Quantifying Phase Analysis Parameters for Normal Cardiac Synchrony

• Published in: 2006 IEEE Nuclear Science Symposium Conference Record Vol. 5; pp. 3180 - 3184
• Main Authors: Dej, B., Wassenaar, R.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.10.2006
• Subjects: Angiography; Cathode ray tubes; Heart; Histograms; Image analysis; Medical treatment; Pacemakers; Pixel; RNA; Smoothing methods
• ISBN: 9781424405602; 1424405602
• ISSN: 1082-3654
• DOI: 10.1109/NSSMIC.2006.356551

Congestive heart failure (CHF) may result in decreased quality of life or death. Cardiac resynchronization therapy (CRT) uses pacemakers to control cardiac contraction and improve the quality of life in CHF patients, but currently suffers from a 30-50% non-responder rate. In phase analysis, the gated counts of each pixel in a radionuclide angiography image are fit to a cosine to obtain a phase shift, thetas, and amplitude, A. Phase analysis of RNA can potentially assess cardiac synchrony and predict CRT outcome. The standard deviation (phaseSD) of a histogram of thetas values was previously used to assess dyssynchrony, but it is unclear whether this is sufficient for determining the extent or underlying cause of dyssynchrony. Synchrony (S) and entropy (E) have been recently proposed to better describe cardiac synchrony. Limited work has been done quantifying these new parameters, particularly the effect of histogram bin size, noise, and smoothing. In this paper, normal values of SD, S and E were quantified for a range of bin sizes, histogram noise levels, and image filtering. Noise was removed using a noise threshold cut-off method. Pre-analysis smoothing was performed using a Butterworth filter (0.5 cm -1 , order of 11). PhaseSD, S and E values were plotted for various noise threshold cut-off values. Values approaching the theoretical values (S = 1, E = 0 for normal), with low inter-subject standard deviation, were found using pre-analysis smoothing, a noise threshold cut-off value of approximately 20% of the maximum histogram value and a bin size of 12deg to 24deg.