Real-time acquisition and data analysis of skeletal muscle
Contents
1. an unknown amount Ad ditionally acquisition must be executed at the highest priority to prevent the system from leaving the job once initiated Finally because the highest priority tasks are interrupt service routines the entire data acquisition routine is entered via an interrupt generated by the system real time clock Data Translations Model 2580 Thus data acquisition proceeds to completion without inter ruption In order to achieve the desired data acquisition rates 3000 Hz for twitch contractions 300 Hz for tetanic contractions in the multi user environ ment the acquisition routine incorporates software counting loops which delay acquisition by a fixed amount Following acquisition of a given data point the counting loop delays operation for a fixed period of time before resuming acquisition The relationship between data acquisition rate and number of counting loops was determined using the acquisition program and a digital interval timer Tektronix Model 504 The LSI 11 23 proces sor was running under the TSX operating sys tem and acquisition time was measured as a func tion of delay value Because the results are ex tremely linear the data were subjected to linear regression analysis and an equation derived to relate desired acquisition speed to number of counting loops Table 1 This relationship was independent of operating system load thus assur ing real time data acquisition under any condition The acqu2. Computer Methods and Programs in Biomedicine 22 1986 259 265 Elsevier CPB 00775 259 Real time acquisition and data analysis of skeletal muscle contraction in a multi user environment Richard L Lieber Dean E Smith R Clark Campbell and Alan R Hargens Division of Orthopaedics and Rehabilitation Department of Surgery V 151 Veterans Administration Medical Center and University of California 3350 La Jolla Village Drive San Diego CA 92161 U S A A data acquisition system is described which acquires data from contracting skeletal muscle The system is designed to run in a multi user environment while acquiring contractile data in real time Time dedicated solely to laboratory experiments is thus eliminated A menu driver is included to allow users to enter experimental commands with or without command arguments Error monitoring functions prevent operator errors from causing data loss Data storage in both ASCII and binary formats maximizes file flexibility readability and accessibility Finally an on line tutorial and help facility is provided for user training The system developed is applicable to any experimental environment involving data acquisition storage and analysis Data acquisition Muscle contraction Time sharing 1 Introduction Recently the number and type of laboratory ap plications involving computers have greatly in creased This trend reflects the increased availabil ity of low cost mini and microc
3. IAL Assigns a file name to the experi ment and queries the user for experimental details animal type muscle type animal identification and experiment type Calibrates tension transducer by either directly entering calibration 2 CALIB factors or sequentially placing weights on the transducer and al lowing the program to perform lin ear regression on the data Calibra tion factors include a slope trans ducer sensitivity Y intercept transducer offset and correlation coefficient indicator of linearity 3 TWITCH Acquisition commands to trigger or stimulator and acquire the force re TETNUS cord 4 STORE Stores the most recently acquired data on disk Commands 3 and 4 are repeated according to experimental requirements Acquisition command use is monitored by variations which provide error handling 3 3 2 Error handling Because many of the acquired data are unique i e the same information cannot be obtained twice from the same muscle due to muscle fatigue it is important to minimize operator errors which cause data loss Thus the following two logical variables were implemented CALIB Indicates that the transducer is calibrated MEMFUL Indicates that acquired data are in computer memory The benefit of the CALIB logical variable is that data acquisition cannot proceed until calibra tion factors are entered Thus uncalibrated force values are never acquired The MEMFUL variable restricts fu
4. ction ANALYZE Perform analysis on data in memory BMDP Sort entire data file for statistical analysis CALIB Calibrate tension transducer CLOSE Close open data file CONVERT Convert data from ASCII to binary DISPLAY Display acquired data on oscilloscope EXIT Exit program FATIGUE Acquire data from muscle during fatigue test FILTER Filter data presently in memory FORCE Acquire and display baseline force HELP Provide general or specific help text INIT Define experimental parameters LOGICAL Report logical variable status and calibration factors OPEN Open data file for input PLOT Plot data on Tektronix terminal PRINT Print analysed data on printer PROCESS Sequentially read filter and analyse a data file READ Read a set of data from an open file into memory STORE Store acquired data on disk SYSTEM Execute operating system command TWITCH Acquire data from muscle during a twitch contraction TETANUS Acquire data from muscle during a tetanic contraction TUTOR Provide tutorial information on acquisition and analysis TEST Test oscilloscope digital to analog converter display 3 Program must be able to operate in a step by step or completely automated mode 4 Acquired data should be protected from acci dental deletion or repeated storage 5 Program should implement appropriate exist ing software packages 3 3 1 Data acquisition sequence The following sequence of commands performs data acquisition 1 INIT
5. ction force is sensed by the force transducer and fed through the analog to digital converter ADC into computer memory Transducer calibration occurs prior to data acquisition under computer control Acquired data are immediately displayed on a nonstorage oscilloscope using two DACs one for the horizontal axis and one for the vertical axis Data are stored on disk and later plotted on a Tektronix Model 4051 graphics termi nal Tektronix Beaverton OR from which the printed copy is obtained 3 Software development The software developed for the system falls into two general categories based on the two separate functions required 1 real time data acquisition display and 2 command and file handling Real time applications are first discussed followed by more general programming considerations 3 1 Real time data acquisition programs All acquisition programs were written in MACRO 11 assembly language in order to permit rapid real time acquisition and execution of the TSX assembly language instructions which are specific to real time programs In the multi user environment real time data acquisition requires certain special programming considerations For example under the TSX operating system the job in this case data acquisition must be locked in memory to prevent it from being removed from memory and temporarily placed on disk by the operating system scheduler These disk swaps delay acquisition by
6. e user environment it is preferable to drive data acquisition by a real time clock However the TSX operating system can handle less than 100 interrupts per second 12 thus precluding the use of real time clock inter rupts for high speed data acquisition in the multi user environment 3 2 Real time data display programs In order to provide feedback to the operator dur ing experiments the program must immediately display the acquired data Unfortunately most plotting packages are relatively large and slow resulting in significant data handling e g writing the data to a temporary disk file and delay before data can be viewed To circumvent this problem 262 an oscilloscope driver was written to output an array of acquired force data sequentially 300 points via the ADCs into X and Y channels of a nonstorage oscilloscope Tektronix Model 2215A Beaverton OR Data are thus immediately availa ble for viewing following acquisition and prior to storage In the initial implementation data display was performed in real time as with acquisition However since this locked out other users during data viewing for about 30 seconds we elected to permit time slicing of the display routine by the operating system As a result a slight amount of flicker in the displayed data is apparent during high system loads In order to provide the operator with im mediate quantitative information regarding the muscle force magnitude calib
7. er and experiment type are printed Similar information is stored on disk for subsequent statistical analysis step by step command execution A sample analy sis printout is shown in Fig 3 All subroutines are structured such that they can be executed directly by the operator e g to open a file for input using the OPEN command or indirectly by another subroutine e g PRO CESS uses OPEN during the processing sequence As a result of this structure keyboard commands are easily constructed which represent complex combinations of existing subroutines see PRO CESS and BMDP commands Table 2 3 3 5 File structure Raw data files were designed for flexibility as well as readability Thus all raw data are initially stored on disk in ASCII format The directory of the file contains the appropriate experimental ini tialization parameters the experiment date the experiment time and transducer calibration fac tors The remaining data are grouped into sets A set contains data from three muscle contractions Three sweeps per set were selected to provide repeated measures of the same conditions with minimal user intervention Therefore under FOR TRAN IV all of the initial data files are formatted sequential access The advantage of this structure is that the entire file can be viewed and if desired modified using the standard screen editor The analysis algorithms can also be checked or de bugged by creating the appro
8. erating system 12 The LSI 11 Q bus was selected for this application due to its powerful input output bus architecture which is well suited for data acquisition applications The 0169 2607 86 03 50 1986 Elsevier Science Publishers B V Biomedical Division 260 COMPUTER STIMULATOR LSI 11 23 ANALOG TO DIGITAL CONVERTER OSCILLOSCOPE DISPLAY DIGITAL TO ANALOG CONVERTER CONSOLE TERMINAL Fig 1 Schematic diagram of data acquisition system Data acquisition is initiated by the user from the console The computer then triggers the stimulator causing muscle contraction Real time data acquisition at the transducer then permits recording of contractile force via the analog to digital converter Acquired data can be immediately displayed on the oscilloscope via the digital to analog converter TSX operating system enables sharing of system resources during execution of low priority tasks see below Finally DEC equipment was selected because of the large amount of software support available for scientific applications 2 The hardware configuration of the system is typical and therefore only briefly described Fig 1 The user initiates data acquisition at the con sole by selecting the appropriate command TWITCH or TETNUS see below The computer then triggers the stimulator via the digital to ana log converter DAC Data Translations Model 2781 Data Translations Boston MA Muscle contra
9. ines created a user friendly system which easily proceeds from data acquisi tion analysis and plotting to statistical analysis The system is applicable to any environment acquiring data from a transducer The only user specific programs requiring development are the data acquisition and analysis subroutines 4 System performance The maximum data acquisition rate is 19 4 kHz This rate is presently limited by the relatively old Data Translations board used for acquisition The system has now been in use for about one year Generally during the experiments a secretary is word processing in addition to several students generating data plots or performing data analysis As mentioned above data acquisition speed is independent of any system load Lower priority tasks e g disk storage terminal input output are slightly affected by the number of users By comparison to our laboratory s previous method where data were stored on chart recorder data acquisition proceeds in approximately the same amount of time However the major ad vantage of the present implementation relates to data analysis Experimental data which may be gathered over several hours are analysed in less than 10 min as contrasted with the previous 4 6 hours Additionally raw data are already dig itized enabling rapid reliable graphic generation see Fig 2 Because the analysed data are stored 265 on disk data need never be entered manually for fur
10. isition rates were then confirmed by in troducing a sine wave of known frequency and using an analysis subroutine to calculate sine wave frequency Results were independent of system load or number of points acquired and demon strate acquisition speed accuracy within 0 3 of the input frequency as measured by the digital timer over the range 1 19 400 Hz 261 TABLE 1 Relationship between data acquisition time and software coun ter loop value These data were acquired using the data acquisi tion program and a digital interval timer See text for details Number of Acquisition Acquisition counting loops time 100 points rate integer ms Hz 10 6 10 16390 50 41 8 2392 100 86 1 1161 200 175 2 570 8 500 440 6 226 9 1000 883 3 113 2 1500 1325 75 43 2000 1769 56 53 2500 2210 45 23 3000 2653 37 69 3500 3096 32 31 4000 3538 28 26 4500 4015 25 12 Regression analysis of the above data yields the following results y 0 888 x 2 427 where y is the acquisition time 100 points in ms and x is the number of counting loops integral Equation slope 0 888 is the delay time in ms per counting loop Equation intercept 2 427 is the hypothetical acquisition time given zero count ing loops In order to achieve a desired acquisition rate of 3000 Hz a delay value of 40 would thus be used yielding an actual acquisition rate of 3022 Hz This relationship is highly linear r 0 999726 Note that in the singl
11. omputer hardware and an explosion in the amount and type of soft ware support available for these systems We have developed a number of real time data acquisition systems for studies of skeletal muscle contraction which rely on state of the art hardware implemen tations 5S 7 9 10 However many biological ap plications require the use of ordinary hardware in an environment maximizing data acquisition ef ficiency and enabling the inexperienced computer user to perform sophisticated biological experi ments This report presents a system designed to acquire contractile data from active skeletal muscle in a multi user environment using commonly Present Address Papyrus Systems 626 Cordova Place Davis CA 95616 U S A On line analysis available hardware The approach implements as much commercially available software as possible and generates original code only for that which is specific to our task The resulting program is ap plicable to virtually any data acquisition environ ment involving transducer calibration disk storage and file analysis Because the program executes all time critical operations at the highest priority several users may use system resources as they become available thus decreasing time dedicated to laboratory experimentation 2 Hardware development The data acquisition and analysis system de veloped is based on the Digital Equipment Cor poration DEC LSI 11 23 3 running under the TSX op
12. priate data file using the screen editor and then running the analysis subroutine The disadvantage of such a file struc ture is that the files are relatively large 19 kbytes of disk space for 6 sets of data and in order to access data sets at the end of the file the entire file must be sequentially read Even on a Winchester disk it can take up to 30 seconds to read one set of data if it is preceded by about ten others There fore the CONVERT subroutine was developed to convert files from formatted sequential access ASCII files to unformatted direct access binary files The binary files require less disk space 11 kbytes for 6 sets of data and any data set can be retrieved in less than 1 second independent of file length Using this dual file type both the safety and flexibility of the ASCII files were retained while providing the speed and compactness of the binary files 3 3 6 Software packages implemented Existing software packages which were imple mented in this program include BMDP Statistical analysis software 4 FILTER Recursive and nonrecursive digital filtering programs 13 PLOT 10 Graphic support programs for the Tektronic plotters 8 RTPACK Provides the ability to execute sys tem commands e g DIRECTORY SQUEEZE EDIT from within the muscle program 11 105LIB Screen handling subroutines for the VT100 family of terminals 2 The use of these programs in conjunction with the developed subrout
13. ration bars are dis played with the data trace The operator uses the oscilloscope variable gain control to align the calibration bars with the oscilloscope reticle en abling accurate estimation of force magnitude Calibration bars are generated by the display pro gram which uses the calibration factors see below to calculate the digital values corresponding to the calibration bar levels usually 0 and 500 grams 3 3 General program considerations The entire program is menu driven The main component of the menu driver is a character han dling subroutine 7 which parses the entered com mand string into its components In this way all commands may be executed with arguments for example OPEN FILE DAT or alone for exam ple OPEN with the subroutine prompting for the argument FILE DAT if it is missing This en ables each user to run the program at an ap propriate level of sophistication maximizing ef ficiency when possible Table 2 lists all of the available commands All program code not in volved with data acquisition or real time display is written in FORTRAN IV The following consider ations influenced the development of the general code 1 Program must have sufficient error handling ability to eliminate data loss and or erasure 2 Program must provide on line tutorial help in order to aid the inexperienced user TABLE 2 Commands available from the muscle program menu Command Brief description of fun
14. rom a file In this way following file opening the user may select the PROCESS command to sequen tially analyse an entire data file providing a printout and a disk file containing the analysed results By setting the PROCESS logical program calls and returns proceed without user supplied carriage returns which are normally required for 5 ANALYSE weneseecceseececceccccceccesscexceescexeceerccrecacccerscesssscs25s5s55 ANALYSES SUMMARY meeeceecccescncens wanes neeeacexmeawcuceeeseeccecerceee enesceeesescscssss Date of Analysis 6 20 65 Ties of Analysis 10 25 24 Filter Parameters Limits 0 000 to 0 990 Gibbs 50 Terms 5 Iterations 1 Type 2 Order 2 File name IN A10111 8IN Identification name Rabbit 14C Left Experiment Twitch Experiment Date 3 22 85 Set Number 1 Sweep Peak Time Peak Tension dP dt Half Relax Time Base Tension 1 31 4 msec 245 2 grams 11 35 g ms 62 0 msec 168 4 grams 2 30 2 msec 245 8 grams 15 51 g ms 62 9 msec 167 6 grams 3 29 1 msec 239 0 grams 16 51 g ms 61 0 msec 160 5 grams Experiment Tetanus Experiment Date 3 22 85 Set Number 2 Sweep Peak Time Peak Tension Max dp dt Base Tension 1 795 1 msec 533 8 grams 23 1 g msec 141 1 grams 2 856 7 msec 674 0 grams 22 7 g msec 124 6 grams 3 846 4 msec 831 7 grams 25 2 g msec 122 9 grams Fig 3 Sample printout of a portion of the analysis Note that filtering and experimental details are both provided Before each data set the set numb
15. rther data acquisition attempts before disk storage of acquired data If the user wishes to acquire more data without disk storage they must answer a warning message before continuing data acquisition Following data storage MEMFUL becomes false preventing further data storage without acquiring new data This prevents the storage of meaningless or repeated data The status of all program logical variables as well as trans ducer calibration factors is reported to the oper ator by executing the LOGICAL command An example of data acquired from an experi ment is presented in Fig 2 Note the excellent signal to noise ratio gt 20 especially at high forces These data represent the force generated by 263 1000g att z AARAU kan 500 ms 10 Hz 40 Hz re f mi f 1000g Sh eae CE A kian 500 ms 60 Hz 80 Hz 100 Hz ml AN 50 ms Twitch Fig 2 Sample of data acquired from a single muscle experi ment representing contraction force as a function of time In each case the stimulator was triggered by the computer caus ing the muscle to contract Stimulation parameters were varied from a single twitch lowest panel to stimulation at frequencies ranging from 10 to 100 Hz Note that the time base for the twitch differs from the time base for the tetanic contractions 50 ms vs 500 ms a muscle as a function of stimulation frequency A single twitch contraction is presented in the lower panel As the stimulation freq
16. s Soc 6 1980 1243 1248 8 PLOT 10 Advanced Graphing II Users Manual Tektronix Beaverton OR 1981 9 K P Roos R J Baskin R L Lieber J W Cline and P J Paolini Digital data acquisition and analysis of striated muscle diffraction patterns with a direct memory access microprocessor system Rev Sci Inst 51 1980 762 767 10 K P Roos and B A Lubell High resolution digital imag ing and mapping of heart muscle striation patterns Proc Digital Equip Comp User s Soc 6 1982 211 218 11 RTPack User s Manual Logicaid Limited Framingham MA 1985 12 TSX Plus System Manager s Guide S amp H Computer Systems Nashville TN 1981 13 R Walraven Digital Filters Proc Digital Equip Comput User s Soc 1 1980 827 883
17. ther computer aided statistical analysis Finally the redundancy built into the system for file iden tification and experimental data protection has resulted in a dramatic increase in data quality and reliability Further technical information and program list ings are available from the first author Acknowledgement This work was supported by the Veterans Admin istration and NIH Grants AM 25501 AM 26344 and AM 35192 References 1 R I Close Dynamic properties of mammalian skeletal muscles Physiol Rev 52 1972 129 197 2 1985 1986 U S Chapter DECUS Program Library Software Abstracts Digital Equipment Computer Users Society Digital Equipment Corporation Marlboro MA 1985 3 Microcomputer Interfaces Handbook Digital Equipment Corporation Marlboro MA 1983 4 W J Dixon BMDP Statistical Software University of California Press Los Angeles CA 1983 5 R L Lieber and B A Lubell Real time data acquisition of diffraction spectra from contracting skeletal muscle fibers Proc Digital Equip Comp User s Soc 9 1982 219 225 6 R L Lieber K P Roos B A Lubell J W Cline and R J Baskin High speed digital data acquisition of sarcomere lengths from isolated skeletal and cardiac muscle cells IEEE Trans Biomed Eng 30 1984 50 57 7 B A Lubell and K P Roos A study utilizing RT 11 and an optical CCD device for sarcomere measurement in single cardiac cells Proc Digital Equip Comp User
18. uency increases from 10 to 100 Hz muscle force increases due to tem poral summation of the stimulus input Note also that the single twitch contraction occurs on a faster time base than the tetanic contraction 3 3 3 On line help and tutorial Two types of on line help are available First a general tutorial demonstrates a typical sequence of commands required to acquire and analyse data during a muscle experiment Second command specific help provides the user with proper com mand syntax and an explanation of the various options and arguments used with each command In practice this assistance is used mainly by the novice and enables the development of program familiarity without assistance thus serving as a training tool 3 3 4 Data analysis sequence After data acquisition and storage the following commands are entered to analyse a set of data 1 CONVERT Converts the raw data file from ASCII to binary see below 264 2 OPEN Opens the binary data file for input 3 READ Reads in selected data set 4 FILTER Applies digital filtering al gorithm to acquired data op tional Analyses the data for the de sired parameters Analysis of subsequent sets occurs by repeating the command sequence 3 to 5 Repeated execution of this command sequence for files with multiple sets is tedious and therefore the PROCESS com mand was developed PROCESS sequentially reads filters 13 and analyses all data sets f
Download Pdf Manuals
Related Search