalent discriminatory power. A further increase in the lower passband limit (20-80 Hz for CWA and 30-80 Hz for DOA) reduced effectiveness of the methods.<<ETX>>