WO2008152549A2

WO2008152549A2 - Device for functional assessment of a shoulder

Info

Publication number: WO2008152549A2
Application number: PCT/IB2008/052236
Authority: WO
Inventors: Kamiar Aminian; Brian Coley; Alain Farron; Brigitte Jolles-Haeberlin
Original assignee: Laboratory Of Movement Analysis And Measurement
Priority date: 2007-06-13
Filing date: 2008-06-06
Publication date: 2008-12-18
Also published as: WO2008152549A3

Abstract

The present invention relates to a device for a functional assessment of a shoulder comprising three gyroscopes, three accelerometers, all designed to be attached to the humerus of a subject and for measuring the anterior elevation- extension (pitch), abduction-adduction (yaw) and internal-external rotation (roll) of the shoulder; and a processing unit at least suitable to process the received data of the gyroscopes to estimate the 3D range of angular velocity according to the equation (I). Further the invention also relates to a method for functionally assessing a shoulder by measuring and scoring angular velocities of a humerus, resulting in a score (RAV), angular velocities and accelerations of the humerus, resulting in a score (P); and the sum of moments of the humerus, resulting in a score (M).

Description

Device for functional assessment of a shoulder

The present invention relates to a functional assessment of a shoulder and to a method for the functional assessment of the shoulder.

The functional assessment of body joints is known in different areas. For example has the importance of recognizing the result of a medical procedure long been recognized in surgery and particularly in orthopaedic surgery.

Outcome assessment has been given new impetus during the past decade as the emphasis has shifted from the era of expansion and technical development to one of assessment and accountability. There are over twenty different assessment methods for judging the functional outcomes of shoulder procedures (see e.g. Kirkley A, Griffin S, Dainty K. Scoring systems for the functional Assessment of the shoulder. Arthro and ReI Surg 2003; Vol.19. 10: 1109-1120). Some of these assessment methods, such as the Disabilities of the Arm and Shoulder score (DASH) (see: Jester A, Harth A, Wind G, Germann G, Sauerbier M. Disabilities of the arm, shoulder and hand (DASH) questionnaire: determining functional activity profiles in patients with upper extremity disorders. Jour Hand surg (British and European Volume, 2005) 2004; 30B: l : 23-28) and the Simple Shoulder Test score (SST) (see: Lippitt SB, Harryman DT , Matsen FA . A practical tool for evaluating function : The Simple Shoulder Test. In : Matsen FA, Fu FH, Hawkins RJ, eds. The shoulder: A balance of mobilty and stability. Rosemont, IL: Am Acad of Ortho Surg 1992; 501-518. ), are widely used, though none has been accepted as the universal standard.

These instruments assign a score to the patient using questionnaires based on separate domains: pain, function and overall satisfaction during special movements and/or positioning. Albeit validated, these instruments give only subjective scores and therefore give an incomplete answer on patient's shoulder evaluation.

Objective assessments like radiographs provide a static estimation of the range of movement of the shoulder girdle but do not measure its dynamic functionality. Although laboratory measurements like video-based motion analysis provide a complete 3D kinematics of the shoulder, they require a dedicated laboratory and assume that data measured from a short period are representative of usual performance. This constraint beside the cost of these technologies and the time needed for the analysis has restricted the use of these technologies in clinical practice.

It is thus an objective of the present invention to provide efficient and objective alternatives to functionally assess the shoulder.

Accordingly the invention relates to a device for a functional assessment of the shoulder which comprises at least three gryoscopes and at least three accelerometers, of which all are designed to be attached to the humerus of a subject. The gryoscopes and accelerometers are designed to measure the anterior elevation-extension (pitch), abduction-adduction (yaw) and internal- external rotation (roll) of the shoulder. Further to that the device according to the invention includes a processing unit which is at least suitable to process the received data of the gryoscopes to estimate the 3D range of angular velocity according to the equation

According to the invention a different approach is described. 3D kinematics during movements from body fixed sensors using ambulatory recording device are measured.

With such a measurement objective parameters (scores) for the assessment of shoulder function or movements, respectively could be achieved. This was achieved with the help of body fixed inertial sensors. The resulting parameters are evaluated to quantify for example the difference of kinematics between a healthy and a painful shoulder. The parameters could also be used to define the mobility of the shoulder in general and define some mechanical restrictions etc..

By an approach according to the invention the clinician could use a device to assess the shoulder's function and to find objective scores of their patients and which not include answering questions and therefore always a high subjective percentage.

Preferably the processing unit is further suitable to process the received data from the gryoscopes and accelerometers to observe the relationship between humerus acceleration and angular velocities according to the equation

Pr = V i— troll , pitch, yaw rang °ei acceleration) ^■

velocity ^J) ^/ ( ^\Eq -ⁱu.4) ^/ .

According to a particularly preferred approach of the present invention is the processing unit further suitable to process the received data from the gryoscopes and anthropometries data of the subject to observe the moments of the humerus according to the equation

→ Whereby I is the inertia matrix and (JQ the angular velocities.

Further to that the present invention relates to a method for functionally assessing a shoulder by measuring and scoring angular velocities of a humerus, resulting in a score (RAV), measuring and scoring angular velocities and accelerations of the humerus, resulting in a score (P) and by measuring and scoring the sum of moments of the humerus, resulting in a score (M).

Preferably the score RAV is a range of angular velocity and is calculated by the difference between the maximum and minimum of angular velocity, preferably in 3 dimensions, measured during one test, according to the equation

(V rangeiangular velocity))

RAVr = ^v-^lroll'^Pltch'^yaw 3 (Equ. l).

According to a particularly preferred approach of the present invention the score P is a product of the range of acceleration and the range of angular velocity, measured during one test, according to the equation Pr = ^ range(acceleration) ^■ range(angular velocity) (Equ.4).

Preferably the score M is based on the angular velocities of the humerus and anthropometrics data of the patient according to the equation

wherein I is the inertia matrix and Q) the angular velocities.

As already mentioned the use of the results according to the invention can be for different assessments. One possibility here is to compare the scores of healthy subjects to other according scores e.g. of pathological subjects.

Method according to any of the preceding claims 4 to 9, wherein the scores P, RAV and/or M are measured during at least two predetermined movements and the scores are an average of all according values of the tests.

Good results could be achieved if nine tests were conducted and the average of each value leads to the scores P, RAV and M.

Further to that according to a preferred embodiment of the invention the scores of left handed subjects are compared to right handed subjects. Or with the results it could be determined whether an object is more left- or right-handed.

If such an assessment is done it can be advantageous if the measurement is done during at least 1 hour, and the score P is estimated at least every second.

The invention will be described in more detail by way of preferred embodiments with reference to the drawings.

The drawings show in Fig. 1 positioning of the sensors module according to a preferred embodiment of the invention; 1 : Inertial sensor, 2, 3: ultrasound microphones (markers), 4: humerus

Fig. 2 a flow chart of the angles estimation according to a preferred embodiment of the invention;

Fig. 3 humerus acceleration as a function of its angular velocity a) for a healthy shoulder b) for a pathological shoulder

Fig. 4 angles estimation compared to a reference system for a) Flexion, extension b) Abduction, adduction c) Internal, external rotation

Fig. 5 P parameter for a patient and a control subject; Fig. 6 Box plot for the P score (a), RAV score (b) and Mscore (c); and Fig. 7 humerus angles for another test according to a preferred embodiment of the invention.

According to a first described embodiment of the invention a functional assessment of the shoulder is presented. 3D accelerometers and gyroscopes, attached on the humerus were used to differentiate a healthy from a painful shoulder.

According to this embodiment objective parameters (scores) could be found for the assessment of shoulder function based on body fixed inertial sensors and evaluating the effectiveness of these parameters to quantify the difference of kinematics between a healthy and a painful shoulder. By validating such approach, it could be provided to the clinician a system to assess the shoulder's function and to find objective scores of their patients.

According to the preferred embodiment described hereafter, 10 healthy subjects (25.1 years old ± 4.1) and 10 patients with unilateral pathological shoulder (7 rotator cuff disease (7 rotator cuff repair) / 3 osteoarthritis (3 prosthetic shoulder arthroplasty) : 4 women, 6 men : 62.4 years old ± 10.4) were studied. Nine tests representing some movements of daily activity based on the Simple Shoulder Test were carried out for both shoulders (see table 1) before surgery, 3 and 6 months after surgery. These tests were also carried out twice with one year interval on the same healthy subjects. Each test lasted 20 seconds and was video filmed for further validation of the movements and estimation of the false movements.

Table 1 : Summary of the 9 tests carried out for painful and healthy shoulders. The subject is in standing position.

Tests Description

1 Rest position

2 Hand to the back

3 Hand behind the head

4 Object ahead

5 4kg in abduction

6 8kg along the body

7 Hand to the opposite shoulder

8 Change a bulb

9 Object on side (Elbow in 90°, ext/int. rotation)

Figure 1 is an illustration of a) the position of the inertial sensors module including 3D gyroscope and 3D accelerometer and b) of the position of the reference markers for abduction/adduction (yaw), flexion/elevation (pitch) rotation and c) the position of the reference markers for internal and external rotation (roll). The reference markers from the reference system were used for assessing our kinematic system.

As shown in Figure 1 in this study, one module 1 comprising three miniature capacitive gyroscopes (Analog device, ADXRS 250, ±400 deg/s) and three miniature accelerometers {Analog device, ADXL 210, ±5 g) were fixed by a patch on the humerus 4. This way, the sensors measured the anterior elevation- extension, abduction-adduction and internal-external rotation of the shoulder. The signal from the sensors was amplified and low-pass filtered (cutoff frequency: 17 Hz) to remove any electronic noise. The sensors and their conditioning electronics were packaged in a very small box (25x25x13 mm). All signals were digitized at 200Hz sampling rate and recorded by the data logger (Physilog®, BioAGM, CH) carried on the subject's waist.

The Simple Shoulder Test (SST) and the Disabilities of the Arm and Shoulder Score (DASH) were filled out by each subject to compare the results with the results of the described embodiment according to the invention. The SST consists of 12 questions with "yes or no" answer. DASH is a 30-item questionnaire designed to evaluate upper extremity-related symptoms and to measure functional status at the level of disability. The SST and DASH scores are both validated scores and patient-reported outcomes measures.

Angles Estimation

Internal and external rotational movements (roll), extension and anterior elevation movements (pitch) and abduction and adduction movements (yaw) were estimated from 3D accelerometers and 3D gyroscopes.

Figure 2 shows the flow chart of the 3D angles estimation. The 3D gryoscopes measure the angular velocity for flexion/elevation, int/ext rotation and adduction/abduction. The accelerometers measure the gravity component, and using this feature, it is possible to measure the segment orientation when it is motionless. Drift and DC components of the angular velocities were removed using wavelet transformation and considering the initial and final orientation of the segment based on the acceleration signals. The 3D angles were obtained after integration of the three angular velocities. The angles were estimated from the integral of angular velocity and by considering initial and final orientation from the accelerometers.

As reference system, a Zebris CMS-HS ultrasound-based motion measurement system was used. This system consists of three fixed sonic emitters which send out a burst of ultrasound, and receivers (microphones) placed on body segments. The time taken for the burst to reach each receiver is recorded. Using this delay, the distances between the receiver and each emitter can be calculated from the sound velocity. Knowing the distance from three emitters, the coordinates of the receiver placed on body segment can be computed by triangulation with an absolute accuracy better than 1.0 mm with a sampling rate of 100Hz. In this study, two ultrasound microphones (marker 2 and marker 3) were attached over the same segment of the humerus 4. Spatial marker positions (x, y, z) were recorded and used for calculation of humerus orientation angles. Synchronization between the reference and the Physilog systems was performed by electrical trigger. The angle data obtained by the body-fixed sensors were down sampled to 100Hz for comparison purpose. The flexion/extension and abduction/adduction angles of the humerus were estimated using the spatial coordinates of the microphone markers 2,3 on the humerus 4 (see Fig. Ib)). The internal/external rotation angles of the humerus 4 were estimated using the spatial coordinates of the microphone markers 2,3 on the radius (see Fig. Ic)). Basic movements like anterior flexion-extension, abduction, adduction and internal/external rotation were performed with our system and the reference system on 10 healthy subjects to assess the accuracy of our angles estimation method.

RAV Score Algorithm

The second investigation was to estimate the difference of kinematics between the healthy and the painful shoulder. It was based only on the angular velocities of the humerus 4. The 3D range of angular velocity (RAV) was calculated by the difference between the maximum and the minimum of angular velocity (deg/s) measured by 3D gyroscopes during each test in internal and external rotational

(roll), flexion/extension (pitch) and abduction/adduction (yaw) directions for each subject. The RAVr parameter was estimated as the average of the sum of the RAV in the three axis of rotation.

Hl_N)

The difference between a healthy and a painful shoulder (ZlRAVr) was expressed as the percentage of RAV of the healthy shoulder (ZlRAVr).

ΔRAVr= RAVheaithy - RAVpamfui / RAV_heaithy (Equ .2)

The RAV score is defined as the average of the ΔRAVr over all 9 tests. RAV score = l -

lOθ[%] (Equ.3)

P Score Algorithm

The main idea was to observe the relationship between humerus acceleration and angular velocities. Figure 3 shows the difference between the healthy and the painful side for one axis and a patient. In order to estimate the difference between both sides, for each test the surface inside the curve was calculated for both sides. The simplest estimation of this surface was to calculate the area of the rectangle, which circumscribes the curve corresponding to the product of the acceleration range by the angular velocity range.

Figure 3 shows in a) a trace representing the humerus acceleration vs. angular velocity for the healthy side of the patient; b) a trace representing the humerus acceleration vs. angular velocity for the painful side. The rectangle, which circumscribes the curve, corresponds to the product of the acceleration range by the angular velocity range (Pr).

Pr = ^ range(acceleration) ^■ range(angular velocity) (Equ.4)

Further the surface for each axis for both sides was calculated and added to obtain a parameter called Pr for a healthy and a painful side. By considering that the product of angular velocity and acceleration is related to power of movement, it was assume that P is a power dependent quantity. This parameter can also be considered as the control of the humerus velocity by its acceleration.

The difference between the Pr parameter of a healthy and a painful side relative by the healthy side was considered as ΔPr parameter.

ΔP_r=(P_healthy-Ppaιnful)/Phealthy (EqU .5)

The first score is defined as the average of the ΔPr over all 9 tests. P score = 1

(Equ.6)

Comparing to RAV where only angular velocities were used, P score used both the angular velocities and the accelerations of the humerus.

M Score Algorithm

The last step was to estimate the difference of moments M between the healthy and the painful shoulder; it was based on the angular velocities CO of the humerus and the anthropometries data of the patient.

M was defined as the moment of the humerus (Equ.7), I as the inertia matrix (Equ.8).

Ipitch 0 0

I = 0 Iroll 0 (Equ.8)

0 Iyaw

Using the mathematical definition of moment of inertia from Vaughan et al. (Vaughan C, Davis B, O'Connor J. Dynamics of human gait. Human kinetics Publishers 1992) and the anthropometries data of the patient (length of the humerus: L_h, circumference of the biceps: Q₁, masse of the humerus: m), the relationship of the moment of inertia about flexion/extension (Ipitch), the moment of inertia about abduction/adduction (Iyaw) and the moment of inertia about internal/external rotation (Iroll) can be derived (Equ.9). m ' ^{■ ■} (O-076 - CΪ + L;)

Ipitch = ^■

12 m Iroll = ^■ ^■ c: (Equ.9)

Iyaw = Ipitch This method was used to evaluate the difference between the healthy and the painful shoulder, calculating the maximum of the norm of the moment (noted by 1 1 1 1 ) during each test for each shoulder.

The difference between the healthy and the painful shoulder was expressed as the percentage of the moment of the healthy shoulder.

AMr = ^- _Jr (EqU- I l) max \\Mhealth\

The M score is defined as the average of the ΔMr over all 9 tests.

M score =

(Equ.12)

A subject with a total mobility of his/her shoulder will have a M score, a RAV score and P score of 100% and a patient without any mobility of his/her shoulder will have a M score, a RAV score and a P score of 0%.

Statistical analysis

The Wilcoxon matched pairs signed rank sum test was used as a non-parametric hypothesis test to show if there were significant differences (at a significance level 5%) between baseline vs. 3 months, and baseline vs. 6 months for 10 patients.

The Wilcoxon rank sum test was used as a non-parametric hypothesis test to show if there were significant differences between baseline vs. 10 control subjects, 3 months vs. 10 control subjects and 6 months vs. 10 control subjects. RESULTS

Angles estimation

Figure 4 shows the angles of the basics movements of the reference system Zebris and the inertial sensors a) for Flexion, extension; b) Abduction, adduction; c) Internal external rotation. The dashed line shows the reference system and the solid line the inertial sensors.

The proposed method gave an accurate estimation of shoulder angles. The results of all the tests are shown in table 2.

Flexion/Elevation Abduction/Adduction Rotation Int./Ext. Subject Error, deg Error, deg Error, deg

RMS mean SD RMS mean SD RMS mean SD

Subject 1 2.50 -0.45 2.47 0.9986 2.95 -2.20 1.97 0.9968 3.19 0.58 3.13 0.9983

Subject 2 5.64 -3.08 4.72 0.9936 3.83 3.34 1.88 0.9940 2.38 -0.95 2.19 0.9972

Subject 3 4.86 6.25 3.36 0.9888 5.53 -4.08 3.63 0.9994 5.72 -1.90 5.39 0.9865

Subject 4 7.49 6.48 7.29 0.9970 9.61 8.59 6.37 0.9653 8.04 -3.97 6.69 0.9491

Subject s 7.25 6.02 6.90 0.9945 5.21 2.54 3.63 0.9880 7.99 1.32 7.88 0.9829

Subject 6 7.17 4.40 5.16 0.9953 8.97 6.55 8.52 0.9863 6.25 -5.92 4.61 0.9657

Subject 7 6.59 4.42 5.01 0.9962 1.41 0.48 1.33 0.9993 3.71 -4.49 3.57 0.9739

Subject 8 8.66 2.95 7.16 0.9984 3.62 0.31 3.58 0.9976 5.82 2.25 3.37 0.9950

Subject 9 6.56 5.16 6.44 0.9975 7.80 7.98 5.55 0.9849 6.50 2.68 6.10 0.9971

SubjectlO 10.03 4.26 9.09 0.9989 1.12 0.09 1.10 0.9991 7.81 4.32 6.51 0.9933

Mean 6.68 3.64 5.76 0.9959 5.01 2.36 3.76 0.9911 5.74 -0.61 4.94 0.9839

Table 2: Comparison between humerus angles obtained inertial sensors and reference system for 10 subjects. The error represents the RMS, mean and SD of the difference between reference and our measuring device. V represents the Correlation Coefficient between the two measuring system. As can be seen from table 2, the results were very close to those of the reference system presenting a small average error in RMS (5.81°), mean (1.80°) and standard deviation (4.82°) of the difference signal, reflecting accurate and precise estimation respectively; and excellent correlation coefficient (0.99) values reflected highly linear response.

Figure 5 al) and bl) show the comparison of P parameters between a patient and a control subject for the nine tests realized. It can be observed that for the patient (Fig. 5 (al)) the P parameter is higher for the healthy side than the painful side for all tests. But for the healthy subject (Fig. 5 (bl)) the Pr parameter is approximately equal between the right and the left shoulder for each test. Table 3 shows the P score for a healthy subject. The P score for the healthy subjects ranged from 85% to 97% (mean : 92%), which is twice compared to patients before surgery (table 3,4).

Table 3 shows all the results in comparison with the baseline (before surgery). The Wilcoxon matched pairs signed rank sum test indicates that significant differences were found between the P score at baseline vs. the P score at 3 months and the P score at baseline vs. the P score at 6 months (p<0.05).

Patients 1 2 3 4 5 6 7 8 9 10 Wilcoxon Test

Rav Scorebaseline 42 80 69 70 66 5 50 64 84 59

Rav Score 3 mo nth 87 94 79 98 76 81 62 60 94 76 p=0.0039

Rav Score 6 mo nth 87 93 93 94 70 95 54 66 97 76 p=0.0020

P score baseline 28 75 57 62 48 3 36 38 67 48

P score 3month 70 74 82 91 67 61 42 39 88 59 p=0.0059

P score θmonth 76 67 98 93 58 97 33 39 87 69 p=0.0195

M score baseline 22 51 48 42 36 22 15 25 55 25

M score 3month 64 90 59 37 65 63 31 44 69 64 p=0.0041

M score θmonth 66 83 97 44 52 70 23 42 86 60 p=0.0020

Dash baseline 137 91 47 74 93 75 93 128 79 47

Dash 3month 137 101 34 49 80 74 115 78 50 65 NS

Dash θmonth 94 93 34 32 81 54 110 72 54 38 p=0.0273

SST baseline 0 7 9 5 1 5 1 1 4 6

SST 3month 0 3 11 11 6 6 1 3 5 2 NS

SST θmonth 5 4 11 10 6 9 1 3 7 10 p=0.0234

Table 3: DASH, SST, P score, RAV score and M score for patients before surgery (baseline) and 3, 6 months after surgery. NS indicates that no significant differences were found at 5%. The DASH (30 is "very good mobility" and 150 is "very bad mobility"), SST (0 is "very bad mobility" and 12 is "very good mobility")

The P score average was 46%, 67% and 72% respectively at baseline, 3 month and 6 month after surgery. Fig. 6(a) shows the improvement of the P score after surgery in comparison to the baseline values and the control subjects.

It was observed that there were significant differences between the P score at the baseline vs. the P score of the healthy subjects and the P score at 3 month vs. the P score of the healthy subjects, but no significant differences were found between the P score at 6 month vs. the P score of the healthy subjects (p=0.074). RAV Score

Figure 5 a2) b2) show the comparison of RAV parameters between a patient and a control subject for the nine. The RAV parameter is higher for the healthy side than the painful side for all tests (Fig. 5 (a2)). But for a healthy subject (Fig. 5 (b2)) the ΔRAV parameter is approximately similar between the right and the left shoulder for each test. The RAV score for healthy subject ranged from 87% to 99% (mean : 94%). While this score was in average 59% for patients preoperatively (tables 3, 4).

Subjects P score, % RAV score, % M score, %

1 91(7) 94(5) 91(2)

2 96(-12) 99(-14) 87(3)

3 93(-4) 98(-4) 88(3)

4 94(3) 98(-l) 82(2)

5 96(-3) 91(5) 97(-9)

6 93(-ll) 95(5) 86(12)

7 97(-13) 95(-8) 95(-15)

8 90(10) 96(1) 93(-3)

9 93(5) 93(6) 72(17)

10 98(-9) 96(-9) 89(5)

Mean Δ( 1-2) -2.7 -1.4 0.7

STD Δ( 1-2) 8.5 7.1 9.4

Table 4: DASH, SST, P Score, RAV Score and M Score for healthy subjects. For all the healthy subjects : the SST was 12 and the DASH was 30. In brackets: difference between the first measurement and the one year measurement (Δ(l-2)) .

Significant differences were found between the RAV score at baseline and the RAV score at 3months, as well as between the RAV score at baseline and the RAV score at 6 months (p<0.05). The average of RAV score was respectively 81% and 83% at 3 months and 6 months after surgery (Table 3). Figure 6(b) shows the improvement of RAV score after surgery in comparison to the baseline values and the control subjects.

The RAV score of the healthy subjects was significantly higher than the RAV score at baseline as well as the RAV score at 3 month, but significant differences were also found between the RAV score at 6 month and the RAV score of the healthy subjects (p=0.037).

M Score

Figure 5 a3) and b3) show the comparison of moment in Nm (Newton-meter) between a patient and a control subject for the nine. The moments are higher for the healthy side than the painful side for all tests (Fig. 5 (a3)); while the moments are similar between the right and the left shoulder for a healthy subject (Fig. 5 (b3)). The M score for a healthy subject ranged from 82% to 97% (mean : 88%), which is more than twice the average for the patients preoperatively (tables 3, 4).

The M score at baseline was significantly lower than the M score at 3 months as well as the M score at 6 months (p<0.05).

Table 3 shows all the results in comparison with the baseline. The M score average was respectively 59% and 62% at 3 months and 6 months after surgery. Fig. 6(c) shows the improvement of the M score after surgery in comparison to the baseline values and the control subjects.

It was observed that there were significant differences between the M score at the baseline vs. the M score of the healthy subjects and the M score at 3 month vs. the M score of the healthy subjects, but significant differences were also found between the M score at 6 month vs. the M score of the healthy subjects (P=O.009). Many investigations of shoulder outcome evaluation previously used the questionnaires on imposed movements. Kirkley et al. (Kirkley A, Griffin S, Dainty K. Scoring systems for the functional Assessment of the shoulder. Arthro and ReI Surg 2003; Vol.19. 10: 1109-1120) presented the differences between scoring systems for the functional assessment of the shoulder. They observed that many of the items may seem irrelevant to patients with specific conditions and none has been accepted as the universal standard. In some case, the patient could not understand the real meaning of the questions and could not answer or did a wrong answer. The DASH instrument is a questionnaire. It depends on subjective evaluation of the patients. In some case, the patient doesn't understand the questions or answers wrongly. It depends also of the psychological condition of the patient. Due to the dichotomous response option (yes or no), the SST instrument is likely to have poor sensitivity to differentiate between patients with varying severity of the same condition.

The outcome evaluation of shoulder surgery according to the method of the preferred embodiment was based on objectives scores derived from accurate 3D measurement (table 2) of shoulder kinematics on healthy and painful shoulder obtained during specific task. These scores concerns acceleration and angular velocity rather than angles' components. Though angles can be estimated accurately with our system, they have not shown pertinent changes between a healthy and a painful shoulder.

Figur 7 shows the 3D angles for a patient for the test n°2, where the subject moved his hand to the back. Shown are the humerus angles of the healthy and the painful side in flexion/elevation (pitch) (a), in internal /external rotation (roll)

(b) and in abduction/adduction (yaw) (c). The angular ranges are rather larger for the painful side in comparison of the healthy side for the abduction/adduction

(yaw) and flexion/extension (pitch) axis. This observation shows that the patient has a strategy to minimize the pain by accomplishing a longer path than normal for the painful shoulder to do the same movement. However this is not the case for all patients, since every patient has a different movement strategy to reduce the shoulder pain. Therefore, it was not possible to use the angle magnitude as an objective parameter to quantify the difference between a healthy and a painful shoulder. Figure 6 shows the comparison between baseline, 3, 6 months after surgery for the three scores. It is shown a box plot for the P score (a), the RAV score (b) and the M score (c). The boxes contain 50% of the results and lines represent the range. The dashed line show the limit for healthy subjects.

It can be observed with these scores that, for all the patients, the mobility increased significantly after surgery (table 3). In addition the scores are clearly distinct between a healthy subject and a painful patient at baseline without any overlapping of the confidence intervals (Fig.6).

Table 3 shows also the results of the Wilcoxon matched pairs signed rank sum test for the clinical scores (DASH, SST). It can be seen while kinematic scores showed significant differences between baseline and follow-up time (p<0.02), the clinical scores (DASH, SST) showed no significant differences between baseline and 3 months evaluation but the differences became significant at 6 months evaluation (p<0.03). These results suggested that our kinematics scores might be more sensitive to the functional changes than the clinical scores, and were able to express a kinematic improvement from the baseline even at 3 months after surgery.

It can be seen from the table 3 that the patient 7 had very bad clinical scores after the surgery. He had an inflammatory capsulite retractile after 6 months. The kinematics scores were also able to show this post-surgery complication because the patient had a lot of pain while performing some movements. The complication of the patient 8 was a chronicle luxation. His clinical scores were improved but the kinematic scores were equal as the baseline, whose show the poor mobility of this patient.

By producing objective score based on 3D kinematics of the shoulder our system assessed the functionality of the shoulder. However, it can't be used yet for the diagnosis of complex pathology or to differentiate the pathologies. Our score is not related directly to pain but to the pain's effect on mobility. For example, if a patient experiences pain in a shoulder and therefore moves lesser his shoulder; our system will detect this lack of functionality. But in the case where there is no recovery of shoulder functionality even if the pain is removed after surgery our scores will remain low.

It is noteworthy that the three described scores compare the patient's affected and non affected shoulder only if the pathology is unilateral. Further study with more measurements on patients is needed to be able to use these scores to assess the functionality of a shoulder independently of the pathology.

Concerning the sensors attachment some precautions should be made. First in order to reduce the effects of skin artefacts a sticking elastic band was used to fix the sensors. In addition the module was placed on the distal and posterior part of the humerus where there are less skin movement and where sensor can detect all rotation of humerus. In fact, if the sensor is positioned at the top of the humerus (near the humeral head), the internal/external rotation can be not measured.

In order to estimate the repeatability of the system measurement where repeated on the 10 control subjects after 1 year. The comparison between the two measurements showed low difference (less than 3% in average with SD less than 10%) (Table 4).

The proposed system has also the potential to be used during daily activity before and after shoulder surgery and to provide a valuable outcome.

Another preferred embodiment of the invention is to measure the humerus activity in a daily life cycle to find out about the dominant shoulder of a subject.

Thirty-five healthy subjects (mean 32 years old±8) were studied. Two inertial modules were fixed by a patch on the dorsal side of the distal humerus and on the thorax. Each module consists of three miniature gyroscopes (Analog device, ADXRS 250, ±400°/s) which measured the limb angular velocity and three miniature accelerometers (Analog device, ADXL 210, ±5 g). The inertial module on the humerus measured the anterior elevation-extension, abduction-adduction and internal-external rotation of the shoulder and the module on thorax was used for detecting daily activities (walking, sitting, standing). Each inertial module, including the sensors and their conditioning electronics, was packaged in a small box. The signal from the sensors was amplified and low- pass filtered to remove any electronic noise. All signals were digitized and recorded by two synchronized data loggers (Physilogl, BioAGM, CH) carried on the subject's waist. Each subject carried the system during 1 day (8 h). Following completion of recording, the datawere transferred to a computer for further analysis.

Body posture allocations (sitting, standing and lying) as well as walking periods were detected by the trunk inertial module. The time of sit-stand or stand-sit transition was detected from the patterns of angular tilt obtained from the gyroscope. Pattern recognition of the vertical acceleration allowed detection of the transition and distinction between standing and sitting positions. The lying position was detected from the inclination of the trunk obtained from the accelerometers. Walking periods were defined as intervals with at least three gait cycles. Walking state was identified by analyzing the vertical accelerometer every five seconds.

The difference between right and left shoulder was shown for each period corresponding to sitting, standing and walking.

As shown the product of range of acceleration and range of angular velocity which inform about the power of the shoulder is a pertinent parameter in evaluating shoulder mobility.

The parameter Pr was defined which considered the 3D components of acceleration and angular velocity of the humerus obtained from the inertial module fixed on this segment:

Pr = velocity )

^j ) ^/ ( xEq ⁱu.4 ⁱ

Pr was estimated every 5 s for the left and the right humerus (Pr_Left, Pr_Rιght). In order to estimate shoulder usage, Pr was compared to a defined threshold (th). If Pr was under th the humerus was considered as motionless; otherwise it was considered as active. The periods where Pr > th were estimated as a percentage of the total monitoring time and were defined as activity. To define the threshold th, we turned on the system in rest position during 1 h to detect the meanvalue of the Pr for the left and right humerus. The mean value for Pr was used to define the optimum th.

If the difference between Pr_Lef_t and Pr_Rιght was positive and Pr_Lef_t was larger than th, the usage was classified as a left shoulder usage (ALS = 1). If the difference between Pr_Rιght and Pr_Left was positive and Pr_Rιght was larger than th, the usage was classified as a right shoulder usage (ARS = 1). The percentage of the left shoulder usage (ALSp) and right shoulder usage (ARSp) were described as

n = total time of measurement/5 s.

For each interval i of 5 s, P(i) parameter was defined as

If Pr_Rιght(i) > Pr_Left(i) + th

P(i) = Pr_Rιght(i) Elseif Pr_Left(i) > th

Else P(i) = 0

To define the threshold (th), we turned on the system in rest position during 1 h to detect the mean value of the Pr for the left and right humerus. The mean value for Pr_Left was 0.859 and the mean value for the PrRight was 0.556. These values corresponded to the average noise of the motion during rest. Activity periods should be several times above this noise level. To find the optimum threshold, we varied th from 1 to 10 per step of 1 for the 31 subjects. The optimum threshold was defined as the value where a difference of 1% was observed in the values of ARSp and ALSp (for the sit and stand postures). We obtained an optimum threshold of 3 which was used to estimate the activity periods.

The results showed, that the activity of both shoulders during standing and sitting postures was 74% and 59% for the right handed subjects and 79% and 69% for the left handed subjects. The activity of both shoulders during the walk was over 99%.

The usage of dominant side (ARSp and ALSp) over the day (8 h) and during each period of usage showed for right handed subjects (N = 23) the right side was on average 18% (±12) and 25% (±12) more used than the left side in standing and sitting postures, respectively. The opposite occurred for the left handed subjects (N = 8) : the right side was on average 8% (±13) and 18% (±21) less used than the left side in standing and sitting postures, respectively. For the walking periods the use of the right and left sides was almost identical (50-50% for the right handed subjects; 48-52% for the left handed subjects).

The mean of the P parameter during the daily activity for all right handed subjects was larger for the right shoulder compared to the left shoulder. The tendency was reversed for the left handed subjects on average although a few left handed subjects had higher intensity for the right shoulder.

Based on kinematic studies using inertial sensors, we were able to identify the difference between dominant and non-dominant shoulder and to determine hand dominance by considering the actual daily activity. This study provides preliminary evidence that this system is a useful tool for objectively assessing upper-limb usage during daily activity.

Claims

1. Device for a functional assessment of a shoulder comprising : three gyroscopes; three accelerometers; all designed to be attached to the humerus of a subject and for measuring the anterior elevation-extension (pitch), abduction- adduction (yaw) and internal-external rotation (roll) of the shoulder; and a processing unit at least suitable to process the received data of the gyroscopes to estimate the 3D range of angular velocity according to the equation

. .

2. Device according to claim 1, wherein the processing unit is further suitable to process the received data of the gyroscopes and accelerometers to observe the relationship between humerus acceleration and angular velocities according to the equation

Pr = V i— troll, pιtch,yaw rang °ei acceleration) ^■

velocity ^j ) ^/ ( ^\Eq -ⁱu.4) ^/

3. Device according to one of the preceding claims, wherein the processing unit is further suitable to process the received data from the gyroscopes and anthropometries data of the subject to observe the moments of the humerus according to the equation

I...inertia matrix ₍IQ ... angular velocities

4. Method for functionally assessing a shoulder by measuring and scoring angular velocities of a humerus, resulting in a score (RAV); angular velocities and accelerations of the humerus, resulting in a score

(P); and the sum of moments of the humerus, resulting in a score (M).

5. Method according to claim 4, wherein the score RAV is a range of angular velocity and is calculated by the difference between the maximum and minimum of angular velocity, preferably in 3 dimensions, measured during one test, according to the equation

(V range{angular velocity))

RAVr = — -

3 (Equ. l)

6. Method according to any of the preceding claims 4 or 5, wherein the score P is a product of the range of acceleration and the range of angular velocity, measured during one test, according to the equation

Pr = ^ range(acceleration) ^■ range(angular velocity) (Equ.4)

7. Method according to any of the preceding claims 4 to 6, wherein the score M is based on the angular velocities of the humerus and anthropometries data of the subject, measured during one test, according to the equation

I...inertia matrix Q_j ... angular velocities

8. Method according to any of the preceding claims 4 to 7, wherein the scores of healthy subjects were compared to other according scores.

9. Method according to any of the preceding claims 4 to 8, wherein the scores of healthy subjects were compared to scores of pathological subjects.

10. Method according to any of the preceding claims 4 to 9, wherein the scores P, RAV and/or M are measured during at least two predetermined movements or movement combinations and the scores are an average of all according values of the tests.

11. Method according to any of the preceding claims 4 to 10, wherein the scores of left handed subjects are compared to right handed subjects.

12. Method according to any of the preceding claims 4 to 11, wherein the measurement is done during at least 1 hour, and where the score P is estimated at least every second.