- Email: [email protected]
Automated data acquisition and analysis in muscle physiology research
Computer Methods and Programs in Biomedicine, 28 (1989) 273-279 Elsevier
273
CPB 00990
Section II. Systems and programs
Automated data acquisition and analysis in muscle physiology research Thomas E. Stanley, III 1, James P. Held and Jorge A. Estrin 3 t Department of Anesthesiology, The Heart Center, Duke University Medical Center, Durham, NC 27710, U.S.A., 2 University of Minnesota, Minneapolis, MN 55455, U.S.A., and 3 Department of Anesthesiology, University of Minnesota Health Sciences Center, Minneapolis, MN 55455, U.S.A.
Research in muscle physiology has progressed in both the volume and complexity of data examined. Dependence on manual methods to analyze and condense this amount of information can present a narrow bottleneck to the efficient completion of a study, and can compromise the reliability of the results. We designed a general purpose computerized system for data acquisition in experiments in our cardiovascular physiology laboratory. In addition, we developed a software program specifically for the analysis of data from studies of isolated, isometrically contracting myocardium. This system has reduced the time for analysis of such data by approximately 50-fold over that of manual techniques, and has contributed significantly to our confidence in the measured results. Measurement, computer-assisted; Isometrically contracting muscle; Data analysis, computerized
1. Introduction Experiments that study the isometric contraction of isolated muscle are extremely common. Many fundamental aspects of muscle physiology as well as the influences of innumerable pharmacologic agents on muscle contraction properties have been described using this type of model. Increasing familiarity among researchers with the contraction waveforms generated during these experiments has led to the development of sophisiicated analysis schemes that examine subtle changes in these data. Moreover, investigations of the effects of pharmacologic agents using this model require that many samples be recorded at high frequencies
Correspondence: Dr. Thomas E. Stanley, Department of Anesthesiology, P.O.Box 3094, Duke University Medical Center, Durham, NC 27710, U.S.A.
in order to illustrate pharmacodynamic profiles. This has resulted in enormous volumes of data being produced during a given experiment. Relying on manual methods of storage and analysis of these data can present a significant burden to the investigator and pose to threat to the accuracy of the results of the experiment. Computerized data acquisition and an::lysis of experimental data is an attractive solution to this problem. Systems that can provide adequate computational power for this purpose are available at an affordable price. However, software that specifically addresses the requirements for analysis of isolated muscle data is lacking. We present a general purpose computerized data acquisition system, designed for our basic physiology laboratory using commercially available hardware. Along with this system, we have developea a program for display and analysis of data from models of isolated, isometrically contracting muscle.
0169-2607/89/$03.50 © 1989 Elsevier Science Publishers B.V. (Biomedical Division)
274
2. Methods
2.1. Data acquisition
Like most laboratories, ours makes use of a number of in vivo and in vitro experimental models and techniques. This fact was a major influence in the design of the computer system. In order to accommodate present and future experimental designs, a general purpose data acquisition system that would utilize a standard format for data storage was conceived. Separate processing programs could then be custom-tailored to read these universal data files and perform the specific analyses required by the particular experimental protocol. Our facility uses a variety of standard transducers and biophysical amplifiers (Gould, Cleveland, OH, U.S.A.) in all experimentation. The data :.cquisition hardware was chosen to interface with the analog output of these amplifiers. The central unit (Plessey Peripheral Systems, lrvine, CA, U.S.A.) is based on the Digital Equipment Corporation (DEC, Maynard, MA, U.S.A.) PDP11/23 Plus microprocessor system, and includes both hard disk and floppy disk mass storage devices. To this base unit was added an analog-todigital converter (ADAC Corporation, Woburn, MA, U.S,A.), a DEC programmable real-time clock/counter, and a graphics enhanced DEC VT-100 console terminal (RetroGraphics, Digital Engineering, Sacramento, CA, U.S.A.). The computer is controlled using the DEC RT-11 operating system. Application programs were written in PASCAL (Oregon Software) and assembly language (DEC Macro-11). The program for digitizing experimental data in real time is called DATACQ (DATa ACQuisition) and is the general purpose utility to serve the entire laboratory. It was important that this program be as flexible as possible in its interaction with tile user in order to provide the greatest variety of data recording possibilities. A series of comprehensive menus for selection of digitizing parameters was designed for this purpose. After the program is started and the experiment to be recorded is assigned a 'run' number, these menus options are presented to the user. Among the
Fig. 1. Console display of the channel selection menu for DATACQ. A full complement of descriptors is available for each channel. Non-selected channels are indicated as 'idle'.
programmable functions of DATACQ are the selection and identification of the analog inputs to be digitized. The program allows up to eight channels to be uigitized essentially simultaneously (see Appendix). For each of these channels, the user defines a descriptive label, the units of measurement, and minimum and maximum ranges in these units according to the calibration scale of the corresponding analog amplifier (Fig. 1). Any unused channels are indicated as 'idle'. As an example, the primary analog input from isolated muscle studies might be labelled 'Tension', given units of 'grams', and scaled from 0 to 5 in these units. These definitions are stored for the remainder of the experiment, but can be altered at any time during the program's use. Other digitizing parameters available to the user are presented in a separate menu (Fig. 2). Sampling rate is entered as the number of times per second the entire bank of selected analog channels is digitized. The maximum rate of sampling is inversely related to the number of channels selected, but even with all eight channels digitized, exceeds 2900 Hz. Sampling duration is selected in seconds, with its maximum value dependent on both the chosen sampling rate and the number of digitized channels.
275
Often, the experimental data to be recorded occurs at a predictable time, such as the contraction of a muscle in response to an electrical stimulus. The program was designed to take advantage of this fact and provides synchronization of triggering of the digitizing process to an external device. If this option is selected, D A T A C Q waits for a signal from the device, such as a stimulator, then begins the data acquisition. Otherwise, a simple console c o m m a n d starts the digitization. Once acquisition is triggered, the analog-to-digital converter is activated, which sweeps through all selected input channels, storing the digitized data points in a large buffer of system memory via direct memory access. The resulting block of data residing in computer memory is termed a "frame'. The digitizing process is controlled by the programmable real-time clock and is repeated at the defined sampling rate for the defined sample durahon. A maximum of 98 304 individual data points can be recorded for each frame. When the recording is complete, D A T A C Q immediately plots the frame data on the screen for approval of the user (Fig. 3). The data is scaled
!i!i+:!3!!i~ii~i!i~!i~l?~i:i~i~ii~:;i~:i~l!i!i!i(ii~; l/:i~ili~i:i:!i;~ i~i)ii?
iii~i;:i~i?i:~i!~i,~i
:++: ++++++::+: ++' ++:+'::+::::+:+++::++:++:+:+:+:+,+++:+:++:+::+:+:+++~++~ R+ +P+++~m+.++++: ++::++:+:+::+?:+:++:++
:++++ ::++:+: ++:++ :++++++ +++:: +,+ ::+:+ +:'+,+:::. . . . . .
!
.+
.............
......
+++++:++::++++++++++++: +++++++++++++++: +++:+........... +++++++: +++?
.............
.......
.....
+
+%++ ¸,+ ¸i ! il;i!;:i?iiiiil ! i
?iiiili
£
Fig. 3. Plot of a frame of data sampled by DATACQ. Note that the data is scaled and labelled according to the specifications defined for each channel.
according to the defined channel range settings, and all axes, labels, and units are drawn. 7 he user may then accept or abandon the frame. If accepted, the original digitized data, as well as the descriptive variables and settings, including the precise time of :he sampling are written to mass storage in files identified by prescribed run and frame numbers. A final step in automating the data recording ++:++:, :+:++: process is provided by an 'auto-sample' mode. +::+:+++;+:,:++:++++With +++ this feature, sampling of entire frames of data can be repeated at predefined intervals for any period of time with no prompting from the user. This allows the system to record large numbers of data sets at precise frequent intervals, or to +++++,+ carry out certain stable phases of an experiment unattended. All data are maintained in logical order on the mass storage device for easy access by later anal~,sis programs. ~+d~:+
ii
2.2. Isolated isometric muscle contraction analysis Fig. 2. Console display of the sampling parameters menu for DATACQ. Maximum sampling duration for a frame is determined by the number of channels selected and the sampling rate. Plot resolution sets the ratio of total data points to be plotted on the display screen after data sampling. Activation of AutoSample mode allows definition of the frequency and duration of automatically repeated frame acquisitions.
Our model of isometrically contracting muscle is designed according to that described by Blinks [1] and uses right ventricular papillary muscle from cats. The muscles are suspended in a chamber and bathed in a physiologic solution that can maintain their viability for many hours (Fig. 4). Using an
276
I~'
Gould ,~" uc-3 !, L II Strain ~m,~ T r a ~ ToElectrodes Stimulating
.~ !
e- ,o J..
:;~.~.%~
Papillary Muscle
~
17' = :,~'~.
.___~~
1
Perfusate Medium ~:.,6 Drain
i~
~
Superfuaing Gas
The large blocks of 'raw' data that form the muscle contraction contour are of little direct use to the experimenter. Another computer program is therefore used to recall, analyze, and condense the data. This program is called PMLYSE (Papillary Muscle anaLYSE). It interfaces directly with the original data files produced by DATACQ and performs specific analyses required for this model. The usual measurements made of a muscle twitch are shown in Fig. 5. Typically, the resting tension on the muscle at the time of stimulus (RT) and the peak tension of the contraction (PT) are determined. The difference of these values is the active developed tension (AT). Similarly, the maximum rate of tension development (mxdT) and the maximum rate of relaxation (nmdT) are measured from the first differential waveform. We are also interested in determining the amount of time required for these aforementioned measurements to occur (tPT, tmxdq, tmndT). The use of DATACQ makes these latter determinations especially straightforward by synchronizing the stimulating pulse to the data acquisition process, thus identify-
Fig. 4. Isolated papillary muscle -- isometric contraction apparatus.
electrical stimulator, the muscles are made to contract against an isometric strain gauge at a constant rate. The tension measured by this gauge and the analog-derived first differential of this tension constitute the data output of this experimental model. With DATACQ, these two data channels are recorded directly to a digital format in real time. Sampling rate for a (rame (corresponding to a single muscle contraction) is typically 1000 points per second, and sample duration is usually 1-5 s. Triggering of the digitizing process is synchronized to the stimulator. For a single experiment, as many as 200 individual muscle contractions are recorded at frequencies as high as 12 frames per m i , u t e . These program settings result in 350000-500000 data points being obtained by DATACQ. Data for each experiment or run is maintained for analysis on the hard disk device and backup copies archived on floppy disk.
tiT/tit (mN/mr~lsec)
Tension (mN/mm 21
i i
qT
i
Tim== (seconds)
Fig. 5. Typical isometric muscle contraction waveforms. Depicted are the nar~m,,ters mcasured or derived. See text for further description.
277
ing the precise time of stimulus to the analysis program. Characterization of the relaxation phase of the muscle twitch is traditionally quantitated as the time required for the muscle to achieve a 50% reduction in developed tension (t50R). Additional analyses include times to 75% and 95% relaxation (t75R, t95R). More sophisticated muscle relaxation analyses that examine the quotient of maxim u m rate of relaxation and absolute tension at that point in time, and the maximum value of the ratio of these factors for the entire relaxation curve have been described [2]. Finally, integration of the area under the tension curve and examination of the percentage of this area falling under the relaxation phase is of interest. PMLYSE executes these computations on every waveform recorded for an experiment. As in DATACQ, the user is offered substantial flexibility in thi~ analysis. When started, P M L Y S E loads preliminary information for the run, then insures that all DATACQ-created data files are present. The user then inputs the weight and length of the muscle studied. With this information, and assuming that the muscle is a right circular cylinder with a density of 1.05 m g / m m 3 [3], the program calculates its cross-sectional area. All tension measurements are then computed as millinewtons per square millimeter of muscle area ( m N / m m 2). In addition, a variable number of the initial frames can be labelled as 'control state' recordings. During the analysis process, PMLYSE maintains a list of averages of the measurements for these control contractions. For all other 'non-control' frames, the measured parameters are also expressed as the percentages of these mean measurements and stored as an additional group of data. Thus, parameter values that represent percentage change from control are immediately available. When the actual analysis is begun, PMLYSE moves sequentially through the entire set of frames in the run. The waveforms from each frame are again plotted to scale on the computer screen. The program then examines each of the digitized points of the waveform, establishing maxima and minima or threshold values where appropriate, x-Axis or time calculations are m a d e in reference to the time of muscle stimulation, which is known to the
II!III iii¸¸i!iiiiii! !i
ii!iiil
iiiii!illii!iliii!i!ili!!ili!i!ii!
Fig. 6. Console display of reconstructed contraction waveform and completed analysis using PMLYSE. Computer-derived points on the waveformsare marked and the numericpresentation of resulting measurementslisted on the right margin of the screen. The user is then prompted to accept, edit, or reject these data.
program. Area measurements are easily computed using a trapezoidal area algorithm with the resting tension taken as the baseline. Once completed, the results of these computations are displayed on the screen not only by printing the actual numerical values, but also by marking the position on the waveforms from which the parameters were chosen (Fig. 6). The user is given the oppoltunity to accept or reject these results. Alternatively, the user may edit the measurements. This is accomplished using an adjustable cursor on the screen to shift a parameter to a different position on the waveform. The program then recomputes the entire series of derived parameters based on the corrected data point(s). The array of measurements for all accepted frames is stored in tabular form in a separate file as easily readable text. This file can be printed outright for inspection or accessed by other programs for statistical analysis. In m a n y studies using this model, trends in a particular measurement are of interest, and graphical representations of these trends are useful. A third program, called PLOTTER was written to access the condensed file(s) of derived measurements for up to four experiments. The user is offered a choice of parameters to examine. This choice is then plotted on the computer screen on a
278
With the introduction of the computerized data acquisition and analysis system, experimental data of similar volume is fully analyzed and tabulated in 45-60 min. This includes the complex measurements that previously had been abandoned. The resolution of the analog-to-digital converter used is 1:2056, yielding a measurement accuracy of +0.05% of the full scale of the input amplifier. Moreover, the programmable clock that generates the timing pulses for the digitization process operates at a resolution of 1 #s, resulting in a very high degree of accuracy for the time axis parameters.
Fig. 7. Display of output of PLOTTER program. The effects on active developed tension (AT) of a period of hypoxia followed by reoxygenation are plotted for four similar, but separate experiments.
scaled x y grid as the y-axis or dependent value versus the time that the sample was recorded on ~he x-axis for all waveforms of the run (Fig. 7). Thus, changes in particular measurement can be traced over the course of the entire experiment.
3. Results Prior to the introduction of this computer system, our laboratory relied on manual methods for analysis of isolated muscle data. Waveforms were recorded on strip chart paper calibrated to known scales in both horizontal and vertical directions. Many of the simple parameters were chosen and measured using a ruler, converting the actual distances on the paper to the appropriate units. Values such as t50R required several intermediate computations in order to be derived. Moreover, complex measurements such as areas under the waveforms were considered to be impossible to attain accurately. In a single study generating 150-200 waveforms, data analysing using this manual approach required 1-2 full days' work. Since several experiments could be completed in a single day, the analysis phase of this model became the bottleneck. In addition, the tedious nature of this repetitive task cast doubts on the accuracy of the measurements.
4. Discussion The development and use of on-line computerized data acquisition and subsequent computer analysis of the recorded !.-.formation in experiments using an isolated muscle preparation have had a marked positive effect on the quali~, of these studies. We have been able to execute experimental protocols that required critical examination of large volumes of rapidly occurring data. Moreover, several of the parameters easily derived from digitized data, such as specific segmentation of area under a curve, are essentially impossible to deternfine using analog or manual methods. As our expertise in this field of investigation grows, the computer system is able to expand to
Fig. 8. DATACQ, recording of data from an in vivo regional myocardial function experiment.
279
fit later data recording and computational needs. For instance, the addition of transducers to measure isotonic motion of these muscles could be easily a c c o m m o d a t e d by D A T A C Q , while employment of new methods of data analysis is as simple as incorporating the additional routines into the PMLYSE program. The work in our laboratory is not limited to the isolated muscle model. Other investigations have utilized in vivo heart preparations that examine regional cardiac function. The data acquisition program DATACQ has been used without modification to record all of the continuous data generated in this model (Fig. 8). Separate programs are currently being developed to perform the specific analyses required for these investigations.
References [1] J.R. Blinks, A convenient apparatus for recording contractions o~ isolated heart muscle, J. Appl. Physiol. 20 (1965) 755-757.
[2] W.H. Frist, i. Palacios and W.J. Powell, Sr., Effect of hypoxia on myocardial relaxation in isolated cat papillary muscle, J. Cfin. Invest. 61 (1978) 1218-1224. [3] N.F. Paradise, J.L. Schmitter and J.M. Surmitis, Criteria for adequate oxygenation for isometric kitten papillary muscle, Am. J. Physiol. 241 (1981) H348-H353.
Appendix Analog.to-digital cc,nversion
The analog-to-digital converter used in this computer system is a multiplexing unit; one converting device services all selected input channels in sequence. The time required for digitization of an input signal and transfer of the result to memory is 10/ts. Thus, there is a small amount of 'skew' along the time axis between sequential channels for a given sampling. At a typical sampling rate of 1000/s, this aznount of deviation from a true simultaneous sampling of all ell, ,reels represents only a 1% error in the time dimension.
273
CPB 00990
Section II. Systems and programs
Automated data acquisition and analysis in muscle physiology research Thomas E. Stanley, III 1, James P. Held and Jorge A. Estrin 3 t Department of Anesthesiology, The Heart Center, Duke University Medical Center, Durham, NC 27710, U.S.A., 2 University of Minnesota, Minneapolis, MN 55455, U.S.A., and 3 Department of Anesthesiology, University of Minnesota Health Sciences Center, Minneapolis, MN 55455, U.S.A.
Research in muscle physiology has progressed in both the volume and complexity of data examined. Dependence on manual methods to analyze and condense this amount of information can present a narrow bottleneck to the efficient completion of a study, and can compromise the reliability of the results. We designed a general purpose computerized system for data acquisition in experiments in our cardiovascular physiology laboratory. In addition, we developed a software program specifically for the analysis of data from studies of isolated, isometrically contracting myocardium. This system has reduced the time for analysis of such data by approximately 50-fold over that of manual techniques, and has contributed significantly to our confidence in the measured results. Measurement, computer-assisted; Isometrically contracting muscle; Data analysis, computerized
1. Introduction Experiments that study the isometric contraction of isolated muscle are extremely common. Many fundamental aspects of muscle physiology as well as the influences of innumerable pharmacologic agents on muscle contraction properties have been described using this type of model. Increasing familiarity among researchers with the contraction waveforms generated during these experiments has led to the development of sophisiicated analysis schemes that examine subtle changes in these data. Moreover, investigations of the effects of pharmacologic agents using this model require that many samples be recorded at high frequencies
Correspondence: Dr. Thomas E. Stanley, Department of Anesthesiology, P.O.Box 3094, Duke University Medical Center, Durham, NC 27710, U.S.A.
in order to illustrate pharmacodynamic profiles. This has resulted in enormous volumes of data being produced during a given experiment. Relying on manual methods of storage and analysis of these data can present a significant burden to the investigator and pose to threat to the accuracy of the results of the experiment. Computerized data acquisition and an::lysis of experimental data is an attractive solution to this problem. Systems that can provide adequate computational power for this purpose are available at an affordable price. However, software that specifically addresses the requirements for analysis of isolated muscle data is lacking. We present a general purpose computerized data acquisition system, designed for our basic physiology laboratory using commercially available hardware. Along with this system, we have developea a program for display and analysis of data from models of isolated, isometrically contracting muscle.
0169-2607/89/$03.50 © 1989 Elsevier Science Publishers B.V. (Biomedical Division)
274
2. Methods
2.1. Data acquisition
Like most laboratories, ours makes use of a number of in vivo and in vitro experimental models and techniques. This fact was a major influence in the design of the computer system. In order to accommodate present and future experimental designs, a general purpose data acquisition system that would utilize a standard format for data storage was conceived. Separate processing programs could then be custom-tailored to read these universal data files and perform the specific analyses required by the particular experimental protocol. Our facility uses a variety of standard transducers and biophysical amplifiers (Gould, Cleveland, OH, U.S.A.) in all experimentation. The data :.cquisition hardware was chosen to interface with the analog output of these amplifiers. The central unit (Plessey Peripheral Systems, lrvine, CA, U.S.A.) is based on the Digital Equipment Corporation (DEC, Maynard, MA, U.S.A.) PDP11/23 Plus microprocessor system, and includes both hard disk and floppy disk mass storage devices. To this base unit was added an analog-todigital converter (ADAC Corporation, Woburn, MA, U.S,A.), a DEC programmable real-time clock/counter, and a graphics enhanced DEC VT-100 console terminal (RetroGraphics, Digital Engineering, Sacramento, CA, U.S.A.). The computer is controlled using the DEC RT-11 operating system. Application programs were written in PASCAL (Oregon Software) and assembly language (DEC Macro-11). The program for digitizing experimental data in real time is called DATACQ (DATa ACQuisition) and is the general purpose utility to serve the entire laboratory. It was important that this program be as flexible as possible in its interaction with tile user in order to provide the greatest variety of data recording possibilities. A series of comprehensive menus for selection of digitizing parameters was designed for this purpose. After the program is started and the experiment to be recorded is assigned a 'run' number, these menus options are presented to the user. Among the
Fig. 1. Console display of the channel selection menu for DATACQ. A full complement of descriptors is available for each channel. Non-selected channels are indicated as 'idle'.
programmable functions of DATACQ are the selection and identification of the analog inputs to be digitized. The program allows up to eight channels to be uigitized essentially simultaneously (see Appendix). For each of these channels, the user defines a descriptive label, the units of measurement, and minimum and maximum ranges in these units according to the calibration scale of the corresponding analog amplifier (Fig. 1). Any unused channels are indicated as 'idle'. As an example, the primary analog input from isolated muscle studies might be labelled 'Tension', given units of 'grams', and scaled from 0 to 5 in these units. These definitions are stored for the remainder of the experiment, but can be altered at any time during the program's use. Other digitizing parameters available to the user are presented in a separate menu (Fig. 2). Sampling rate is entered as the number of times per second the entire bank of selected analog channels is digitized. The maximum rate of sampling is inversely related to the number of channels selected, but even with all eight channels digitized, exceeds 2900 Hz. Sampling duration is selected in seconds, with its maximum value dependent on both the chosen sampling rate and the number of digitized channels.
275
Often, the experimental data to be recorded occurs at a predictable time, such as the contraction of a muscle in response to an electrical stimulus. The program was designed to take advantage of this fact and provides synchronization of triggering of the digitizing process to an external device. If this option is selected, D A T A C Q waits for a signal from the device, such as a stimulator, then begins the data acquisition. Otherwise, a simple console c o m m a n d starts the digitization. Once acquisition is triggered, the analog-to-digital converter is activated, which sweeps through all selected input channels, storing the digitized data points in a large buffer of system memory via direct memory access. The resulting block of data residing in computer memory is termed a "frame'. The digitizing process is controlled by the programmable real-time clock and is repeated at the defined sampling rate for the defined sample durahon. A maximum of 98 304 individual data points can be recorded for each frame. When the recording is complete, D A T A C Q immediately plots the frame data on the screen for approval of the user (Fig. 3). The data is scaled
!i!i+:!3!!i~ii~i!i~!i~l?~i:i~i~ii~:;i~:i~l!i!i!i(ii~; l/:i~ili~i:i:!i;~ i~i)ii?
iii~i;:i~i?i:~i!~i,~i
:++: ++++++::+: ++' ++:+'::+::::+:+++::++:++:+:+:+:+,+++:+:++:+::+:+:+++~++~ R+ +P+++~m+.++++: ++::++:+:+::+?:+:++:++
:++++ ::++:+: ++:++ :++++++ +++:: +,+ ::+:+ +:'+,+:::. . . . . .
!
.+
.............
......
+++++:++::++++++++++++: +++++++++++++++: +++:+........... +++++++: +++?
.............
.......
.....
+
+%++ ¸,+ ¸i ! il;i!;:i?iiiiil ! i
?iiiili
£
Fig. 3. Plot of a frame of data sampled by DATACQ. Note that the data is scaled and labelled according to the specifications defined for each channel.
according to the defined channel range settings, and all axes, labels, and units are drawn. 7 he user may then accept or abandon the frame. If accepted, the original digitized data, as well as the descriptive variables and settings, including the precise time of :he sampling are written to mass storage in files identified by prescribed run and frame numbers. A final step in automating the data recording ++:++:, :+:++: process is provided by an 'auto-sample' mode. +::+:+++;+:,:++:++++With +++ this feature, sampling of entire frames of data can be repeated at predefined intervals for any period of time with no prompting from the user. This allows the system to record large numbers of data sets at precise frequent intervals, or to +++++,+ carry out certain stable phases of an experiment unattended. All data are maintained in logical order on the mass storage device for easy access by later anal~,sis programs. ~+d~:+
ii
2.2. Isolated isometric muscle contraction analysis Fig. 2. Console display of the sampling parameters menu for DATACQ. Maximum sampling duration for a frame is determined by the number of channels selected and the sampling rate. Plot resolution sets the ratio of total data points to be plotted on the display screen after data sampling. Activation of AutoSample mode allows definition of the frequency and duration of automatically repeated frame acquisitions.
Our model of isometrically contracting muscle is designed according to that described by Blinks [1] and uses right ventricular papillary muscle from cats. The muscles are suspended in a chamber and bathed in a physiologic solution that can maintain their viability for many hours (Fig. 4). Using an
276
I~'
Gould ,~" uc-3 !, L II Strain ~m,~ T r a ~ ToElectrodes Stimulating
.~ !
e- ,o J..
:;~.~.%~
Papillary Muscle
~
17' = :,~'~.
.___~~
1
Perfusate Medium ~:.,6 Drain
i~
~
Superfuaing Gas
The large blocks of 'raw' data that form the muscle contraction contour are of little direct use to the experimenter. Another computer program is therefore used to recall, analyze, and condense the data. This program is called PMLYSE (Papillary Muscle anaLYSE). It interfaces directly with the original data files produced by DATACQ and performs specific analyses required for this model. The usual measurements made of a muscle twitch are shown in Fig. 5. Typically, the resting tension on the muscle at the time of stimulus (RT) and the peak tension of the contraction (PT) are determined. The difference of these values is the active developed tension (AT). Similarly, the maximum rate of tension development (mxdT) and the maximum rate of relaxation (nmdT) are measured from the first differential waveform. We are also interested in determining the amount of time required for these aforementioned measurements to occur (tPT, tmxdq, tmndT). The use of DATACQ makes these latter determinations especially straightforward by synchronizing the stimulating pulse to the data acquisition process, thus identify-
Fig. 4. Isolated papillary muscle -- isometric contraction apparatus.
electrical stimulator, the muscles are made to contract against an isometric strain gauge at a constant rate. The tension measured by this gauge and the analog-derived first differential of this tension constitute the data output of this experimental model. With DATACQ, these two data channels are recorded directly to a digital format in real time. Sampling rate for a (rame (corresponding to a single muscle contraction) is typically 1000 points per second, and sample duration is usually 1-5 s. Triggering of the digitizing process is synchronized to the stimulator. For a single experiment, as many as 200 individual muscle contractions are recorded at frequencies as high as 12 frames per m i , u t e . These program settings result in 350000-500000 data points being obtained by DATACQ. Data for each experiment or run is maintained for analysis on the hard disk device and backup copies archived on floppy disk.
tiT/tit (mN/mr~lsec)
Tension (mN/mm 21
i i
qT
i
Tim== (seconds)
Fig. 5. Typical isometric muscle contraction waveforms. Depicted are the nar~m,,ters mcasured or derived. See text for further description.
277
ing the precise time of stimulus to the analysis program. Characterization of the relaxation phase of the muscle twitch is traditionally quantitated as the time required for the muscle to achieve a 50% reduction in developed tension (t50R). Additional analyses include times to 75% and 95% relaxation (t75R, t95R). More sophisticated muscle relaxation analyses that examine the quotient of maxim u m rate of relaxation and absolute tension at that point in time, and the maximum value of the ratio of these factors for the entire relaxation curve have been described [2]. Finally, integration of the area under the tension curve and examination of the percentage of this area falling under the relaxation phase is of interest. PMLYSE executes these computations on every waveform recorded for an experiment. As in DATACQ, the user is offered substantial flexibility in thi~ analysis. When started, P M L Y S E loads preliminary information for the run, then insures that all DATACQ-created data files are present. The user then inputs the weight and length of the muscle studied. With this information, and assuming that the muscle is a right circular cylinder with a density of 1.05 m g / m m 3 [3], the program calculates its cross-sectional area. All tension measurements are then computed as millinewtons per square millimeter of muscle area ( m N / m m 2). In addition, a variable number of the initial frames can be labelled as 'control state' recordings. During the analysis process, PMLYSE maintains a list of averages of the measurements for these control contractions. For all other 'non-control' frames, the measured parameters are also expressed as the percentages of these mean measurements and stored as an additional group of data. Thus, parameter values that represent percentage change from control are immediately available. When the actual analysis is begun, PMLYSE moves sequentially through the entire set of frames in the run. The waveforms from each frame are again plotted to scale on the computer screen. The program then examines each of the digitized points of the waveform, establishing maxima and minima or threshold values where appropriate, x-Axis or time calculations are m a d e in reference to the time of muscle stimulation, which is known to the
II!III iii¸¸i!iiiiii! !i
ii!iiil
iiiii!illii!iliii!i!ili!!ili!i!ii!
Fig. 6. Console display of reconstructed contraction waveform and completed analysis using PMLYSE. Computer-derived points on the waveformsare marked and the numericpresentation of resulting measurementslisted on the right margin of the screen. The user is then prompted to accept, edit, or reject these data.
program. Area measurements are easily computed using a trapezoidal area algorithm with the resting tension taken as the baseline. Once completed, the results of these computations are displayed on the screen not only by printing the actual numerical values, but also by marking the position on the waveforms from which the parameters were chosen (Fig. 6). The user is given the oppoltunity to accept or reject these results. Alternatively, the user may edit the measurements. This is accomplished using an adjustable cursor on the screen to shift a parameter to a different position on the waveform. The program then recomputes the entire series of derived parameters based on the corrected data point(s). The array of measurements for all accepted frames is stored in tabular form in a separate file as easily readable text. This file can be printed outright for inspection or accessed by other programs for statistical analysis. In m a n y studies using this model, trends in a particular measurement are of interest, and graphical representations of these trends are useful. A third program, called PLOTTER was written to access the condensed file(s) of derived measurements for up to four experiments. The user is offered a choice of parameters to examine. This choice is then plotted on the computer screen on a
278
With the introduction of the computerized data acquisition and analysis system, experimental data of similar volume is fully analyzed and tabulated in 45-60 min. This includes the complex measurements that previously had been abandoned. The resolution of the analog-to-digital converter used is 1:2056, yielding a measurement accuracy of +0.05% of the full scale of the input amplifier. Moreover, the programmable clock that generates the timing pulses for the digitization process operates at a resolution of 1 #s, resulting in a very high degree of accuracy for the time axis parameters.
Fig. 7. Display of output of PLOTTER program. The effects on active developed tension (AT) of a period of hypoxia followed by reoxygenation are plotted for four similar, but separate experiments.
scaled x y grid as the y-axis or dependent value versus the time that the sample was recorded on ~he x-axis for all waveforms of the run (Fig. 7). Thus, changes in particular measurement can be traced over the course of the entire experiment.
3. Results Prior to the introduction of this computer system, our laboratory relied on manual methods for analysis of isolated muscle data. Waveforms were recorded on strip chart paper calibrated to known scales in both horizontal and vertical directions. Many of the simple parameters were chosen and measured using a ruler, converting the actual distances on the paper to the appropriate units. Values such as t50R required several intermediate computations in order to be derived. Moreover, complex measurements such as areas under the waveforms were considered to be impossible to attain accurately. In a single study generating 150-200 waveforms, data analysing using this manual approach required 1-2 full days' work. Since several experiments could be completed in a single day, the analysis phase of this model became the bottleneck. In addition, the tedious nature of this repetitive task cast doubts on the accuracy of the measurements.
4. Discussion The development and use of on-line computerized data acquisition and subsequent computer analysis of the recorded !.-.formation in experiments using an isolated muscle preparation have had a marked positive effect on the quali~, of these studies. We have been able to execute experimental protocols that required critical examination of large volumes of rapidly occurring data. Moreover, several of the parameters easily derived from digitized data, such as specific segmentation of area under a curve, are essentially impossible to deternfine using analog or manual methods. As our expertise in this field of investigation grows, the computer system is able to expand to
Fig. 8. DATACQ, recording of data from an in vivo regional myocardial function experiment.
279
fit later data recording and computational needs. For instance, the addition of transducers to measure isotonic motion of these muscles could be easily a c c o m m o d a t e d by D A T A C Q , while employment of new methods of data analysis is as simple as incorporating the additional routines into the PMLYSE program. The work in our laboratory is not limited to the isolated muscle model. Other investigations have utilized in vivo heart preparations that examine regional cardiac function. The data acquisition program DATACQ has been used without modification to record all of the continuous data generated in this model (Fig. 8). Separate programs are currently being developed to perform the specific analyses required for these investigations.
References [1] J.R. Blinks, A convenient apparatus for recording contractions o~ isolated heart muscle, J. Appl. Physiol. 20 (1965) 755-757.
[2] W.H. Frist, i. Palacios and W.J. Powell, Sr., Effect of hypoxia on myocardial relaxation in isolated cat papillary muscle, J. Cfin. Invest. 61 (1978) 1218-1224. [3] N.F. Paradise, J.L. Schmitter and J.M. Surmitis, Criteria for adequate oxygenation for isometric kitten papillary muscle, Am. J. Physiol. 241 (1981) H348-H353.
Appendix Analog.to-digital cc,nversion
The analog-to-digital converter used in this computer system is a multiplexing unit; one converting device services all selected input channels in sequence. The time required for digitization of an input signal and transfer of the result to memory is 10/ts. Thus, there is a small amount of 'skew' along the time axis between sequential channels for a given sampling. At a typical sampling rate of 1000/s, this aznount of deviation from a true simultaneous sampling of all ell, ,reels represents only a 1% error in the time dimension.