LVP31a, our new Left Ventricular Pressure Analyzer
11 May 2012

LVP31a, our enhanced Left Ventricular Pressure Analyzer, is now in testing. As stated in a previous Newsletter, this new module will replace LVP30a and offer new and enhanced features.

The first improvement is the built-in cardiac cycle detection. LVP31a has only one input and does not need a cycle detector module anymore.

End-Diastolic Pressure point detection accuracy

One of the major improvements was performed on the End Diastolic Pressure point (EDP) detection. LVP31a detection accuracy was determined on a reference database. This database contains 34 left ventricular pressure recordings from dogs, non-human primates, pigs and minipigs, guinea-pigs, rats and mice. The files also include various pressure patterns such as normal beats, ventricular beats, sinus pauses, beats under cardiac hypertrophy, vasodilatation, heart rate increases and decreases, venous cava occlusion, baseline drifts, artifacts, signal dropouts, etc.

The absolute error of EDP amplitude computation is less than 1 mmHg on almost all files; the mean absolute error on the database is 0.33 mmHg. LVP31a reduces the average error of EDP amplitude estimation by 2.07±1.5 mmHg when comparing to LVP30a on the same database (see graph below).

Average error of EDP amplitude computation in mmHg with LVP30a and LVP31a on 34 files.

 

General detection performance

LVP31a performance was also assessed by comparing automated analysis with manual marking in terms of:
- Sensitivity: ability of the analyzer to detect Systolic, Diastolic and EDP points.
- Positive predictivity: ability of the analyzer to detect only identified Systolic, Diastolic and EDP points.
Sensitivity and positive predictivity of Systolic, Diastolic and EDP detections are above 99% for almost all files (see table below).

LVP31a detection performance results. Results are presented as “mean [minimum value – maximum value]".

 

Left ventricular isovolumic relaxation index computation

Calculation method of the isovolumic relaxation index (Tau) was optimized considering the current knowledge and techniques.

In LVP31a, Tau is computed on the LVP signal segment between the pressure point at minimum derivative of pressure (dP/dt) and a second point relative to EDP of the next cycle (see figure below), whereas LVP30a was using EDP of the current cycle. Using EDP of the next cycle allows determining the signal segment for Tau computation even if the baseline pressure suddenly increases between 2 cycles.

Impact of the EDP selection for determining the LVP signal segment used for Tau computation.

LVP31a offers 5 Tau computation methods: Weiss, Half-pressure, Non-Zero Asymptote, Levenberg-Marquardt, and Logistic.
Weiss, Non-Zero Asymptote and Levenberg-Marquardt methods are now derived from the general equation below:

where P(t) is the left ventricular pressure (in mmHg) as a function of time, is the Pressure at minimum dP/dt when t = 0 ms, is the pressure asymptote to which ventricular pressure tends, and  (Tau) is the time constant (in ms) of isovolumic relaxation.

The Logistic method uses a different equation to estimate Tau and may provide a better curve fit to the left ventricular isovolumic pressure fall in both isovolumic and ejecting contractions (Matsubara et al., 1995).

References

Matsubara H, Takaki M, Yasuhara S, Araki J and Suga H. Logistic Time Constant of Isovolumic Relaxation Pressure–Time Curve in the Canine Left Ventricle – Better alternative to exponential time constant. Circulation 1995; 92: 2318-2326.