Apparatus for determining right ventricular ejection fraction from high fidelity thermodilution curves.

BACKGROUND OF THE INVENTION

0001The right ventricle receives blood from the right atrium through the tricuspid valve and pumps the blood through the pulmonic valve to the pulmonary artery. In pumping blood, the right ventricle expands during diastole to take in blood through the tricuspid valve and contracts during systole to discharge blood through the pulmonic valve into the pulmonary artery.

0002It is sometimes necessary or desirable to determine the effectiveness of the right ventricle in pumping blood to the pulmonary artery, and for this purpose, right heart ejection fraction is determined. Ejection fraction is determined by comparing the expanded volume or end diastolic volume (EDV) of the right ventricle with the contracted volume or end systolic volume (ESV) of the right ventricle. Mathematically, ejection fraction (EF) can be expressed as follows:<maths id="math0001" num=""><img file="EP0246920A2_D0001.tif" /></maths>

0003Ejection fraction is calculated from thermodilution curves by hand computation. Thermodilution is typically carried out by injecting a cold indicator into the right ventricle or right atrium and allowing the indicator to be diluted with blood. As this cold mixture is pumped through the right ventricle into the pulmonary artery, the temperature of the fluid in the pulmonary artery falls and then rises. The temperatures of the fluid in the pulmonary artery are measured during the time that the temperature is rising and compared with a prebolus baseline temperature of blood in the pulmonary artery to establish temperature differentials. The temperature differentials are then used to determine ejection fraction. Ejection fraction calculation is widely discussed in the literature, such as <ul id="ul0001" list-style="none"><li>1. <u style="single">Journal of Surgical Research</u>, "Measurement of Ejection Fraction by Thermal Dilution Techniques", H.R. Kay et al, Vol. 34, 337-346 (1983).</li><li>2. <u style="single">The Journal of Trauma</u>, "Thermodilution Right Ventricular Volume: A Novel and Better Pre­dictor of Volume Replacement in Acute Thermal Injury", J.A.J. Martyn et al, Vol. 21, No. 8, 619-626 (1981).</li><li>3. <u style="single">ASA</u>, "Ejection Fraction By Thermodilution" (Abstract), G.G. Maruschak et al, Vol. 55, No. 3, Sept. 1981.</li></ul>

0004Hand calculation of ejection fraction is typically performed using the plateau method which can be mathematically stated as follows:<maths id="math0002" num=""><img file="EP0246920A2_D0002.tif" /></maths> where, T_i is the temperature at any of the plateaus on the downslope (i.e. during the time the temperature in the pulmonary artery is rising) of the thermodilution curve, T_i+1 is the temperature at the immediately following plateau, and T_B is the baseline temperature. Hand calculation of ejection fraction using the plateau method is, of course, slow and, as shown by equation 2, utilizes only two points in the cardiac cycle, see <u style="single">American Journal of Cardiology</u>, "Usefulness and Limitations of Thermal Washout Techniques in Ventricular Volume Measurement", E. Rapaport, Vol. 18, 226-234, August 1966.

0005In addition, calculation of ejection fraction using the plateau method is consistently lower than ejection fraction as calculated using radionuclear techniques. Accordingly, it is desirable to refine the ejection fraction determinations so that more accurate results are obtained.

SUMMARY OF THE INVENTION

0006This invention automates EF determinations and provides other important advantages. This invention is based, in part, upon the recognition that some of the assumptions underlying the calculation of ejection fraction using equation 2 are erroneous. For example, one such assumption is that the baseline temperature T_B of the blood entering the right ventricle does not change from a time just before injection of the bolus through the times that the temperature is measured in the pulmonary artery for the purpose of determining EF. This is not true if atrial injection is utilized. Because injection of the cold indicator is carried out over several heart beats, some indicator is pooled in the atrium, and this changes the baseline temperature T_B of the blood entering the right ventricle from the prebolus baseline temperature prior to injection. This is one reason why ejection fraction calculations made by the plateau method tend to be low.

0007This invention compensates for the baseline temperature drift without adding another temperature sensor to the catheter used in taking the temperature measurements. This is accomplished by establishing a temperature versus time relationship or curve using at least some of the measured temperatures in the pulmonary artery when the temperature of fluid in the pulmonary artery is rising. This curve projects temperatures between the measured temperatures and beyond the highest measured temperature used to establish the curve. A temperature established by the curve which is below the last measured temperature is then used to establish the post bolus baseline temperature.

0008For the best results, the curve preferably is or approximates a first order exponential curve. This curve can be used in various different ways to approximate a post bolus baseline temperature T_B2 which produces more accurate results than the prebolus baseline temperature which is the blood temperature entering the right ventricle prior to injection of the cold indicator into the atrium. In any event, the post bolus baseline temperature will be lower than the prebolus baseline temperature due to the various cooling factors identified above.

0009A preferred technique for establishing the post bolus baseline temperature is to use as the post bolus baseline temperature, the temperature identified by the first order exponential curve as the curve approaches its asymptote. Although the use of this curve to establish a post bolus baseline temperature is particularly adapted for injection of the indicator upstream of the right ventricle, such as in the atrium, it may also be used for ventricular injection.

0010The plateau method as determined by equation 2 also assumes that the temperatures sensed in the pulmonary artery accurately represent the temperature of the fluid in the pulmonary artery. However, testing has shown that the response time of the catheter mounted temperature sensor is not adequate to monitor 100 percent of the temperature change within a given heartbeat interval. Numerical modeling provides a means by which the response time can be compensated for when the heartbeat or R-R interval is known. This enables more accurate computer measurement of the temperature changes in the pulmonary artery.

0011This can be accomplished, for example, by determining the response time for a group of the catheter-mounted sensors which are to be utilized. A correction factor can then be applied to the calculated EF.

0012Another assumption underlying the use of equation 2 is that there is no arrhythmic heart activity during the time that the temperatures in the pulmonary artery are being measured to determine ejection fraction. This invention recognizes the possibility of such arrhythmic heart activity and provides for arrhythmia detection. If arrhythmic heart activity is detected during the time of interest, an alarm is activated to advise of the arrhythmic heart activity.

0013The invention, together with additional features and advantages thereof, may best be understood by reference to the following description taken in connection with the accompanying illustrative drawing.

BRIEF DESCRIPTION OF THE DRAWINGS

0014<ul id="ul0002" list-style="none"><li>Fig. 1 is a sectional view through the right heart showing one form of catheter which can be used in the determination of ejection fraction.</li><li>Fig. 2 is an idealized, illustrative plot of temperature of the fluid in the pulmonary artery versus time and of the corresponding "R" waves and detected "R" waves.</li><li>Fig. 3 is a flow chart showing the basic steps in the determination of ejection fraction.</li><li>Fig. 4 is a plot of temperature versus time illustrating how the post bolus baseline temperature is established.</li><li>Fig. 5 is a plot of percent response versus time for one catheter-mounted thermistor that may be used in carrying out this invention.</li><li>Fig. 6 is a block diagram showing one apparatus embodying the invention.</li></ul>

DESCRIPTION OF THE PREFERRED EMBODIMENT

0015Fig. 1 illustrates one form of catheter 11 inserted into a human heart 13 for the purpose of carrying out the present invention. Although the catheter 11 may be of various different constructions and be just a temperature probe, in this embodiment, the catheter includes an elongated flexible tube 15 having an injectate port 17, a balloon 19 closely adjacent the distal end of the tube and a temperature sensor in the form of a thermistor 21 proximal of the balloon but adjacent the distal end of the tube. A more complete description of the catheter 11 can be obtained from Webler et al application Serial No. 570,631 filed January 13, 1984, now U.S. Patent No. 4,632,125, which is incorporated by reference herein.

0016The catheter 11 is inserted into the heart 13 using conventional medical techniques to place the balloon 19 and the thermistor 21 in pulmonary artery 23 and to place the injectate port 17 in right atrium 25. Thus, the catheter extends through tricuspid valve 27, right ventricle 29 and the pulmonic valve 31 to the pulmonary artery 23.

0017To determine right heart ejection fraction or cardiac output, a bolus of cold indicator, such as saline solution, is rapidly injected over several heart beats through a lumen of the tube 15 and the injectate port 17 into the right atrium 25. In the illustrated embodiment, the injectate is directed generally toward inferior vena cava 32 counter current to the blood flow from the inferior vena cava so as to provide good mixing with the blood. During diastole, the right ventricle 29 expands, and the tricuspid valve 27 opens to allow some of the blood-indicator mixture to enter the right ventricle. During systole, the right ventricle 29 contracts to force or pump the blood-indicator mixture through the pulmonic valve 31, into the pulmonary artery 23 and across the thermistor 21. Because the injection of the cold indicator takes place over multiple heart beats, the temperature of the fluid in the pulmonary artery 23 reduces from a prebolus baseline temperature T_B1 (Fig. 2) in multiple increments to a lowermost temperature or temperature peak T_PK and then increases in increments such that the temperature curve asymptotically approaches the prebolus baseline temperature T_B1. As shown in Fig. 2, each temperature step or plateau of the thermodilution (TD) curve 33 immediately follows an "R" wave or right ventricle 29 contraction or heart beat. It should be noted that the TD curve 33 is inverted in that the prebolus baseline temperature T_B1 is a higher temperature than the temperature peak T_PK. Using the TD curve 33 and equation 2, it is possible to calculate ejection fraction as more fully described in, for example, Webler et al application Serial No. 570,631 now U.S. Patent No. 4,632,125 referred to above.

0018The present invention provides for the automated determination of ejection fraction, although hand computation utilizing the principles of this invention is also possible. This invention can be implemented with the catheter 11 and a suitable instrument 34 (Fig. 1) which may include suitable electronic hardware, software and a microcomputer or a combination of the two. A software implementation is preferred to carry out the steps shown in Fig. 3, and in that connection, it is only necessary to make certain modifications to a program known as COM-1 used by equipment available from American Edwards Laboratories of Santa Ana, California, for the purpose of computing cardiac output.

0019With the catheter 11 in the heart 13 as shown in Fig. 1, the thermistor 21 provides continuous temperature information concerning the temperature of the fluid, i.e., blood or blood-indicator mixture in the pulmonary artery 21 to the instrument 34. In a digital system, this temperature information is sampled periodically, such as every 71 milliseconds by a sampler in the instrument 34. With reference to Fig. 3, and excluding the usual preliminaries of the type used in the COM-1 program, such as noisy baseline identification and thermodrift detection, the first step is prebolus baseline determination, i.e., determining the prebolus baseline temperature T_B1 To accomplish this, the temperature samples are averaged in any suitable manner, such as by calculating a running average of the samples. Although T_B1 is actually measured in the pulmonary artery, it can be safely assumed that, prior to injection of the cold indicator, the temperature of the blood entering the right ventricle is the same as in the pulmonary artery.

0020With T_B1 determined, the operator can initiate a start command, and a bolus of cold indicator is injected through the injectate port 17 into the right atrium 25 with such injection taking place over multiple heart beats. The indicator cools the blood and forms a blood-indicator mixture which is pumped through the right ventricle 29 to the pulmonary artery to provide the TD curve 33 shown in Fig. 2. The start command also brings about the storage of raw dT values and the associated detected "R" waves 35 (Fig. 2) as, for example, a twelve-bit word. The dT values are the difference between the temperature defined by the TD curve 33 and the prebolus baseline temperature T_B1 as shown in Fig. 2. The dT samples are taken periodically, such as every 71 milliseconds, and the buffer for storing such samples may be, for example, 1024 words in length for approximately 73 seconds of storage. Every "R" wave is stored in the "R" wave buffer in the instrument 34 and is stored as a detected "R" wave 35. The "R"-wave buffer and the buffer for the dT values are synchronized in time so that, for each dT value, the presence or absence of an "R" wave during the associated dT sample time is stored as a "1" or "0" in the "R"-wave buffer. This buffer synchronization facilitates correlation between heart electrical activity and fluid movement by the heart as manifested in temperature changes, i.e., dT values. Although the dT values can be manipulated in various different ways, it is preferred to store all of the dT values. In addition, a real time running average of recent dT values is maintained, as represented by the half-second window storage. For example, every 7 dT samples may be averaged and stored as a .5 second average. Thus, the first 7 dT values are averaged, then dT values 2 through 8 are averaged and so on.

0021Next, the peak temperature T_PK , or lowest temperature, is determined. This can be accomplished, for example, by identifying the largest stored .5 second dT average as the peak temperature T_PK. T_PK and the time when it occurs are stored.

0022Next, the deviation 80% and 30% values are calculated by multiplying .8 and .3, respectively, times the peak temperature T_PK. These calculated values are stored.

0023Using the stored .5 second dT averages, the 80 percent deviation determination is located on the TD curve 33. Specifically, the first of such average dT values following T_PK that is equal to, or less than .8 times T_PK is stored as T_D1. The time at which the temperature T_D1 occurs is also stored.

0024Similarly, the 30 percent of deviation determination is also located. The first of the stored average dT values following T_PK that is equal to or less than .3 times the temperature T_PK is identified as T_D2 and is stored along with its reference time.

0025Establishing T_D1 and T_D2 as approximately equal to .8 times T_PK and .3 times T_PK, respectively, is desirable but not critical. However, other points on the downslope of the TD curve 33 between about .95 times T_PK and .15 times T_PK can be used, if desired.

0026The evaluation interval is then determined as the first step of post processing. To enhance repeatability and allow for a good curve fit, it is desired to consistently locate the evaluation interval in accordance with a particular program. Generally, this can be accomplished by determining the "R" waves occurring closest to T_D1 and T_D2 and their respective times of occurrence. Various programs for choosing such "R" waves can be used. For example, if T_D1 occurs between "R" waves, the first of such "R" waves, i.e., the "R" wave nearer T_PK, is used to establish the upper limit temperature T_u if such "R" wave's corresponding temperature amplitude is within 12.5 percent of the temperature T_D1 and is less than 90 percent of the peak temperature T_PK. Also, this "R" wave must occur after the occurrence of the peak temperature T_PK. If these synchronization conditions are met for such first "R" wave, then the temperature corresponding to the time of occurrence of such "R" wave will be used as the upper limit temperature T_u. If these synchronization conditions cannot be achieved for such first "R" wave, then the temperature corresponding to the time of occurrence of the "R" wave immediately following the occurrence of the temperature T_D1 will be used as the upper limit temperature T_u.

0027The "R" wave nearest the temperature T_D2 must be at least two R-R intervals beyond the upper "R" wave. If the "R" wave which is two R-R intervals forward down the TD curve 33 is from 15 to 30 percent of T_PK, then this point is used as T_L as shown in Fig. 2. If this "R" wave is above 30 percent of T_PK, then the temperature that corresponds to the "R" wave that is closest to 30 percent of T_PK is used. If the temperature along the TD curve 33 at the end of the second R-R interval is less than 15 percent of T_PK, then the temperature at the end of the first R-R interval is used for T_L. If the temperature at the first R-R interval is still less than 15 percent of T_PK, an error message is given.

0028The respective upper and lower limit temperature values T_u and T_L are stored and each is preferably an average, such as a 3-point average, of the data on each side of the associated "R" wave. For example, if T₂ corresponds to the lower "R" wave synchronization point, the actual temperature used for T would be as follows:<maths id="math0003" num=""><img file="EP0246920A2_D0003.tif" /></maths> where, T₁ and T₃ are stored temperatures dT on opposite sides of T₂.

0029The temperatures T_L and T_u which also constitute evaluation limits always coincide with "R" wave events as shown in Fig. 2.

0030Next a post bolus baseline temperature T_B2 is determined utilizing T_u and T_L and force fitting a first order exponential curve to these two points as shown in Fig. 4. This calculation is made using the prebolus baseline temperature T_B1 as follows: TD = Δe^-t/τ<u style="single">Equation 4</u> where, TD is the value on the curve 41 Δ is the temperature T_u t is time and <maths id="math0004" num=""><img file="EP0246920A2_D0004.tif" /></maths> where, t_L is the time at which T_L occurs and t_u is the time at which T_u occurs.

0031Thus, by implementing equation 4, the curve 41 of Fig. 4 can be plotted and extrapolated beyond T_L . Because the curve 41 is or approximates a first order exponential curve, it asymptotically approaches the prebolus baseline temperature T_B1 .

0032This invention establishes as the post bolus baseline temperature T_B2 the temperature which exists near the time when the curve 41 closely approaches T_B1 . Although various levels can be employed, a preferred approach is to utilize a threshold of .01 to .05 of T_PK with .03 of T_PK being optimum. The time t_f at which this threshold temperature occurs can be obtained by solving for "t" in equation 4 which yields t_a, which is the difference between t_F and t_u. The time t_F can be found from the equation t_F=t_u +t_a. Next, the raw temperature data that corresponds to the time t_F is located, and this is established as the post bolus baseline temperature T_B2. Preferably, an average, such as a 70-point average of the temperature data beginning at t_F is used to establish the post bolus baseline T_B2 , i.e., an average of the temperatures occurring in the pulmonary artery in the next 2.5 to 5 seconds may be used to establish T_B2.

0033The post bolus baseline temperature T_B2 is subtracted from the curve 41 to provide a new upper limit temperature T_nu and a new lower limit temperature T_nL. A new curve 43 can then be force fit to the new upper and lower limit temperatures T_nu and T_nL as shown in dashed lines in Fig. 4. The curve 43 asymptotically approaches the post bolus baseline temperature T_B2. The number of "R" waves occurring during the evaluation interval, i.e., between t_u and t_L are determined, and the duration of the evaluation interval is calculated. From this, preliminary ejection fraction can be calculated from the following equation:<maths id="math0005" num=""><img file="EP0246920A2_D0005.tif" /></maths> EIT = t_L - t_u n = the number of "R" waves occurring during the evaluation interval.

0034The preliminary ejection fraction calculation is then corrected based upon the response time of the catheter mounted thermistor 21. For this purpose, the response of the catheter mounted thermistor 21 is plotted as a function of time using any suitable technique, and one such plot for the thermistor 21 as mounted on the catheter 11 is shown in Fig. 5.

0035Although various techniques can be utilized to determine the response, to plot Fig. 5, a group of the catheters 11 having the thermistor 21 thereon were tested for response time data at the 63 percent, 90 percent and 95 percent responses. The average of these data points at these responses are shown by the points D, E, F on the response curve 51. Beyond the point F, the curve 51 approaches an asymptote A which represents the maximum percent response for the catheter-mounted thermistor. For a catheter mounted thermistor, a second order exponential curve 51 is a good approximation of the percent response as a function of time, with the curve 51 being influenced primarily by the first order component from the origin to a division response or point G, which, in this example, is beyond the point D and at about 70 percent response, and with the curve 51 being primarily influenced by the second order component above, response G. By placing a time delay before the second order component, its effect on the first order component is delayed; hence the shape of the curve between the points G and E. The functional form of the equation for the second order exponential curve 51 is: A [1-e^(-Bt)-(t)e^(-Ct)] = % response <u style="single">Equation 6</u> where, A is the % response at the asymptote, B and C are the first and second order slope coeffi­cients, and t is time.

0036The best curve fit using this form is achieved with A equal to 97, B equal to 12 and C equal to 1.8.

0037As shown by Fig. 2, the temperature during the downslope of the curve 33 changes with each heart beat. Accordingly, the time for the thermistor 21 to react to a temperature change is equal to the R-R interval. The model shown by way of example in Fig. 5 shows the known approximate percent response to temperature change as a function of time, i.e., about how rapidly the catheter-mounted thermistor 21 responds during any given time interval. Although this could be used to correct every temperature value, this would be quite complex, and it has been found that a good approximation for correcting the ejection fraction can be determined as follows:<maths id="math0006" num=""><img file="EP0246920A2_D0006.tif" /></maths> To utilize equation 7, the length of the R-R interval or the average length of such intervals between t_u and t_L determine the time in seconds, and from this the approximate percent response can be determined from the curve 51. Thus, an R-R interval of one second would provide an 80 percent response which in turn would provide a correction factor of 1.25 which should be multiplied by the preliminary EF to obtain the corrected ejection fraction.

0038Of course, the ejection fraction can be calculated multiple times from multiple injections of cold indicator, if desired. The mathematical functions and the steps described above can be readily implemented with software.

0039An optional, but important, feature of the invention is the setting of a flag or the providing of an alarm if any arrhythmic heart activity occurs during the evaluation interval. This can be accomplished by appropriate monitoring of the "R" wave data stored in the "R" wave buffer. Although this can be accomplished in different ways, the present invention provides a 4-beat running average of the "R" wave intervals with all abnormal beats and the beat following any abnormal beat not being used in the average; i.e., with each new beat, a new average of the 4 most recent normal beats is taken. Although various factors could be monitored, this invention considers heart activity which is out of range, premature, or delayed to be abnormal or arrhythmic in nature. For example, individual heart beats and the preceding R-R interval which are equivalent to heart rates below 35 per minute or over 180 per minute are considered out of range. Delayed beats are those which are separated by more than one-and-one-half times the current 4-beat average, and a premature beat is any beat which has its preceding R-R interval 20 beats per minute faster than the current 4-beat average. Thus, the present invention provides an alarm if any out of range, delayed or premature beats occur during the evaluation interval by monitoring the stored R-wave data.

0040Fig. 6 shows by way of example a block diagram of the components of the instrument 34. Analog temperature data from the catheter 11 is fed through an isolation amplifier 101 to an A/D converter which samples the raw temperature data periodically, such as every 71 milliseconds, to provide dT temperature samples or values to a microcomputer 105. Similarly, "R"-wave information is fed through an isolation amplifier 107 to an "R"-wave detector 109 which provides the detected "R" waves 35 to the microcomputer 105. The microcomputer 105 has the storage capability to store the temperature samples dT and the detected "R" waves 35 and to perform the other functions discussed above. A display 111 may be provided to display, for example, the calculated ejection fraction.

0041As used herein, the term "catheter" refers to any probe or catheter. The injection and temperature-measuring functions of the catheter can be carried out by separate catheters, if desired.

0042Although an exemplary embodiment of the invention has been shown and described, many changes, modifications and substitutions may be made by one having ordinary skill in the art without necessarily departing from the spirit and scope of this invention.