ESTIMATING A VALUE ASSOCIATED WITH HEART WALL TENSION

Abstract

A method of estimating a value associated with heart wall tension. The method comprises: using motion data recorded with a sensor in communication with the heart to identify motion in the heart; and estimating a value associated with heart wall tension based on the identified motion in the heart. The motion in the heart that forms the basis of the estimation may be a vibration in the heart wall. A heat monitoring system for carrying out the method of estimating a value associated with heart wall tension comprises a sensor configured to be placed in communication with the heart in order to identify motion in the heart; and a data processing device arranged to receive motion data from the sensor and to then carry out the steps of the method.

Claims

1. A method of estimating a value associated with tension in the heart wall, the method comprising: using motion data recorded with a sensor in communication with the heart to identify motion in the heart; and estimating a value associated with tension in the heart wall based on the identified motion in the heart.

2. A method as claimed in claim 1, wherein the identified motion in the heart wall is inherent to the heart as part of the normal functioning of the heart.

3. A method as claimed in claim 2, wherein the identified motion in the heart results from the opening and/or closing of at least one of the mitral, aortic, pulmonary and tricuspid valves, or wherein the identified motion in the heart results from the contraction and expansion of the heart.

4. A method as claimed in any preceding claim, wherein the sensor comprises at least one accelerometer, magnetometer and/or at least one gyro.

5. A method as claimed in any preceding claim, wherein the sensor is an invasive sensor implanted at the heart, for example at the epicardium, the endocardium and/or the myocardium.

6. A method as claimed in claim 5, wherein the sensor is comprised within a pacemaker lead positioned at the heart.

7. A method as claimed in any of claims 1 to 4, wherein the sensor is a non-invasive sensor, such as a sensor based on computed tomography, echocardiography, magnetic resonance imaging, radar.

8. A method as claimed in any preceding claim, wherein the motion data is recorded in real-time such that the estimated value associated with tension in the heart wall is a real-time estimation.

9. A method as claimed in any preceding claim, wherein the step of estimating a value associated with tension in the heart wall based on the identified motion in the heart comprises repeatedly estimating a value associated with tension in the heart wall over a period of time covered by the recorded motion data such that dynamic changes in the value associated with tension in the heart wall can be assessed.

10. A method as claimed in any preceding claim, further comprising estimating a volemic state of the heart based on the estimated value associated with tension in the heart wall.

11. A method as claimed in any preceding claim, wherein the value associated with tension in the heart wall is a tension in the heart wall.

12. A method as claimed in claim 11, wherein the method further comprises the step of estimating a ventricular preload based on the estimated tension in the heart wall.

13. A method as claimed in any of claims 1 to 10, wherein the value associated with tension in the heart wall is a ventricular preload.

14. A method for producing a patient specific Frank-Starling curve comprising: carrying out the steps of the method as defined in claim 12 or 13 such that an estimate of ventricular preload is obtained; obtaining a measurement of patient cardiac function contemporaneous with the estimate of ventricular preload; and producing a patient specific Frank-Starling curve based on the measurement of patient cardiac function and the estimate of ventricular preload.

15. A method as claimed in claim 14, comprising: carrying out the steps of the method as defined in claim 12 or 13 at least once more at a different point in time covered by the motion data such that at least one further estimate of ventricular preload is obtained; obtaining at least one further measurement of patient cardiac function contemporaneous with the at least one further estimate of ventricular preload; and producing the patient specific Frank-Starling curve based on each measurement of patient cardiac function and each estimate of ventricular preload.

16. A method as claimed in claim 15, wherein at the different point in time covered by the motion data the heart is of a different volemic state due to fluid loading and/or unloading of the heart.

17. A method as claimed in any of claims 14 to 16, wherein the, or each, measurement of cardiac function is measured using motion data recorded with the sensor in communication with the heart.

18. A method as claimed in any of claim 13, wherein the identified motion in the heart includes heart wall motion over at least one ventilation cycle of the patient.

19. A method as claimed in claim 18, wherein the method comprises the step of determining a variation in the heart wall motion over the at least one ventilation cycle of the patient, and wherein the step of estimating ventricular preload comprises estimating a ventricular preload based on the variation in the heart wall motion over the at least one ventilation cycle of the patient.

20. A method as claimed in any of claims 1 to 17, wherein the identified motion in the heart includes a vibration in the heart wall.

21. A method claimed in claim 20, further comprising: determining a frequency of the vibration in the heart wall; and calculating an estimated tension in the wall of the heart or an estimated ventricular preload based on the frequency of the vibration.

22. A heart monitoring system comprising: a sensor configured to be placed in communication with the heart in order to identify motion in the heart; and a data processing device arranged to: receive motion data from the sensor; identify motion in the heart from the received motion data; and estimate a value associated with tension in the heart wall based on the identified motion in the heart.

23. A heart monitoring system according to claim 22, wherein the data processing device is arranged to carry out the method of any of claims 1 to 21.

24. A heart monitoring system according to claim 22 or 23, wherein the sensor comprises at least one accelerometer, magnetometer and/or at least one gyro.

25. A heart monitoring system according to claim 22, 23, or 24, wherein the sensor is an invasive sensor configured to be implanted at the heart, for example at the epicardium, the endocardium and/or the myocardium.

26. A heart monitoring system as claimed in claim 25, wherein the sensor is comprised within a pacemaker lead that is to be positioned at the heart.

27. A heart monitoring system according to any of claims 22 to 24, wherein the sensor is a non-invasive sensor, such as a sensor based on computed tomography, echocardiography, magnetic resonance imaging or radar.

28. A computer programme product comprising instructions that, when executed, will configure a data processing apparatus to perform the method of any of claims 1 to 21.

Description

[0071] Certain example embodiments of the present invention will now be described, by way of example only, and with reference to the accompanying embodiments, in which:

[0072] FIG. 1 shows a graphical illustration of Starling's Law;

[0073] FIG. 2 shows a graphical illustration of varying airway pressure and arterial pressure over time;

[0074] FIG. 3 show a graphical illustration of varying stroke volume over time;

[0075] FIG. 4 shows graphical illustrations of data recorded using epicardially implanted accelerometers over time, along with various other hemodynamic parameters;

[0076] FIG. 5 shows other graphical illustrations of data recorded using epicardially implanted accelerometers over time;

[0077] FIG. 6 shows graphical illustrations of data recorded using epicardially implanted accelerometers, intravascular PiCCO™ catheters, and from a pulmonary artery occlusion pressure analysis over time, along with regression analyses thereof;

[0078] FIG. 7 shows two Bland-Altman plots representative of the agreement between data recorded with epicardially placed accelerometers and sonomicrometric measurements, and between data recorded with intravascular PiCCO™ catheters and using sonomicrometric measurements;

[0079] FIG. 8 shows graphical illustrations of Frank-Starling relations derived from epicardially implanted accelerometers and other modalities; and

[0080] FIG. 9 shows a graphical illustration of a receiver operating characteristics analysis performed on data recorded using epicardially implanted accelerometers and other modalities.

[0081] The following discussion pertains to a study carried out by the inventors to estimate left ventricular preload, and thereby left ventricular filling status, using an epicardially placed accelerometer within a number of pigs. The accelerometer was also used to assess fluid responsiveness of the heart. The study as presented herein is merely an example of the aspects of the invention as set forth above, and as defined in the claims. The scope of protection should not be seen to be limited to this specific implementation. For instance, whilst the below discussion pertains to the estimation of ventricular preload, it will be understood that the invention is not limited to preload estimations and that, as described, the invention has broader applications in the estimation of values associated with tension in the heart wall more generally (e.g. tension in itself). Moreover, whilst the below discussion exemplifies the use of epicardially implanted accelerometers, it will be appreciated that other sensors and modalities may be used to determine heart wall motion in accordance with aspects of the invention. Furthermore, the below discussion outlines how heart wall vibration is used as the basis of estimation; however variation in heart wall motion may equally be used for estimation of ventricular preload.

SUMMARY

[0082] In the inventors' study, heart wall vibrations were continuously measured using a three-axis accelerometer sutured to the left ventricular (LV) epicardium in 9 pigs during baseline, fluid loading and phlebotomy in a closed chest condition. So as to evaluate the accuracy and reliability of the results using the three-axis accelerometer, sonomicrometry was used as gold standard for continuous measurements of LV volume, and a change in end diastolic volume (EDV.sub.SONO) was used as an index for a change in preload. Additionally, global end-diastolic volume was estimated using PiCCO™ (GEDV.sub.PICCO) and pulmonary artery occlusion pressure (PAOP) was measured with a pulmonary arterial catheter. Linear regression and Bland-Altman analyses were performed to determine the accuracy of f.sub.s1, GEDV.sub.PICCO, and PAOP referenced to EDV.sub.SONO. A receiver operating characteristics (ROC) analysis was performed to determine the diagnostic accuracy (area under ROC curve) of identifying fluid responsiveness (defined as stroke volume variation (SVV) by sonomicrometry >11.6%) for f.sub.s1, SVV.sub.PICCO, pulse pressure variation (PPV.sub.PICCO), GEDV.sub.PICCO, and PAOP.

[0083] Methodology

[0084] 9 land-race pigs (4 male) having an average weight 45 kg (±2 kg SD.) were used as the basis of the study. The animals were premedicated with an intramuscular injection of ketamine 20 mg/kg, azaperone 3 mg/kg and atropine 0.02 mg/kg. Anaesthesia was induced by intravenous thiopental 2-3 mg/kg and morphine 0.5-1 mg/kg. Immediately after induction of anaesthesia, tracheotomy was performed and anaesthesia upheld by inhalation (Isoflurane 1-2%) and morphine 0.50-1.0 mg/kg/h, adjusted by the animal's autonomous stress response. A Leon respirator was used for ventilation and gas monitoring with inspired oxygen fraction of 0.35. The animals were monitored by electrocardiogram (ECG), peripheral oxygen saturation (SpO2), temperature and diuresis. The internal and external jugular veins were cannulated for introduction of a central venous pressure catheter and pulmonary artery catheter (Edwards Lifesciences Corporation, Irvine, Calif., USA). The carotid arteries were cannulated to introduce two 5-Fr Millar pressure catheters into the left ventricle (LV) and aortic outlet (AO). A PiCCO™ catheter was introduced via a femoral artery.

[0085] After introduction of anaesthesia and hemodynamic monitoring, sternotomy was performed, and the pericardial sack was split from apex to base exposing the heart. To measure the LV volume, sonomicrometry crystals were placed sub-endocardially in a long axis pair (apex to anterior base), and short axis pair (equatorial, postero-lateral to antero-septal). From these two pairs the continuous volume could be estimated using the formula V=π/6W.sup.2H, where W and H were the distance between the short axis pair and long axis pair, respectively.

[0086] To measure the myocardial vibrations, an inertial sensor (MPU9250, InvenSense Inc, San Jose, Calif., USA) incorporating a 3-axis accelerometer was placed in the anterior LV apical region.

[0087] Lastly, the thorax was closed by suturing sternum and skin. The intervention protocol was started after a 30 minutes stabilisation period following instrumentation.

[0088] After baseline measurements for each pig using the various different sensor modalities, including baseline measurements of ventricular preload for each pig, the ventricular preload of each pig was artificially altered by volume loading of 250 ml 0.9% NaCl solution in intervals until a 10% increase in end diastolic volume (EDV) was observed in the sonomicrometric volume trace. Recordings were obtained between each interval of loading. To decrease preload, blood was drained from the central venous catheter into heparinised bags in 250 ml or 500 ml intervals, in between recordings, until a 10% decrease in EDV from baseline was achieved.

[0089] Pulmonary artery occlusion pressure (PAOP) was recorded to assess the LV filling pressure. Using sonomicrometry, end diastolic volume (EDV.sub.SONO), stroke volume (SV.sub.SONO), cardiac output (CO.sub.SONO), and stroke volume variation (SVV.sub.SONO) were recorded as reference values. From the PiCCO recordings, cardiac output (CO.sub.PICCO), global end-diastolic volume (GEDV.sub.PICCO), stroke volume (SV.sub.PICCO), stroke volume variation (SVV.sub.PICCO), and pulse pressure variation (PPV.sub.PICCO) were extracted for comparison with sonomicrometry and accelerometer derived values.

[0090] The epicardially placed accelerometers were used to measure the vibrations within the myocardium at the time of mitral valve closure and the first heart sound, events, which are contemporaneous with the start of systole. From the frequency of these measurements left ventricular preload could then be estimated. This estimation made was based on the derivation from Mersenne's law of a vibrating monochord, which is set out in equation (3) above. By modelling each muscle fibre in the wall of the heart as a monochord, and under an assumption that at any given time the density and length of the muscle fibres remain constant, it was possible to estimate the left ventricular preload due to its direct proportionality with the square of the frequency of a vibration in the wall of the heart at the time of mitral valve closure and the first heart sound. The frequency of the vibrations measured in the heart wall using the epicardially placed accelerometer were analysed in order to estimate left ventricular preload and were correlated with alterations in preload as described in further below.

[0091] FIG. 4 shows representative data recorded using the epicardially implanted accelerometer along with various other heart measurement data recorded using the different sensor modalities over a common time period. Specifically, the top panel 41 of FIG. 4 shows raw acceleration data recorded using the implanted accelerometer as the sum of the three axes of the accelerometer. The second panel 43 shows the frequency spectrum over time (wavelet scalogram) for the sum of the acceleration data traces. The black trace in the frequency spectrum (fc) represents the centre frequency for each time point, which was extracted by calculating the weighted average frequency for each time point.

[0092] The third panel 45 shows a plot of LV pressure (the blue trace) and aortic pressure (the orange trace), the fourth panel 47 shows a plot of LV volume, and the fifth panel shows electrocardiographical (ECG) data recorded. Each panel 41-49 is plotted on a common time abscissa as shown at the bottom of panel 49. The vertical grey lines 51 in each panel indicate the time of Q in the ECG, which is considered to be the event indicating the end of the diastolic phase of the heart/the start of the systolic phase of the heart. The vertical green lines 53 in each panel indicate the time that the rate of change of LV pressure is at a minimum. This minimum was considered to indicate the end of the systolic phase of the heart.

[0093] Python™ Software Foundation, version 3.6.9, was used for all signal pre-processing, including the pre-processing of the raw acceleration date (e.g. as in panel 41). From the data recorded, data from at least 3 ventilation cycles were used for the basis of analysis to ensure that the values were unaffected by changes due to respiration.

[0094] The raw accelerometer signals (e.g. as in panel 41) were filtered using a 5.sup.th order Butterworth band pass filter with cut off frequencies at 20 Hz and 250 Hz to remove high frequency noise and unwanted respiration and base heart rate effects on the frequency analysis. Since the accelerometer senses in three spatial directions (x, y, and z), the frequency analysis on each axis were performed individually and the average of these were taken for basis of analysis so as to nullify the orientation of the accelerometer placement and therefore the effects of gravity.

[0095] Frequency (wavelet) analysis of the raw accelerometer signals was used to analyse the frequency components of the first heart sound/mitral valve closure using MATLAB™. A continuous wavelet transform was used to produce a spectrogram (e.g. as in panel 43) containing the power of each frequency component in a signal over time. The colours in the spectrogram represent the magnitude of the particular frequency components in the transform, with the yellow/orange regions representing higher magnitude frequency components and the bluer regions representing lower magnitude frequencies.

[0096] FIG. 5 shows further traces of raw acceleration data and corresponding wavelet transforms and spectrograms similar to those shown in panels 41 and 43 of FIG. 4 respectively. The traces of FIG. 5 pertain to data recorded during the first 150 ms after end diastole (shown in panels 61a-61c) and the corresponding continuous wavelet transforms and spectrograms (shown in panels 63a-63c) produced from the raw acceleration data. Panels 61a and 63a are representative of data recorded for a hypovolemic heart, panels 61b and 63b pertain to normovolemic heart, panels 61c and 63c pertain to a hypervolemic heart. Again, the black trace represents the centre frequency (f.sub.c) for each timepoint, whilst the red circle (f.sub.s1) represents the point of maximum frequency within the time window.

[0097] To investigate preload changes, an analysis of the shift in the centre frequency f.sub.c was undertaken on the recorded data. An increasing centre frequency f.sub.c was predicted to indicate an increased preload. To determine a shift in the centre frequency f.sub.c, the maximum value of the centre frequency f.sub.s1 during the first heart sound within 0.03 and 0.1 s after end diastole was used as a comparator. To further reduce noise, the centre frequency trace for all heart cycles spanning the recording were averaged and the centre frequency f.sub.s1 was calculated based on this average. A correlation analysis was then performed to assess the relationship between the myocardial frequency and end diastolic volume as determined by sonomicrometry measurement (EDV.sub.SONO).

[0098] The EDV.sub.SONO values were used as an index for changes in ventricular preload. Linear regression and Bland-Altman analyses were performed to determine the accuracy of the first heart sound frequency f.sub.s1, GEDV.sub.PICCO, and PAOP referenced to EDV.sub.SONO. In order to perform a comparison of how well different measurements with different units (Hz, ml, and mmHg) correlated with EDV.sub.SONO (ml), the relative percentage change in the parameter values from baseline were also assessed and correlated.

[0099] The preload parameters estimated using the implanted accelerometer was subsequently used to continuously monitor the patient's Frank-Starling relation. SV.sub.SONO, as a measure of cardiac function, was correlated with the estimated preload determined using the implanted accelerometer in order to produce a Frank-Starling curve. The produced curve allowed for comparison with Frank-Starling curves derived using different sensor modalities.

[0100] The estimated preload parameters were also used in monitoring and assessing fluid responsiveness of the heart. In the assessment of fluid responsiveness a threshold value for stroke volume variation (SVV) was set as 11.6% as has been previously been determined in the art. Using SVV.sub.SONO as reference, the accuracy of the accelerometer based method to correctly identify the fluid responsiveness was investigated and compared with PiCCO™ based measurements and pulmonary artery occlusion pressure. A receiver operating characteristics (ROC) analysis was performed, and the area under the ROC curve (AUC) for each method was used to assess and compare their accuracies.

[0101] All statistical analyses were performed using R (v3.5.1), developed by Foundation for Statistical Computing. A total of fifteen experiments were conducted. Six experiments were excluded due to either equipment malfunction (n=2), fatal bleeding due to left atrial rupture (n=1), or ventricular fibrillation (n=3). To test for significant effects of the volemic interventions in Table 1 (see below) students T-test was used. Normality of distributions was determined using Shapiro-Wilks test. To test differences in the area under two receiver operating characteristic curves (AUC) we used DeLong's test. Significance was determined as p≤0.05.

[0102] Results

[0103] Table 1 below sets out the results determined from the above nine experiments. The results from experiments carried out on one animal were excluded from the results given in Table 1 since it did not reach a state of overloading during fluid loading. Three columns are presented in Table 1, showing the baseline hemodynamic variables, the variables determined after maximum fluid loading of the heart, and the variables determined after maximum fluid unloading. The values in Table 1 are reported as mean (SD). *: p<0.05 vs baseline using paired sample t-test. The acronyms set out in Table 1 are as follows: EDV—end diastolic volume; SV—stroke volume; CO—cardiac output; EF—ejection fraction; SW—stroke work; SVV—stroke volume variation; f.sub.s1[x, y, z, avg]—first heart sound frequency; GEDV—global end-diastolic volume; PPV—pulse pressure variation; MAP—mean arterial pressure; CVP—central venous pressure; HR—heart rate; SvO2—venous oxygen saturation; PAOP—pulmonary artery occlusion pressure. As noted, one animal was excluded from this table as it did not reach a state of overloading since unloading was performed before loading.

[0104] The EDV.sub.SONO changed significantly both at maximum loading and maximum unloading. So did SV.sub.SONO, CO.sub.SONO, EF.sub.SONO, SW.sub.SONO and SVV.sub.SONO. The first heart sound frequency f.sub.s1, measured using the accelerometer, showed a significant decrease for all axes (x, y, and z) in addition to the average, during maximum volume unloading. The first heart sound frequency f.sub.s1 showed a significant increase for x and z axes, as well as the average, during maximum volume loading, but not for the y-axis. The PiCCO based parameters (GEDV.sub.PICCO, SV.sub.PICCO, CO.sub.PICCO, SVV.sub.PICCO, PPV.sub.PICCO) all showed a significant change during both unloading and loading. Pulmonary artery occlusion pressure did not significantly change for either extreme.

TABLE-US-00001 TABLE 1 Baseline Fluid loading Fluid unloading Sonomicrometry EDV [mL] 112 (27) 121 (30)* 94 (23)* SV [mL] 42 (10) 50 (14)* 32 (9.6)* CO [L/min] 4.5 (1.7) 5.3 (2.3)* 3.6 (1.4)* EF [%] 38 (6.3) 42 (7.1)* 34 (6.1)* SW [mL .Math. mmHg] 2840 (1246) 3727 (1473)* 1586 (837)* SVV [%] 16 (7.5) 9.6 (4.0)* 24 (9.7)* Accelerometer f.sub.S1 x [Hz] 83 (8.8) 89 (13)* 75 (7.2)* f.sub.S1 y [Hz] 85 (11) 88 (11) 75 (8.6)* f.sub.S1 z [Hz] 79 (12) 86 (14)* 71 (12)* f.sub.S1 avg [Hz] 82 (8.7) 88 (8.4)* 74 (8.6)* PiCCO GEDV [mL] 588 (80) 624 (118)* 483 (69)* CO [L/min] 4.8 (1.1) 5.9 (1.3)* 4.2 (1.4)* SVV [%] 12 (4.1) 6.2 (1.3)* 15 (7.5)* PPV [%] 15 (6.3) 8.6 (3.8)* 24 (6.0)* Hemodynamics MAP [mmHg] 70 (15) 80 (15)* 53 (11)* CVP [mmHg] 11 (1.7) 14 (3.4)* 9.1 (1.8)* HR [bpm] 104 (31) 104 (29) 113 (32) SvO2 [%] 59 (9.6) 55 (7.1) 44 (12)* PAOP [mmHg] 13 (2.2) 16 (4.2) 12 (3.6)

[0105] FIG. 6 shows a graphical representation of results of the experiments carried out in the study and statistics derived therefrom. In FIG. 6, subplots A, C and E show the correlation between EDV.sub.SONO and f.sub.s1, GEDV.sub.PICCO and PAOP, respectively for the individual pigs. Subplots B, D and F show the correlation between relative change in EDV.sub.SONO (ΔEDV.sub.SONO) and relative change in f.sub.s1 from baseline (Δf.sub.s1), relative change in GEDV.sub.PICCO (ΔGEDV.sub.PICCO) and relative change in PAOP (ΔPAOP) respectively for data pooled from all experiments. In subplots B, D and F the solid black line signifies a regression line, whilst the black, stippled line signifies y=x.

[0106] Ventricular preload estimated by the first heart sound frequency f.sub.s1 showed a very strong correlation to the reference EDV.sub.SONO. This is shown in plot A of FIG. 6. The global end-diastolic volume (GEDV.sub.PICCO), shown in plot C of FIG. 6, also showed a very good correlation to EDV, although with a lower average coefficient of determination (r.sup.2). The PAOP had the lowest average r.sup.2, shown in plot E of FIG. 6. When pooling the data for all subjects and looking at the relative change from baseline in the estimated preload parameters Δf.sub.s1, ΔGEDV.sub.PICCO, PAOP and ΔEDV.sub.SONO, the correlation was stronger for Δf.sub.s1 with an r.sup.2=0.81, as shown in plot B of FIG. 6, than for ΔGEDV.sub.PICCO (plot D) and PAOP (plot F) with r.sup.2 values of 0.59 and 0.36 respectively.

[0107] FIG. 7 shows two Bland-Altman plots, A and B. Plot A demonstrates the agreement between Δf.sub.s1 and ΔEDV.sub.SONO. Plot B demonstrates the agreement between ΔGEDV.sub.PICCO and ΔEDV.sub.SONO. The bias (1.96 SD) for Δf.sub.s1 was 1.0 (8.9) %. The bias (1.96 SD) for ΔGEDV.sub.PICCO was 4.0 (17.5) %. The plots are based on data pooled from all experiments. Black, solid lines signify bias, red lines signify bias±1.96 SD.

[0108] As can be seen in FIG. 7, the Bland-Altman analysis showed that the bias of f.sub.s1 compared to the EDV was 1.0% with limits of agreement at −7.9 and 10.0%. The mean bias, and spread, for GEDV were higher than for f.sub.s1, though not significantly (p=0.11, paired sample t-test), having a value of 4.0% and limits of agreement at −13.6 and 21.5%.

[0109] FIG. 8 shows the estimated Frank-Starling relation using the reference SV.sub.SONO, and the results determined for EDV, the accelerometer derived index for preload f.sub.s1, GEDV.sub.PICCO and PAOP. The results from the linear regression analysis assessing the correlation between SV.sub.SONO and EDV.sub.SONO (plot 81), the estimated preload parameters f.sub.s1 (plot 83), GEDV.sub.PICCO (plot 85), and PAOP (plot 87) are also shown as solid lines for each pig. The average r.sup.2 over all pigs for each plot 81, 83, 85, 87 is also shown.

[0110] As can be seen in plot 83, there was a very strong correlation between SV.sub.SONO and f.sub.s1, which was stronger than for GEDV.sub.PICCO as shown in plot 85, when assessing the average correlation coefficient for individual animals. The correlation between SV.sub.SONO and GEDV.sub.PICCO was in turn stronger than for PAOP shown in plot 87. The pooled results, based on relative changes of EDV.sub.SONO, f.sub.s1, GEDV.sub.PICCO and PAOP from baseline, are also shown in FIG. 8 as plots 71, 73, 75 and 77 respectively. The correlation to stroke volume for f.sub.s1 was stronger than for PAOP, which in turn was stronger than GEDV.sub.PICCO.

[0111] FIG. 9 is a plot showing ROC curves for f.sub.s1 (red trace), SVV.sub.PICCO (gold trace), PPV.sub.PICCO (light green trace), GEDV.sub.PICCO (dark green trace), PAOP (blue trace) and heart rate (HR— purple trace) for identifying the measurements where the SVV values determined by sonomicrometry were above 11.6% which, as noted above, was used as the indicator for fluid responsiveness. The areas under each curve (AUC values) for the different predictors are also shown.

[0112] As shown in FIG. 9, the AUC-value was highest for f.sub.s1 but the difference was not statistically significant compared to SVV.sub.PICCO (p=0.21), PPV.sub.PICCO (p=0.09), GEDV.sub.PICCO (p=0.23), or PAOP (p=0.10). There was only a statistically significant difference as compared to heart rate (p=0.01). The sensitivity and specificity for f.sub.s1, choosing the optimal cut-off value from the ROC analysis, was 87 and 67%, respectively. For SVV.sub.PICCO, PPV.sub.PICCO, GEDV.sub.PICCO, PAOP, and heart rate the sensitivities and specificities were, 81 and 60%; 70 and 60%; 56 and 71%; 55 and 64%; and 57 and 67%, respectively.

[0113] The foregoing discussion is evidence of the inventors' realisation that motion of the heart can be used to monitor changes and estimate a value associated with tension in the heart wall (i.e. in form of ventricular preload). In particular the foregoing discussion demonstrates that changes in ventricular preload can be monitored by an accelerometer attached to the epicardium by virtue of an assessment of the changes in the first heart sound frequency (f.sub.s1).

[0114] Further, the foregoing discussion shows a strong correlation between the frequency of the myocardial vibrations associated with the first heart sound f.sub.s1 with EDV and myocardial stiffness. Thus, monitored heat wall motion, in particular vibrations at end diastole/start systole etc., can be used to monitor a volemic state of a patient continuously and in real time. Hence, advantageously an accelerometer placed, for example, epicardially may be used to monitor a patient's volemic state post-operatively.

[0115] It has also been demonstrated that the accelerometer-based measurements as presented herein for the estimation of ventricular preload and identifying fluid responsiveness are comparable to the current clinical standard methods, e.g. PiCCO™ and PAOP.

[0116] Furthermore, since the miniaturized accelerometer used as the basis of the study presented herein may be incorporated in a temporary pace lead, which is routinely placed on open heart surgery patients, the methods presented may not require an additional procedure (e.g. as in the case of PiCCO systems) for the assessment and monitoring of a value associated with tension in the heart wall (e.g. preload) and/or a volemic state of the heart. They can instead rely on equipment which has already been implanted at the heart for a different purpose. This is clearly advantageous in respect of patient safety.

[0117] The use of an implanted accelerometer as presented above ensures that contact with the blood stream is avoided.

[0118] Both the accelerometer based method as detailed above and the prior art PiCCO™ systems allow for a continuous measure of hemodynamic values. However, PiCCO™ systems are dependent on a thermodilution calibration and will drift based on vascular compliance and must therefore be recalibrated when changes in vascular compliance occur. In addition, the accuracy and precision of PiCCO™ based monitoring will also be affected by aortic valve regurgitation and over- or under-damped arterial pressure waveforms. As the accelerometer based method as presented above, and indeed alternative methods (e.g. those reliant on variations in heart wall motion) encompassed within the scope of the present invention, are reliant on an identification of heart wall motion, and are not dependent/reliant on pressure waveforms, the accuracy and precision of such methods should not be affected by the aforementioned conditions.

[0119] Combined with a measure of heart function, like peak systolic velocity, pressure-displacement loop area or others, it is also clear that a robust estimate of a patients position on the Frank-Starling curve can be determined based on the ventricular preloads estimated using methods in accordance with the aspects of the invention. This was reflected in the results of the linear regression analysis, assessing the correlation of the preload parameters to stroke volume. The correlation between SV.sub.SONO and f.sub.s1 was stronger than both the clinically available measures GEDV.sub.PICCO and PAOP, when assessing both the individual and the pooled data. The PiCCO also showed large variation in the slopes of the individual regression lines, while the slopes for f.sub.s1 were more aligned, which suggests that accelerometer based measurements may be less susceptible to inter-subject variability.

[0120] The above described methods have also been shown to be capable of predicting fluid responsiveness, which is of significant importance as fluid resuscitation is the first line of defence in critically ill patients. It is shown above that the diagnostic accuracy (area under the ROC curve) for identifying fluid responsiveness using f.sub.s1 (i.e. using the epicardially implanted accelerometer) was comparable to SVV.sub.PICCO and PPV.sub.PICCO methods and thus the methods of the invention are at least comparable to prior art systems when it comes to determining fluid responsiveness.

[0121] Furthermore, since at least the exemplary method of the invention that is reliant on vibrations in the heart wall as set out in the above discussion is not reliant on mechanical ventilation of a patient, such methods should prevail in predicting fluid responsiveness over SVV measured using traditional PiCCO™ systems in in open chest conditions, in patients on pressure support ventilation, during pericardial effusion, etc. (i.e. in conditions under which there is no, or artificially altered, mechanical ventilation of the patient).

[0122] Additionally, since the methods of the invention as disclosed herein are reliant on identified motion in the heart, such methods most likely not be affected by complications in the peripheral system, like peripheral vascular disease and thus can be used in the hemodynamic assessment of patients with arrhythmias, valvular disease, intracardiac shunts, peripheral vascular disease and decreased ejection fraction, which prior art systems, such as PiCCO™ systems reliant on SVV cannot.

[0123] As noted above, whilst the above exemplary embodiment of the invention is in the context of estimating ventricular preload, the invention is not limited as such. This is demonstrated for example by the frequency spectrum in panel 43 of FIG. 6. As will be appreciated from the above discussion, this frequency spectrum can be seen as a surrogate for tension in the myocardium across the whole heart cycle. Thus, tension in the myocardium (and associated parameters) can be estimated for any point within the heart cycle. For instance, at around 0.5-0.55 s of panel 43 it can be seen there are two peaks at the end of systole. These peaks will likely represent the contractility of the myocardium and as such the invention may comprise estimating contractility. The frequency spectrum is a continuous measure and may represent many different physiological parameters of the heart at different timepoints.