Ανάδραση > Βιοανάδραση (Βιοανατροφοδότηση – Biofeedback) > Εφαρμογές βιοανατροφοδότησης (βιβλίο) > Κεφάλαιο 6 – SELECTIVE TRAINING OF THE VASTUS MEDIALIS AND VASTUS LATERALIS USING EMG BIOFEEDBACK


Αυτό είναι το “Κεφάλαιο 6” από το βιβλίο “Εφαρμογές Βιοανατροφοδότησης” και έχει τίτλο: “SELECTIVE TRAINING OF THE VASTUS MEDIALIS AND VASTUS LATERALIS USING EMG BIOFEEDBACK”. Η Α.Ρ.Α. βιβλιογραφική αναφορά του είναι: “Χρηστίδης, Δ.Α. (2001). Εφαρμογές Βιοανατροφοδότησης. Αθήνα: Έλλην.”

Εάν χρησιμοποιήσετε ως παραπομπή αυτό το άρθρο μου (ή κάποιο άλλο) σε κάποια εργασία σας ή δημοσίευση, σας παρακαλώ πολύ να με ενημερώσετε. Θα χαρώ πολύ εάν μου στείλετε την εργασία σας για ενημέρωση. Σας ευχαριστώ προκαταβολικά.


Abstract of a doctoral dissertation at the University of Miami
Dissertation supervised by professor Ray Winters

Twenty-one subjects suffering from various knee problems of a neuromuscular nature were randomly divided into a treatment (N=10) and a control (N=11) group. Their clinical picture presented with conditions such as patella dislocation and subluxation, patella alta, lateral tracking syndrome, chondromalacia and with symptoms such as pain, swelling, tenderness, VMO atrophy, locking, and limited ROM. In an attempt to investigate whether selective training of the vastus medialis(VM) and the vastus lateralis (VL) with the use of EMG biofeedback is possible, treatment group subjects were provided with EMG biofeedback, whereas control group subjects were placed in a “no-treatment-treatment” waiting condition. All subjects were evaluated during baseline, treatment and one month follow¬-up, on five exercise conditions that are characteristic for adversely influencing the patellofemoral mechanism: knee extensions, knee extensions with resistance, deep knee bends, deep knee bends with the affected leg only, and taking a step on a stool with the affected leg. The averages of the difference in the recorded EMG values of these two muscles served as indicators of change, and these data were treated statistically. Multivariate analyses of variance showed that the treatment effect was significant for the treatment group but there was no change in the control group. Consequently, the control group was provided with treatment as well, and multivariate analysis of variance testing revealed that, following treatment, subjects in this group improved as much as those in the original treatment group and that the treatment effect remained until the follow-up evaluation for all subjects. The clinical picture of all subjects improved markedly, closely following progress in their ability to selectively coordinate the activity of these two muscles.


Misalignment of the patella (kneecap) is the most common problem involving the knee in orthopedic practice (Runow, 1983; Fox, 1975; Hughston, 1968). The patella is part of the patellofemoral joint, whose other components are the femur (the bone of the thigh), and the tibia (the larger of the two lower leg bones). The patellofemoral joint, along with the muscles of the thigh (quadriceps muscle) and other muscles of the leg whose muscles connect to the patella, form the patellofemoral joint mechanism. (See Appendix I for figures of anatomical drawings; Appendix II for term definitions).

Problems occur when the patella acquires a tendency towards dislocation, i.e., when it begins to pull on the side and to ride outside the patellofemoral joint. In many of the disorders that result from patellar misalignment, the patella does not actually dislocate but it has a strong tendency towards dislocation. What prevents the actual dislocation is usually the lateral (or in some not so common cases, the medial) condyle of the femur whose anterior surface comes in direct contact with the patella; such conditions are commonly diagnosed as “patellar subluxation” or, “lateral tracking patellar syndrome” Thus, the term “dislocation” will be used to indicate the tendency of the patella to dislocate, whether or not a dislocation actually occurs.

The correct alignment of the patella is controlled by the forces produced by two of the heads of the quadriceps (thigh) muscle: the Vastus Medialis muscle (VM) and the Vastus Lateralis muscle (VL) (Runow, 1983; Bentley & Dowd, 1983; Larsen & Laurisen, 1982). The most common method of estimating the correct alignment of the patellofemoral mechanism is a measurement called the “Q” angle. The “Q” angle is formed by the intersection of a line extending down the quadriceps muscle to the patella, with a line extending along the ligament that connects the patella to the tibia (patellar ligament). This angle is measured with the leg at full extension (see Appendix I, fig. 2). Movement of the patella is influenced by the force and angle of the patellar ligament, as well as the relative force and angle of pull of the quadriceps muscle. If the vastus medialis muscle is weak, the forces pulling on the patella are uneven and the patella dislocates (Runow, 1983; Reynolds et al, 1983; Aglieti, Tnsall, & Cerulli, 1982). There is a natural tendency for the patella to pull outwardly and dislocate over the femur because the lines of pull of the patellar tendon and the quadriceps muscle produce a vector that pulls it in that direction. An increase in the magnitude of this vector is related to an increased “Q” angle or a weak VM. On the other hand, factors that inhibit dislocation are: a) the size of the “bulge” of the femur that forms part of the patellar housing (most anterior projection of the lateral femoral condyle), b) the tightness that the Vastus Medialis imposes on the patella (resistance of the medial retinaculum), c) the normal force of the patellofemoral joint which varies along the range of motion of the leg as the result of the degree of knee flexion and the resultant forces of the extensor muscles and d) the strength of pull of the VM fibers (medial pull) (Runow, 1983; Aglieti, Insall, & Cerulli, 1982). Thus the amount of resistance needed from the medial retinaculum and the amount of force exerted on the “bulge” of the femorally produced housing would both be lessened if the contractile forces of the VM are sufficient to pull the patella and center it correctly in the femoral groove. Therefore, the burden for the correct alignment of the extensor mechanism rests upon the correct and sufficient work of the VM (Runow, 1983; Reynolds et al., 1983).

Prevention of patellofemoral problems focuses on correcting the alignment of the patellofemoral mechanism. Traditional treatment for patellofemoral problems consists of providing the patient with corrective belts and braces that keep the patella tight during activities and long periods of inactivity, as well as exercises designed to strengthen the quadriceps muscle (Reynolds et al, 1983; Runow, 1983). Phylogenetically speaking, the VM is the newest and weakest of the four heads of the quadriceps muscle. In problematic situations it is the first one to atrophy and the last one to be rehabilitated (Reynolds et al., 1983; Runow, 1983), a fact that makes orthopedic surgery very desirable as an alternative approach in many cases. This involves a bilateral tendon release, where the length of the VM and the VL tendons are surgically adjusted for correction of the “Q” angle and the optimal alignment of the patella within the patellofemoral joint.

Proposed treatment and its rationale

EMG biofeedback has received a lot of attention in recent years in the field of neuromuscular rehabilitation. Proper application of this technique can lead to the recruitment of new motor fiber pools and increase the neuromotor firing of muscles, along with improving the contractile coordination of antagonistic muscle groups. The purpose of the present study was to determine if EMG biofeedback could be used to train patients to coordinate the firing activity of their VM and VL muscles selectively, and thereby to correct extensor mechanism malalignments and to treat patella dislocations non-invasively.

Several studies have investigated the EMG activity of the quadriceps muscle under a variety of circumstances (e. g., walking, isometric exercises, squatting) in both normal and afflicted knees (Reynolds et al, 1983; Basmajian, 1978; Mariani & Carouso, 1979; Wheatly & Jahnke, 1951; Hallen & Lindahl, 1967; Lieb & Perry, 1971; Hirschberg, 1958). In general, they provide the parameters of what the correct EMG activity and coordination of the different extensors should be during variable degrees of extension and variable loads of resistance, as well as pointing out that dislocations and limitations in range of motion are largely the result of decreased or maladaptive VM activity. Based on the availability of similar information and knowledge (normal and maladaptive EMG patterns), EMG biofeedback has been successfully implemented in the rehabilitation of various other neuromuscular disorders such as cerebral palsy, spinal cord injuries, CVA’s, and myofacial pain (Miller, 1985; Goldsmith, 1985; Wolf, 1979; Dohrmann & Laskin, 1978). EMG biofeedback seems to be superior over other therapeutic techniques in terms of interrupting maladaptive co contractive patterns among groups of antagonistic muscles and shaping new adaptive patterns of coordination among them (for related reviews on EMG applications with the upper extremity for example, see Ince, Leon, & Christidis, 1984; 1985). The results of this investigation will indicate whether EMG biofeedback techniques can be implemented in the treatment of disorders of patellar malalignment and, possibly, other related joint laxity problems of a neuromuscular nature.



Twenty-two subjects were selected among those seeking treatment for knee problems at the orthopedic clinic of Miami Children’s Hospital. These subjects were randomly divided into an experimental and a control group in the following manner: Upon referral for biofeedback treatment, a flip of a coin (heads = experimental; tails = control) randomly assigned the first subject into one of the two treatment groups. The next subject was automatically assigned to the other group. The very same procedure determined group membership for consecutive pairs of subjects. This randomization procedure was chosen on the merit of its simplicity and adaptability to the clinical environment of this investigation, since recruitment of the total patient population at the beginning of the investigation was rendered impossible. The follow-up data from one of the treatment group subjects could not be obtained because she moved to a new city: this subject’s data were not considered in the statistical analysis. The ages of the remaining subjects that completed the study (N = 21) ranged from ten to thirty-seven years (mean = 16; median = 15; SD = 6.124). The randomization process assigned eight females and three males to the control group; eight females and two males to the treatment group.

Experimental group: Subjects in this group went through three stages of treatment: BASELINE, TREATMENT, and FOLLOW-UP.

Control group: Subjects in the control group went through five stages of treatment: BASELINE, “NO-TREATMENT TREATMENT” with a CONTROL FOLLOW-UP, TREATMENT, and a second FOLLOW-UP.


The investigation was conducted at Miami Children’s Hospital in a collaborative effort between the department of Orthopedics and the Division of Psychiatry. Dr. W.F. King, Assistant Chief of the Department of Orthopedics, provided his expertise and clinical supervision during the investigation. Dr. King served as the main referral and screening source for the patient population. Subjects were screened for candidacy in the investigation according to the criteria described by Runow (1983). Basically, all patients presented with a patellofemoral mechanism problem of a neuromuscular nature (i.e., atrophied VMO or lateral patella tracking), and all those presenting with a mechanical problem of the patellofemoral joint, such as torn patellar meniscus or ligament, were excluded. Dr. W.W. Schlanger, Chief of Psychiatry coordinated and oversaw the ethical and administrative aspects of the investigation.

Following each subject’s assignment to group, he/she received the prescribed treatment blocks. The treatment blocks or stages were:

a. BASELINE: The first office visit was designated as the “baseline session” and no treatment was given to the patient. During BASELINE, EMG records of the activity of the VM and VL were collected during the following exercise conditions: a) KNEE EXTENSIONS (E), b) KNEE EXTENSIONS WITH RESISTANCE (R), c) KNEE BENDS ON BOTH LEGS (B), d) KNEE BENDS ON THE AFFECTED LEG ONLY(ALO), and e) TAKING A STEP ON A STOOL (STEP-UP) WITH THE AFFECTED LEG (S). Collectively, a, b, c, d, and e, will be referred to as ERBALOS. No patient received any feedback of his/her neuromuscular signal during this period. The patient’s medical history was also recorded during the baseline stage. Subjects were asked the date at which the condition begun, the number of dislocations per time, the history of pain, and the degree of life changes resulting from the patient’s knee condition. Table 1 summarizes this information for each patient.

b. TREATMENT: The design of the study called for a maximum of ten treatment sessions. Patients were encouraged to visit five times per week for a two week period, but since this was not realistic for many of them, their appointments were scheduled conveniently. During the treatment period, all subjects received the standard biofeedback protocol. If they managed to reach their goal earlier, their treatment was terminated. No subject was to receive more than ten treatment sessions. During the week that followed their termination from treatment, EMG records were collected in the absence of feedback. These records were designated as their TREATMENT evaluation, and were used to determine whether any changes had indeed occurred in the patient’s neurosignal patterns.

Each treatment session consisted of a “diagnostic” and a “training” portion. During the diagnostic portion, ERBALOS data were collected in a No-feedback condition, immediately following electrode placement. These No-feedback data guided the investigator’s interpretation of each patient’s session-to- session progress. Upon collection of the diagnostic data the training portion of the training session commenced. In-between treatment sessions, subjects were instructed and encouraged to practice what they were learning. Practice consisted of repeating the ERBALOS exercises on their own, while envisioning that they had the feedback screen in front of them. In addition of enhancing their performance in coordinating the firing of the two muscles, continuous practice was serving the purpose of strengthening the VM and increasing its muscle bulk.

c. FOLLOW-UP: Four weeks following their TREATMENT evaluation, follow-up data were collected in order to determine whether any changes that might have occurred during the TREATMENT block were retained by the patient. Again, the patient was not given any feedback at this time.

d. NO-TREATMENT TREATMENT: Following BASELINE, control group patients went through a one month period void of medical attention. This period of no special attention was termed “no-treatment” treatment, since, quite commonly, orthopedists use it as a form of conservative treatment for knee problems characteristic of this population. From a methodological perspective this was a waiting period, contrived to indicate trends, if any, of time maturational factors in the patient’s condition.

e. CONTROL FOLLOW-UP: The week following their (“no-treatment” treatment) waiting period, control group subjects were evaluated (CONTROL FOLLOW-UP) so that changes, if any, in their EMG records could be assessed. The comparison of their BASELINE and CONTROL FOLLOW-UP data was designed to evaluate the relative efficacy of the traditional conservative (no treatment) treatment method that prescribes rest periods and inactivity, as well as to tease out time maturational factors in the EMG activity of these subjects. Having completed the CONTROL FOLLOW-UP stage of evaluation, control group subjects went through TREATMENT and FOLLOW-UP as described above for the treatment group.Table 1
Age, Sex, and Symptom Check List before treatment

Subject’s Consent and Compliance

Subjects signed a consent form advising them of the nature, risk, and putative benefits of the study and stating that they could withdraw from the study at any time. See Appendix III.


The focus of this investigation was to determine whether patients suffering various knee problems of a neuromuscular nature would be able to alter the neurosignal pattern combination of their VM and VL. Electromyographic records of the activity of these two muscles were considered to be the only accurate indicators of whether such training had indeed occurred. Thus, EMG records were the only measurement to be analyzed statistically.
The integrated EMG for each muscle during each ERBALOS exercise was averaged over half second periods and the VL EMG average was subtracted from that of the VM. This difference is indicative of the strength of deviation of the two muscles’ neurosignal activity. Negative numbers imply higher VL values and positive ones signify higher VM values.
For each patient, the average of the differences during each of the ERBALOS activities was used for each of the blocks. In order to get a more consistent indication of EMG values and to eliminate possible stray scores, the average of two of these average differences during each activity was used in the final analysis of data.


The EMG biofeedback instrumentation used in this investigation consisted of the following basic elements: one analog input (AI) section, one analog to digital conversion (ADC) section, one Apple II computer, two disc drives, two video monitors, one complex sound generator, one audio amplifier, and one hard copy printer. The system was designed to deliver feedback of up to four channels of EMG simultaneously, with a minimum amount of delay and noise. Dedicated components and software processed each channel of EMG independently throughout the operation. Integration and rectification was achieved in software rather than in hardware.

The analog input section, of each channel, received EMG signal from the electrodes through a long shielded cable. The EMG passed through two stages of amplification. In the first stage an (284J Analog Devices Inc.) isolation amplifier was used, providing for low input noise and protection from ground fault currents. In addition, the gain of the 284J isolation amplifier was software programmable; the utility of this characteristic will be explained later. During the second stage, the signal was fed into an operational amplifier (National Semiconductor LM324) for further amplification. Then the amplified signal was fed into a “True rms-to-dc Converter” (AD536, Analog Devices, Inc.) that delivered a dc voltage between 0-10 volts, which is the equivalent of the true root-mean-square level of the complex ac input wave. At this point the signal was ready to enter the ADC section, described below.
The ADC section’s basic components were a multiple-channel 12 bit data acquisition system employing a (AD363) D/A converter, and a (INTIL 8255A) programmable peripheral interface adaptor. The basic functions performed by the ADC section are real time integration of the EMG signal, and back and forth communication between the AI and ADC sections with the (6502) microprocessor of the Apple and its mother board. Software programming controls the 8255A peripheral interface adaptor (selection of proper PORT assignments for INPUT and OUTPUT modes of operation is accomplished by manipulation of control words and bit assignments), which in turn controls and commands the A/D converter. The integration of the raw EMG data is performed in software by the Apple II in the following way: The amplified signal (0-10 volts) undergoes a complete 12-bit successive approximation analog-to-digital conversion as it goes through the A/D converter. Then, a bipolar, 12-bit, two’s compliment binary value is made available to the peripheral interface port of the 8255A peripheral interface adaptor. This value is rectified as it is read in memory by a real time interrupt driven program that takes the absolute value and stores the information in a software accumulator. Then the integration begins through another interrupt driven process: An onboard real time clock provides interrupt request pulses to the 6502 microprocessor at equal intervals of 1/1028 (.00097) seconds. The interrupts occur in the “background”. In this manner, a data display program (written in the higher level program APPLESOFT) can be performing floating point computations and plotting the data, while the interrupt service integration routine is resident in another memory location. The instrumentation’s range is from 0 to 600 microvolts: The frequency response ranges between 0-1500 Hz at different gain settings that are software controlled, and the sensitivity is better than +/-.5 microvolt. Calibration is also software controlled, which allows the device to remain stable during shifts of atmospheric pressure and temperature. With a time constant set a 1/10 seconds, the integrated data are stored at an accumulator/counter (for each channel) that is reset every time the Apple reads them.

The Apple II computer occupies a central role in this instrumentation. With menu driven software it controls the executive commands that specify the gain, integration, selection of channels, and audio feedback generation parameters in the AI and ADC sections. Having those parameters set with menu operations, the Apple itself does not have to dedicate time during routine operations. The Apple plots one point per channel on the monitor screen every 1/10 seconds. The points are plotted on an X-Y axis configuration, where the Y-axis manifests the amplitude of the EMG (in microvolts), and the X-axis the passage of time (in 0-22.5 seconds per frame). To the human eye this point plotting is perceived as a continuous line whose height is directly proportional to the frequency and amplitude of the EMG signal. When the Apple is running a “two channel routine”, a high resolution graphics routine (HI-RES) splits the video screen in two halves: a lower and an upper halve. The upper halve is dedicated to the first channel of EMG, and the lower halve to the second. The X-Y axis configuration appears in each halve: When the “four channel routine” is in operation, the HI-RES plots four bars at the bottom of the screen that are dedicated to channels 1-4 respectively from left to right. These bars ascend and descend to heights that are proportional to the frequency and amplitude of the EMG for each channel.

Main menu commands select routines that set gain, audio feedback, and data acquisition parameters: There are four different gain setting that are software determined, based on the desired range of EMG that needs to be monitored. The EMG ranges are: 0-12%, 0-25%, 0-50%, and 0-100% of the total range. For example, in the case of a spinal cord injured patient who suffered significant neurofiber losses, the 0-12% range setting will provide great magnification of low levels of EMG activity with high resolution and low noise. When, through biofeedback, the patient is able to increase his motor-unit recruitment above that range, the next range setting will be selected. From now on, the 0-12% range setting will be called the “8 scale”, 0-25% the “15 scale”, 0-50% the “30 scale”, and 0-100% the “60 scale”, since the full 0-600 μv/sec range of the instrumentation is represented on the screen as 4-60 μv/sec and these values are stored and printed as such.

For the production of audio feedback, one (Texas Instruments SN76477N) complex sound generator is used per channel. The device receives input from a (Motorola MC1408L) digital-to-analog converter and it responds to EMG signals with audio signals. The circuit is designed to allow microprocessor control over frequency, pitch, noise, mixing, and envelop of generated sounds. Main menu subroutines provide access to three modes: an amplitude modulation (AM) mode that produces “clicks” whose frequency is proportional to the EMG integral, a frequency modulation (FM) mode that generates “tones” whose (voltage controlled oscillating) frequency and pitch vary proportionally to the EMG integral, and a “criterion mode” that generates a pure tone only when a pre-set threshold is met. In the “criterion mode” the threshold can be set anywhere within the 0-600 microvolt range in various combinations: channel 1 only, channel 2 only, channel 1-2, channel 2-1, channel (1+2)-(3+4), etc. The output of the complex sound generator feeds directly into the input of an audio amplifier (Realistic SA-102). The volume setting on the audio amplifier controls the sound that is delivered through a pair of audio speakers (Sansui SP-50).

Data acquisition software allowed twenty-two and one half second frames to be dumped from memory onto a floppy disc while the system was in the operating mode. At the end of the treatment session a “print routine” retrieved the data from the disk and printed them on a hard copy printer. Each patient’s data were organized on the floppy disk by name, date, frame number, activity, muscle groups, and integration time.


The equipment’s arrangement within the space of the laboratory was such that measurements could be collected in the absence or presence of feedback. All the equipment was located at one side of the room, except for one of the monitors and the loudspeakers that were on the opposite side of the room, at eye level from the sitting position. Once the patient had been prepared, he/she was always facing towards the side of the room with the single monitor and the loud speakers, thus, facing away from the rest of the instrumentation and the therapist. The therapist then, controlled the feedback presentation to the patient by switching the patient monitor’s signal on or off, when at the mean time he could continuously monitor the signal on his own monitor. All the data collected during BASELINE, SECOND-BASELINE, TREATMENT, and FOLLOW-UP, were collected in the total absence of feedback.

The BASELINE records of each patient guided the therapist’s decision making process for the direction of treatment. Typically, all of these patients presented, with a weak VM and an overpowering VL. In some patients the VM was markedly weak and atrophied while the VL was in the normal-to-strong range; in some others the VM was close to normal while the VL was unrestrainedly stronger. It was assumed that during any of the given tasks that the patients were asked to perform, the forces of both muscles add to a total of 100% of workload. When the force of the VM drops by 20%, that of the VL has to increase by 20% to compensate for the drop of the total 100%. Thus the goal of treatment for each patient was to increase the motoneuron activity of the VM to the maximum recruitment level possible, and then, to teach the patient to inhibit the activity of the VL to the extend that would supplement the activity of the VM towards the 100% of the assumed workload. It was assumed that this empirically derived formula would allow for overcompensation in the following manner: In this specific instrument configuration, at 100% of workload demand, each muscle would register 60 μv/sec. The sum of both muscles would be 120 μv/sec. During the ERBALOS activities though, which do not call for 100% of muscle use, the readings of normal subjects are about 40 μv/sec for the VM and 30 μv/sec for the VL, for a total of 80 μv/sec. If the initial readings of a patient were 20 μv/sec for the VM and 45 μv/sec for the VL, the goal would be to bring the VM at 60 μv/sec (40 plus [the lost 40-20=20]), and the VL at 20 μv/sec (80 total, minus 60 for the VM). It was further assumed that in due time the system would lean towards its own equilibrated level of coordination, once it was given the capacity to do so.

The electrodes were placed about two inches on each side of the belly of the VL and about half an inch apart of the belly of the Vastus Medialis Oblique (VMO) portion of the VM: In patients with severe atrophy of the VMO the electrodes were placed anatomically, at a 45 degree angle two inches above the horizontal midline of the patella. As much as conditions allow in such cases, these placements have remained constant throughout the duration of the experiment.


Table 2 shows the means and standard deviations for all five conditions during BASELINE, TREATMENT, FOLLOW-UP, and CONTROL FOLLOW-UP for both the control and the treatment group. These statistics are derived from eleven subjects in the control group and ten in the treatment, since one of the treatment group subjects was lost in FOLLOW-UP.

In order to examine whether there was a treatment effect and whether such a treatment effect persisted until the follow-up period, a series of Multivariate Analysis of Variance (MANOVA) tests were performed. Multivariate procedures, which permit the study of more than one dependent variable in an experimental design, are recommended for their economy, strict statistical control and generality of results in situations where a number of dependent variables need to be studied jointly (Tatsuoka, 1971; Winar, 1971; McCall, 1970; Keppel, 1982). Table 3 shows the significance level and degrees of freedom for all the MANOVA tests that are described below and were used to investigate the first question, that is, whether there was a treatment effect or not. All five of the tested exercise conditions (ERBALOS) are included in each one of the reported tests.

In order to ensure that MANOVA testing would be appropriate for these comparisons, it was necessary to examine first whether the means of both groups were drawn from the same population. The first MANOVA test compared the BASELINE for both groups. Failing to reach significance, F(5,15) = 2.82, P > .05, it indicated that the groups did not differ on their BASELINE measures.

The second MANOVA compared the pre and post-treatment measures of both groups. That is: the BASELINE and CONTROL FOLLOW-UP of the control group and the BASELINE and TREATMENT of the treatment group. This group-by-time interaction effect was significant F(5,15) = 28.19, P < .05, indicating that there was a treatment effect. In order to investigate the direction of this treatment effect, a third MANOVA compared the BASELINE and CONTROL FOLLOW-UP (pre and post measures) of the control group. The result was not significant, F(5,15) = 1.72, P> .05, indicating that there was no change during the waiting “No-Treatment treatment” period for these patients. This finding, in addition of implying that the significant effect of the previous comparison (second MANOVA) should be credited to the biofeedback treatment that the treatment group received, also points out that, for this particular population, the period of inactivity that is so commonly recommended as treatment, was ineffectual.

In order to validate the attribution of the treatment effect to the biofeedback intervention of the treatment group, a fourth MANOVA compared the post-treatment records of both groups (the treatment group’s TREATMENT and the control group’s CONTROL FOLLOW¬-UP). The results of this comparison were significant, F(5,15) ¬= 29.55, P < .05, indicating that after being provided with treatment, the treatment group subjects were different from those of the control group (which of course, showed no change from their baseline). Following the authentication of the EMG biofeedback treatment’s therapeutic effect, the focus of the MANOVA testing shifted towards the investigation of the second question, that is whether the treatment effect persisted until the follow-up period. Table 3 shows the significance level and degrees of freedom for these MANOVAs. Again, all five of the tested exercise conditions (ERBALOS) are included in each one of the MANOVAs that are described below.

As it was explained in the “methods” section, subsequent to their CONTROL FOLLOW-UP data collection, control group subjects were also provided with EMG biofeedback treatment. The sixth MANOVA compared the treatment group’s BASELINE and TREATMENT to the control group’s BASELINE and TREATMENT in order to investigate whether the biofeedback treatment had a therapeutic effect in those subjects as well. In this comparison the group effect was not significant, F(5,15) = 1.91, P > .05; the group by time interaction effect was significant, F(5;15) = 4.22, P < .05; and the time effect was highly significant, F(5,15) = 63.81, P < .05.

This pattern indicated that besides group membership, all patients improved with treatment. To cross-validate this finding with stricter statistical criteria, another MANOVA compared the TREATMENT of both groups only. Not reaching significance, it showed that after treatment, both groups had the same amount of improvement: F(4,16) = 2.55, P > .05.

Since the two groups did not differ neither in their BASELINE nor in their TREATMENT scores, the records of all twenty-one subjects were pooled together. With a repeated measures design, the eighth MANOVA addressed the question of whether the treatment effect was retained, by comparing the BASELINE, TREATMENT, and FOLLOW-UP of both the treatment and control groups. The outcome was significant, F(10,11) = 19.35, P < .05, and since all the univariate F-tests were significant as well, it indicated that the treatment effect remained until the follow-up period. The univariate tests were: for treatment extension F(1,20) = 141.04, P < .05; for treatment extension with resistance F(1,20) = 105.98, P < .05; for treatment knee bend F(1,20) = 148.05, P < .05; for treatment knee bend affected leg only F(1,20) = 107.61, P < .05; for treatment step-up F(1,20) = 118.94, P < .05; for follow-up extension F(1,20) = 13.31, P < .O5; for follow-up extension with resistance F(1,20) = 8.24, P < .O5; for follow-up knee bend F(1,20) = 10.45, P < .05; for follow-up knee bend affected leg only F(1,20) = 13.00, P < .05; and for follow-up step-up F(1,20) = 5.74, P < .05.

Finally, in order to delve into the extent of which the treatment effect was retained until FOLLOW-UP, a ninth MANOVA compared the TREATMENT and FOLLOW-UP for both groups. The outcome was significant, F(5,16) = 7.20, P < .05, and the higher means for TREATMENT indicate that there was a drop during FOLLOW-UP. Putative reasons for this drop are presented in the “discussion” section.


As it was stated in CHAPTER ONE, the purpose of this study was to investigate whether EMG biofeedback could train patients to coordinate the firing of their VM and VL muscles, selectively. The results of this investigation show that all of the treated subjects were able to selectively increase the motoneuron firing of their VM while decreasing that of the VL. Furthermore, they were able to maintain their ability to fire these muscles selectively for at least one month following their treatment. This being the case, did this new pattern of muscle firing bring any changes in their symptoms and reduce their suffering? The implied treatment goal was to “tailor” the firing patterns of these two muscles according to each patient’s needs, so that extensor mechanism malalignments could be corrected and the dislocating or subluxating patella could be treated non-invasively. This “tailoring” process involved the strengthening of the VM, and specifically its VMO portion, by increasing the amount of its motoneuron firing by recruitment, while at the mean time, weakening the force of the VL by inhibition. The overall result was that of adjusting the amount of resistance needed from the medial retinaculum (b: the degree of tightness of the tendon of the VM) and reducing the amount of force exerted on the “bulge” of the femorally produced lateral housing (a: femoral condyle). This is very similar to the outcome of surgery or arthroscopy where a lateral release weakens the pull of the VL and, by implication, grant strength to the VM.

Since there is no readily available method of evaluating these anatomical changes, indirect measures, such as the registered changes in the reported symptoms of these subjects were used. Such changes began to occur very early in their treatment stage when, for example, pain during knee bends, step-ups and other painful activities showed marked improvement as the coordination of the two muscles was heading towards the goal direction in their recorded EMG’s. Soon, muscle-fatigue pain started to replace the “bone¬ against-bone” pain of these patients, as recruitment processes in the VM began to activate new, previously unused, motor units and increase its muscle bulk. In due time, when the VM was sufficiently strong, the muscle-fatigue pain itself diminished, even during conditions of prolonged exercises. None of the nine patients whose patella was dislocating before treatment (see Table 6) experienced a dislocation from TREATMENT to FOLLOW-UP. None of the twenty-one treated patients presented with signs of tenderness, swelling, locking or limitations in ROM. Sixteen subjects were totally pain free during FOLLOW-UP, and the remaining five reported their pain as “occasional”, “nothing like it used to be”, or “nothing that I can’t handle or stop me from doing things”. In addition, one of these five subjects (subject #11) reported a “giving away” feeling about twice a week. Table 5 graphically depicts the movement of the means from negative to positive through the course of treatment, as the newly strengthened VM took over the inhibited VL during the five ERBALOS activities. Table 6 lists symptom presence for all twenty-one subjects during BASELINE, TREATMENT, and FOLLOW-UP. Since, in Table 6, there are a total of nine symptoms that each subject could possibly have, for the twenty-one subjects an area of(21 * 9 =) 199 boxes is created. From these, a total of 124 were checked during BASELINE which translates into a symptom presence of 62.3%. During both TREATMENT and FOLLOW-UP only six of these boxes were checked, which translates into a symptom presence of 3.1%. In other words, during TREATMENT and FOLLOW-UP, these subjects presented with only 4.8% of their originally reported symptoms, meaning that there was a 257.3% improvement! Applying more conventional statistical tools in this comparison, it shows that during BASELINE each subject presented with an average of 5.9 (SD=1.48, N=21, Median=6) symptoms, whereas during TREATMENT and FOLLOW-UP each subject presented with an average of 0.29 (SD=0.56, N=21, Median=0) symptoms. A paired t-test indicates a significant difference (T=23.05, df=20, p> .O1) between BASELINE and TREATMENT to FOLLOW-UP symptom presence.

These follow-up data for the treated subjects compared quite discrepantly to those collected at CONTROL FOLLOW-UP for subjects that went through the “no-treatment” treatment. During their CONTROL FOLLOW-UP evaluation, five of the “no-treatment” treatment subjects reported their overall condition as worsened, three as more or less the same, and three as slightly improved. Improvement for these last three subjects was reported as reduction in swelling and tenderness, some pain relief, but no changes in their refraining from such activities as sports, prolonged periods of walking, and others. Even these five improved subjects reported that their condition demanded further medical attention and treatment. Quite clearly then; improvements in each subject’s symptomatology is closely related with changes in the motoneuron firing coordination of the two muscles, as manifested by their EMG recordings.

The observation that there was a drop-in the FOLLOW-UP from the TREATMENT scores of these subjects, allows for at least three independent interpretations. The first, and probably less likely one, is that this drop is due to lesser values being recorded from the VM due to increases of its muscle bulk following treatment. Since the electrode placement on the VMO portion of the VM has remained constant (half an inch around the muscle belly), putative increases in its bulk would have restricted the electrode “pick-up” area, and thus, give smaller readings. In addition, increases in muscle bulk imply increases in muscle strength, and according to the principles of recruitment, demand for additional motoneuron recruitment reduces when the workload itself lessens. The second interpretation is that the assumption that the system will impose its own equilibrium on the coordination of the two muscles is correct. In this case, having increased the motoneuron firing of the VM excessively during treatment, in time due, the neuromuscular system adjusted itself pertinently, so that the patella could ride correctly within the joint. The third interpretation is that the patients’ condition begun to worsen having being cut-off from feedback for that time period. As plausible as this third interpretation seems, it might not really be the case with these patients, since their clinical picture improved further from TREATMENT to FOLLOW-UP. To shed some light as to which of the interpretation is more likely than the others, a longer follow-up period is required, so that a more intelligible clinical picture can be obtained.

The fact that these subjects continued to improve clinically, despite their observed drop of their scores from TREATMENT to FOLLOW-UP, warrants some additional attention. Presenting symptoms such as tenderness, swelling, and pain, are not only due to the overexerting strain and fatigue of the VM fibres as they endlessly attempt to counterbalance the existing malalignment, but to chondromalacia as well. Chondromalacia is the pre-arthritic-like, degenerative condition that is characterized by wear and softness of the cartilage. As the malaligned patella rubs forcefully against the lateral head of the femoral condyle, the protective cartilage of the two contacting surfaces begins to wear away and degenerative processes enter the histological picture. That area then gets “soft” and tender, and very frequently swells. Chondromalacia, being a reversible process, can be corrected and this is the objective aim of the various knee braces whose purpose is to keep the patella tight and away from the lateral head. Wearing a knee brace though does not always guarantee symptom relief, as manifested, in this study for example, by all eleven subjects who were previously fitted with braces. Instead, it might even worsen a patient’s condition by facilitating VM atrophy as it reduces its workload. For subjects in this study then, it seems that an additional benefit of treatment was halting and reversing their chondromalacia, a fact that is manifested by their continuing improvement and reduction of tenderness, swelling and pain, as the softened cartilage was allowed to heal.

APPENDIX A: Anatomical Figures 1-3   APPENDIX A: Anatomical Figures 4-7

APPENDIX B: Definitions and terminology

  • Chondromalacia: Softness of the cartilage.
  • Condylar angle: The angle between the femoral condyles in the condylar groove.
  • Dislocation of the patella: The temporary displacement of the patella from its normal position in the joint.
  • EMG: Electromyography, electromyographic, etc.
  • Patella: The kneecap: The lens-shaped and largest sesamoid bone (fibrocartilage in a tendon playing over a bony surface), situated in front of the knee within the tendon and the quadriceps femoris muscle.
  • Quadriceps muscle: The rectus femoris, vastus intermedius, vastus lateralis, vastus medialis, and vastus medialis oblique. (The vastus medialis oblique, a portion of the vastus medialis, is the one mostly responsible for the medial alignment of the patella).
  • Quadriceps tendon: The patella and the patellar ligament.
  • “Q” angle: The angle between the quadriceps muscle and the quadriceps tendon.
  • Retinaculum: In general use it refers to the membrane or the band that holds any organ or part in place. When it refers specifically to bones, it means the thickening of the deep facia in the distal portions of the limps which hold the tendons into position when the muscles contract and it is also called: retinaculum tendinum. In other words it is the part that bonds the tendons to the bone.

APPENDIX C: Consent Form


Dear …………………………………………………………………………………………………………….: you are invited to participate as one of the (approximately) twenty subjects for this clinical investigation. Since I believe that it is very important that you clearly understand a) what this study is about, and b) what your exact role is as a subject, I prepared this written summary for you. This summary is written following the guidelines of Miami Children’s Hospital regarding a research project. One of the requirements or these guidelines is that you and a witness sign the informed consent statement at the end. Please read it very carefully and make sure that you understand all of the information that is given to you regarding the treatment and the rest of the procedures to be followed. After you finish reading it, if you still have any questions, or if there is something that you do not fully understand, please feel free to ask me.
Your participation is voluntary. Refusal to participate will involve no penalty or loss of benefits to which you are otherwise entitled, and you may discontinue participation at any time without penalty or loss of benefits to which you are otherwise entitled.**********************************************************************************

As it is stated in the title above, this is a clinical investigation. The treatment that you are going to receive is EMG Biofeedback. The following two paragraphs will give you some understanding of what biofeedback is and how it is used. EMG stands for ElectroMyoGraphy and refers to the measurement of electrical activity of muscles.

Biofeedback involves the use of sensitive instrumentation (electronic or electromechanical devices) to mirror bodily processes of which the individual is normally unaware and which may be brought under voluntary control. The most important feature of the instrumentation is that it tells the individual about the measurement that it just made. There are many ways to communicate that information back to the individual. Body processes such as muscle activity, skin temperature, skin conductivity, heart rate, and blood pressure may be presented in the form of visual, auditory, or tactile signals.

Biofeedback can be described then as the treatment method in which immediate and precise information is provided to a patient about small variations of bodily processes that are not ordinarily perceived, and based on that information, the patient can, in turn, alter these bodily processes in desirable ways. As more refined methods and scientifically documented procedures are being developed, biofeedback is becoming the treatment of choice for many physical problems. The underlying principle of biofeedback is learning, and learning is the direct consequence of the enormous amount of plasticity that our nervous system is capable of.

Hoping that you understand by now what biofeedback is all about, I will go on to explain how it is to be used in your specific situation. Two sets of special EMG electrodes will be placed on your skin, right above the two muscles that are important in keeping your kneecap in correct position. (Do not worry about pain. The only pain you will feel during the whole treatment will be at the end of each session when the electrodes are being pulled off. It is the same “pulling” feeling you get as you remove a band-aid or a piece of scotch-tape from your skin). These electrodes measure the messages that your brain sends to those two muscles when it wants them to move. Keep in mind that this is the only thing that these electrodes are doing. They do not give you any stimulation. From the electrodes, the signal travels via special cables to five computers, one of which is an Apple. These computers analyse the signal and “feed it back” to you in a way that you can see and hear. By looking at those brain messages on a computer screen, I will teach you how to train your brain to work these muscles the correct way.

There are no risks, pain, or side-effects associated with this procedure. Even though it has never been used to treat knee problems before, it has been used to treat many other muscular problems. If you can learn how to correctly coordinate those two muscles, I anticipate that your knee problem may improve. To date, the best alternative treatment that you can receive is a surgical procedure called: arthroscopic lateral release, which means that the muscles will be corrected with surgery.

Next, I would like to explain to you how the experimental design is arranged. It is very important for things to be done this way, so the results can be accurately analysed and verified with statistical tools. As previously mentioned, you are one of (approximately) twenty subjects. These twenty subjects are going to be divided into two groups in the following way: As the subjects come in to be evaluated, I will flip a coin. If it turns out “heads”, that subject goes to group “A”. If it turns out “tails”, he/she goes to group “B”. The next subject that comes in automatically goes to the other group. The same procedure will be followed for the next pair of subjects until every one is assigned to either group “A” or “B”.

* Both group “A” and “B” will be evaluated the first day. Then, subjects of group “A” will be treated for the next two weeks: five sessions per week, each session lasting about one hour. Subjects of group “B” will not be treated during this two week treatment period.
* One month later, both groups will come in for a second evaluation. This way, the results of those that received the treatment will be compared to those that did not.
* Immediately after this second evaluation, subjects of group “B” will start being treated for a two week treatment period.
* One month later, group “B” will come in for a follow-up evaluation. These results will be compared to their first and second evaluations, as well as to those of group “A”.

Now that you know what to expect as a subject in this study, it is time to know what I, the investigator, expect from you:
* The treatment that you are about to receive involves learning, and learning is a very active process. This means that you are going to be a very important part of your own treatment: you will be very active rather than sitting back and receiving the treatment passively. You will not gain anything if you are not highly motivated. It is very important that you practice constantly what you will be learning during the two weeks of your treatment.
* Scheduling appointments is going to be very tight due to the nature of the study’s design. Please cooperate.
* Your records will be confidential per the guidelines established at Miami Children’s Hospital. You are encouraged to keep track of your progress by going through your records with me, and by keeping your own computer print-outs.********************************************************************************

* There will be no charge for your treatment for as long as you remain a subject in this clinical investigation. This fee waiver cannot be extended to services rendered to you at Miami Children’s Hospital outside of the Biofeedback and Behavioral Medicine Laboratory.
* After you have read carefully and understand the information that has been given to you in these three pages, please sign on the next page if you agree to participate.

I have discussed the above points with the subject or his/her legally authorized representative, using a translator if necessary. It is my opinion that the subject understands the risks, benefits, obligations, and provisions for confidentiality involved in participation in this project.



I have read this summary. The investigator who signed above discussed it with me.