Announcement

Collapse
No announcement yet.

APAS Response

Collapse
This topic is closed.
X
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • APAS Response

    Dear Biomch-L Subscriber:
    I read with interest many of the statements and comments about the
    Ariel Performance Analysis System (APAS) which I developed. This system
    was initially developed in 1968 and, since then, has been subjected to
    continuous modification and improvement. All of you, the real ones and
    the Anonymous, recognize that the technological changes in the last
    decade are truly phenomenal. There are 1426 APAS systems currently
    serving more then 5000 people with applications in the medicine, sports,
    aerospace, NASA, universities, research institutes, and industry, from
    Israel to China from Australia to Bahrain and, of course, the United
    States. The people involved in this continuous development include
    scientists, Ph.ds with expertise in Nuclear Physics, Mathematics,
    Physiology, Biomechanics, and more, which enabled our products to remain
    on the cutting edge of technology. Utilization of the Internet and
    Multi-Media in the APAS system reinforces this statement.
    At my last reading, only 19 people had responded in the Biomch-L
    publication and 6 of those were Anonymous . I hope that there are more
    people out there reading and attempting to make decisions based on
    accuracy and facts rather than rumor and innuendoes. In addition, Brian
    Bergemann s comment regarding an Ariel Users bulletin board may
    facilitate interchange among scientists perhaps more rapidly than being
    limited to attending conferences.
    I seem to have a reputation, which may or may not be justified
    regarding my personality. However, whether I appeal to or antagonize
    people on a personal level is irrelevant with regards to the System and
    its performance. After all, most people do not purchase automobiles or
    television based on the personality of the inventor. Presumably, most
    people are primarily concerned about what makes a system accurate. The
    validity and reliability of any system is probably one of the most
    important factors for consideration. In order to address these factors,
    one must determine what are the essential processes that would make ANY
    Motion Analysis system accurate and reliable. My apologies to those of
    you who are already familiar with this information. It seems most
    appropriate, however, to discuss some of the building blocks of a
    System.
    In the most simple form, Biomechanical analyses are performed to
    determine the mechanical characteristics of some movement -- usually
    linear displacement, velocity, and acceleration. Data collection
    methods are generally video/film systems, tracking only systems (e.g.,
    Motion Analysis, Elite, Selspot, Vicon), and magnetic devices (e.g.
    flock of birds). Each of these types of systems has strengths and
    weaknesses and will be more or less appropriate depending on the people
    and the tasks to be performed.
    Since the APAS is primarily a video-based system, I will make a few
    comments regarding data collection for that type of biomechanical
    measurement system. Initially, the only medium available was high speed
    film. In the late 1960 s, we used a spring loaded camera, the famous
    Kodak cine special, at a rate of 64 frames per second film. In addition
    to the difficulty of trying to maintain the spring tension during the
    filmed event for each camera, there was the further delay associated
    with developing the film before the scientist even knew whether the
    films were good or useless. An additional, but necessary expense, were
    timing lights which marked the films so that camera speeds could be more
    accurately assessed during the digitizing process. Following this data
    collection, the next problem was accurately determining the coordinates
    of the different joint centers of the body. The film projectors
    employed register pins which were intended to advance the film
    discretely one frame at a time but were frequently inaccurate or slipped
    between frames which obviously introduced errors. Furthermore, the
    registration pins did not locate each frame of the film firmly in place
    thus creating some movement. In an effort to eliminate or reduce this
    false movement , 4 known, stationary points in the backgrounds were
    digitized to anchor each frame.
    By 1971, 3-Dimensional calculations of displacement, velocities and
    accelerations could be accomplished by setting the cameras at 90 degrees
    to each other, eliminating some of the degrees of freedom, and utilizing
    straight forward trigonometry.
    Identification, recording, and maintaining sequential information
    was yet another tedious process. A cumbersome process of locating each
    joint center and identifying it s numerical sequence on a large piece of
    paper, measuring the lengths and angles relative to the right
    horizontal, and manually punching the information on computer data cards
    was required for each camera view. A major advancement came with the
    introduction of a device which performed these steps faster and more
    accurately -- the Sonar Pen. Unfortunately, as with every silver-lined
    cloud comes the rain, there was some distortion due to the resolution of
    the microphones. This problem was resolved satisfactorily with the
    development of an in-house smoothing algorithm which was similar to the
    subsequently developed DLT. Direct Linear Transformation (DLT) was not
    introduced until 1971. Digital Filters of various kinds as well as
    different Polynomial techniques were developed for enhanced signal
    processing. At that time, in an area of 4x3x2 meters, we could minimize
    the error to approximately 1 cm.
    The purpose of this history is not to bore the younger readers with
    stories of the dinosaurs or to remind the other ancients what it was
    like in the good old days !! Rather it is to demonstrate an attitude
    of constantly trying to improve and the adjustments that adopting
    leading-edge technologies frequently require. The Sonar Pen was a
    significant leap in data processing but it was discovered in the
    medical school in Dartmouth College in order to trace brain tumors on
    X-rays. The first one had to be modified by the manufacturer for my use
    in Biomechanics but with its adoption the required adaptations in
    software followed. This process of new hardware followed by new
    software has never stopped nor will it. Another source of software
    enhancements have been generated by User needs which continues as an
    on-going process. Biomechanical analysis is not a fixed unchanging
    structure but rather a living, evolving process.
    The progression from film to video solved some problems and created
    others. It was a real benefit to have immediate confirmation whether
    the event had been satisfactorily captured. Also the cameras and tapes
    are less expensive than film cameras and associated supplies. As
    everyone knows, one of the problems with video is that there are fixed
    number of lines and these lines are interlaced. In other words, what
    you see really is half of the picture while the other half belongs to
    different time domain. Regular video has 60 fields (NTSC=60, PAL=50),
    but most of the currently available VCRs display only half of each field
    as a single Frame . This reduces the number of lines and, of course,
    the resolution. In other words, the limiting factor is the VCR since
    the pictures which appear on the screen are not advanced one field after
    another but rather EVERY OTHER FIELD. Unless a specialized VCR is used
    (the APAS utilizes this type of equipment), which allows each FIELD to
    be viewed, what you actually see on the screen with a regular VCR from a
    tape taken with a 60 field video camera are 30 frames. In other words,
    your camera filmed at 60 fields/sec but you viewed only half of the
    fields.
    Clearly video presents a big problem which must be resolved for all
    biomechanical applications regardless of the system selected for
    quantification. The limitations associated video are inherent to the
    technology; the solution(s) for coping with these difficulties are
    subsequently resolved by each company in their own way. The efficacy of
    the solutions, the scientific/mathematical appropriateness, in
    conduction with cost are probably the more important focus for the
    biomechanist. One solution is to use faster video systems. NAC has
    video cameras up to 1000 fields per second which are not interlaced thus
    providing an increased number of lines per field. Unfortunately, the
    cost for one camera exceeds $50,000.00!. Film based cameras up to 24000
    frames per second are available from Hi Cam and video cameras up to
    10000 fields per second can be purchased from Spin Physics. Since there
    were so many factors associated with camera selection, ranging from cost
    to application, I decided to develop an open system which could accept
    input from any camera. Because the system is independent of the camera,
    decisions appropriate for the user(s) and applications can be made by
    them. Therefore, purchasing a regular VHS camera may be economically
    acceptable but you must be willing to deal with the accompanying
    resolution. An S-VHS video camera will have more lines and higher
    resolution as far as the displayed picture is concerned but the price
    will be higher than for the regular VHS. Another option selected by Dr.
    Hans Gros in Stuttgart was to use two different cameras simultaneously,
    one is the NAC at up to 500 Fields per second and the other one a
    regular S-VHS 60 fields per second. The APAS algorithms, DLT and PPT,
    adjust for the differences in fields speed.
    Obviously there are a variety of considerations which also influence
    the choice of cameras. If the preponderance of studies involve high
    speed events, such as the time of impact of tennis balls on tennis
    rackets which vary between 3.8 to 4.1 ms., then the selected camera
    would have to achieve speeds of 10000 frames per second. For impacts of
    golf balls on golf clubs, camera speeds of 24000 frames per second would
    be necessary since the time of impact is less then one millisecond.
    However, for normal human activities in sports, gait, and normal
    locomotion, ordinary camera speeds of 30 frames per second are usually
    sufficient. For slightly faster activities, 60 and 120 fields per
    second should suffice. Due to these widely ranging needs and financial
    abilities of Users, there was really no other option than to create the
    APAS as an open system in order to accommodate all Users.
    Another difficulty associated with all cameras, including video,
    relates to camera speed. Regardless of the speed associated with each
    camera, e.g. 64, 500, 1000 f/s, etc., this rate does not occur
    instantaneously but must accelerate from zero to the selected speed
    choice. An LED device can insert time marks on the film which software
    can detect and calculate the variability in film speed. The APAS system
    does this computation. Video cameras present new and quite serious
    problems since they exhibit inertial constraints from start up until
    they reach full speed cameras as well as variability even when their
    maximum rate has been reached.. This variability has been demonstrated
    to be as great as 5 fields per 100 fields, that is a 5 percent
    variation. In other words, the camera might start at 59 f/s go to 62 f/s
    after one second and then go to 64 f/s and back to 58 f/s. Three video
    cameras operating simultaneously would varied non-linearly with each
    other. This is an inherent problem of hardware which cannot be
    corrected by using Gen Lock on the cameras!!! Utilization of the
    variable video tape from each camera will introduce additional error.
    Additional error can be produced if the proper VCR is not utilized.
    After struggling for a few years, I met an electronic genius who had
    designed one of the first Frame Grabber for video. He designed his
    board to capture the timing line or Retrace line . This line is
    equivalent to the old LED time line on the edge of high speed film. By
    grabbing both the picture and this line, the APAS software program could
    adjust for the discrepancy in the video. In order to perform this task,
    the video picture had to be captured (grabbed) and stored on the hard
    disk. Unfortunately, since one video picture required as much as 250 K
    of memory without compression, we had to develop a special algorithm to
    compress 250K to approximately 10K. This was possible since the
    background in most settings for biomechanics has little or no movement.
    Frame grabbing, then, is performed primarily to allow for time
    adjustment. If you want to increase accuracy, you must account for the
    discrepancy in the video recording. That does not means that you cannot
    digitize on the fly as in other system. The APAS system allow you to
    grab to the hard disk or digitize on the fly. This is well documented
    in our manual. Currently more then 10000 frames can be grabbed
    depending on the size of your hard disk. (hard disks sizes and types are
    optional). Although frame grabbing requires a little extra time
    (approximately 60 fields per minute to grab to the hard disk), it is
    optional. In other words, the User can choose which option is most
    appropriate for his/her needs.
    Some systems do not use the video picture but only track markers.
    This technique introduces is a different type of inaccuracies. When you
    track a marker, you are not necessarily following the movement of the
    joint center. Skin movement about the joint could account for more then
    10 percent error and even greater problems result when points cannot be
    detected because of limb rotations or one joint passing in front of
    another. Furthermore, assume that the data has been collected with a
    marker system and the experiment was concluded. If a need arises
    subsequently to trace another point, it cannot be done. Video always
    allows reexamination or redigitizing of the original data.
    Of course, the APAS can be utilized to track markers if they are
    applied to the subject. However, should there be an occasion in the
    future necessitating the reprocessing of the data at least the markers
    would be available since they were stored with the video picture. Other
    tracking systems currently available do not have this video restoration
    capability.
    After the data has been acquired and digitized, it is necessary to
    transform the data. Transformation is the process which takes the x,y
    coordinate information from each camera view and calculates the x,y,z or
    3-D coordinates. If only one camera is used, the User has the 2-D
    coordinates. The APAS provides several choices of transformation
    techniques including a regular or linear DLT and a Physical Parameter
    Transformation (PPT) which is a non linear DLT. (The mathematical
    details have been published by us and others. However, special credit
    is due to Herman Woltring, Grame Wood, and Jim Walton who were involved
    with us for many years in developing the best algorithm for the DLT.)
    In my opinion, one of the most important steps is the signal
    processing or the filtering which follows the transformation. All raw
    data is contaminated with noise from various sources (with only a few
    previously mentioned here) and, therefore, the signal must be smoothed
    in an effort to separate the noise from the signal. The APAS allow you
    to display the raw data at any time and to apply various filtering
    techniques such as Cubic and Quintic Splines, Various digital filtering
    techniques, as well as a Fourier Analysis. Again, many of these methods
    would not be available without the contribution of scientists such as
    Hatze, Woltering, Wood, Jennings, Walton, and others. It is also
    possible to combine various filtering techniques on the same curve. If
    one were analyzing a golf stroke, a digital filter can be applied to the
    portion of the curve where an impact occurred and a Quintic Spline used
    for the remainder of the curve.
    Obviously smoothing creates a new curve which has interpolated data
    on the time base chosen by the User. This procedure does NOT create new
    raw data. The raw data is always preserved; an additional or second
    smoothed version is merely created. However, if the investigator
    prefers to change the time base from that of the original data to 10 ms
    intervals, for example, this change can be made. However, it is always
    recommended to keep the time base as the raw data. For example, for 30
    frames per second it would be 33.333 ms per frame. For 60 Hz, it will
    be 17 ms per frame and so on. (One Anonymous comment was that the data
    was displayed at .010 rather then .016666. Reading the manual would
    have made it clearly obvious that this time can be chosen by the
    investigator. Of course, an old versions of the manual with a new
    versions of the software might not match exactly.)
    The APAS system was designed and has evolved with maximum
    flexibility. All data, raw as well as any smoothed data, can be dumped
    to various files type such as ASCII, WKS, DBS, etc. . This output
    possibility has been available for many years. Analog signals,
    including force platforms and EMG, can be used in conjuction with the
    video analysis or as independent, stand-alone systems. Again,
    flexibility was a major consideration since there are many different
    hardware options available. Therefore, raw data can be captured and
    stored in the computer from nearly any analog input device and
    thereafter, the signal can be processed. As more people continue to use
    these analog options, the software continues to increase and new
    features are added.
    Another relatively new feature is the ability to Pan the cameras.
    The software allows the investigator to use up to 9 cameras panned
    simultaneously. The Manuals and details for the panning is included in
    the documentation and available on FTP at:

    ftp.uci.edu at the \incoming\adi directory

    or at FTP at :

    k12.ucs.umass.edu at the \adi directory

    If you have any difficulties, you can contact the APAS User group set up
    and managed by Dr. Brian W. Bergemann at e-mail address:

    BWBERGEMANN@USSA-SPORT.USSA.EDU

    Since our software is continuously changing and improving, it is not
    unusually for the Manual to lag behind the software. However, our
    system provides on-line context sensitive Help screens. By pressing
    the question mark, you have an immediately explanation. Help screens
    are updated with each new revision of the software. The software
    updates sent to customers for free of charge and the new manuals are
    available on FTP or sent on a compressed diskette. The complaint
    expressed regarding free upgrades for years which should be obtained in
    writing is certainly a wish that everyone who utilizes computers and
    software shares. Only those who live beyond the real world could
    believe that the computer-based industries, including but not limited to
    such giants as Microsoft, Lotus, Borland, and most other software
    companies, operate under this edict. The fact is that most of the APAS
    customers have, in fact, received free upgrades. However, new software
    is often developed for newer hardware options on newer versions of DOS
    and it is obviously beyond any reasonable person s expectation that new
    hardware should be provided free of charge with upgraded software.
    I realize that many people do not know me and even some of them
    don t like me! Perhaps my looks, personality, or some other aspect is
    objectionable to some people. However, I have always tried to make and
    improve my System in an effort to create the best biomechanical tool. I
    originally did this to enhance and facilitate my own research efforts
    and now I have additional wish list input from other scientists to
    incorporate into the system. Even my worst enemies will acknowledge
    that I am always working to add to, refine, and improve my analysis
    system. Neither I nor this concept are perfect. As I said before, this
    isn t something that reaches a state of completion rather, it is a
    growing, developing, evolving entity.

    Gideon Ariel, Ph.D.
    ------------------------------------------------------------------------
    ------------------------
    It is not the critic that counts, not the man who points out how the
    strong man stumbles or where the doer of deeds could have done them
    better. The credit belongs to the man who is actually in the arena,
    whose face is marred by dust, and sweat, and blood; who strives
    valiantly, who errs and comes short again and again because there is no
    effort without error and shortcomings; but who actually strives to do
    the deed, who knows the great devotion; who spends himself in a worthy
    cause, who at the best knows in the end the high achievement of triumph
    and who at worst, if he fails while daring greatly knows his place shall
    never be with those timid and cold souls who know neither victory or
    defeat.
    Theodore
    Roosevelt
    ------------------------------------------------------------------------
    ------------------------
Working...
X