- Email: [email protected]
Knee pathology classification using the instantaneous screws parameter and Support Vector Machines classifier
$178
Journal o f Biomechanics 2006, Vol. 39 (Suppl 1)
5013 Mo, 16:30-16:45 (P12) Knee p a t h o l o g y classification using the instantaneous screws parameter and Support Vector Machines classifier A. Wolf 1,2, A. Degani 2. 1Dept. of Mechanical Engineering, Technion -Israel Institute of Technology, Haifa, Israel, 2Robotics Institute, Carnegie Mellon University, Pittsburgh, PA, USA Assessment of joint pathology is not always trivial and requires, in most cases, a combination of visualization data (i.e. CT, MR) combined with physical subjective tests performed by the physician. Most of the tests performed today are performed manually by the physician while manipulating the patient lower limb. In our study, we demonstrate a method in which kinematic measurements of the knee are used to automatically identify the knee pathology. The flexions of knees with different pathologies are tracked using an optical tracking system. The screw parameters which describe the transformation of the tibia with respect to the femur in each two successive observation are represented as the instantaneous screw axis of the motion given in its PlUcker line coordinate, along with its corresponding pitch. The set of screw parameters associated with a particular knee with a given pathology is then identified and clustered in R 6 to form a "signature" of the motion for the given pathology. Plastic Sawbones model and two cadaver knees with different pathologies were tracked, and the resulting screws were used to train a classifier system. The Support Vector Machines (SVM) classifier was used to cluster the different pathologies in R 6. The system was then tested successfully with new, never trained before data. The SVM classifier was able to classify the different pathologies with a high success percentage of 80 to 90 percent. We further demonstrate that the capability of the classifier to identify pathologies does not depend on the sampling frequency, i.e. sampling frequency and motion velocity do not have to be synchronized between patients.
Oral Presentations joint were used to determine control of landing angle and limb shortening during deceleration (telescopic inverted pendulum). Independent t-tests were used to compare between groups. Pearson's correlations between approach velocity and dependent variables were explored. All participants were able to carry out this deceleration task without events of giving way. Despite the significantly slower approach velocity (ACLD: 2.15±0.36 m/s vs. Control: 2.38±0.39 m/s; p < 0.01 ), ACLD patients took significantly longer to decelerate (205.3±37.5 ms vs. 157.5±29.5 ms; p <0.05) and took longer to achieve stable stance (p<0.01). Landing style differed significantly, with ACLD patients landing more upright (p<0.01). They used change in landing angle more (p<0.01) and limb shortening less (p<0.01) than control subjects. No difference in net ankle moment was observed, but net knee moment was significantly reduced in the ACLD patients (230.8±75.1 N m vs. 278.6±80.3 Nm; p <0.05). Absorption work during deceleration was significantly increased at the ankle joint (23.6±4.1 J vs. 16.2±12.1 J; p<0.05) and significantly decreased at the knee joint in the ACLD group (54.7±35.3J vs. 107.5±70.9 J; p < 0.05). Approach velocity was not significantly correlated with most dependent variables except peak knee moment. No between-group differences in EMG onset were observed. Adaptations in landing style for the ACLD group meant their landing technique relied more on ankle and less on knee control compared to healthy subjects. This knee avoidance technique was less effective to achieve deceleration but presumably protects the knee from instability. 6714 Tu, 08:45-09:00 (P16) A biomechanical study o f graft failure in early period o f graft fixation in acl s u r g e r y M. Bhattacharyya. Investigation performed at Imperial College London, UK Aim: To study mechanism of failure of graft fixation method in the proximal
6.1.3. Traumatology-Orthopaedics 5284 Tu, 08:15-08:30 (P16) Reconstruction o f the anterior cruciate ligament (ACL) in doublebundle-technique: Biomechanical evaluation o f possible insertion points S. Lorenz 1, M. Berr 1, W. Unkh~r 1, S. Eichhorn 2, A.B. Imhoff 1. 1Dept. ef Sports Orthopedic, Klinik f#r Orthop~die und Sportorthop~die, Klinikum rechts der Isar, Technical University of Munich, Germany, 2Dept. of Biomechanics, Klinik f#r Orthop~die und Sportorthop~die, Klinikum rechts der Isar, Technical University of Munich, Germany Though there is a renaissance of reconstruction of the ACL in a double-bundle technique reconstructing the anteromedial and posterolateral bundle there is no consensus about the insertion sites. Neither the exact points are described for the femoral and tibial tunnels nor the angle, in which the bundles should be fixed. The aim of this study was to find the best insertion points femorally and tibially in relation to the center of gravitiy of each bundle. Therefore the normal passive motion of 7 fresh frozen specimens was registered in a fixation device with 6 degrees of freedom. After dissection of all soft tissue including the anterior and posterior cruciate ligament the center of gravity (COG) was defined for both bundles. Around this bundles 4 points were predefined that had a distance of 3 mm in anterior, posterior, medial and lateral direction at the tibial side. At the femur points were defined that had a distance of 3mm in the same manner in relation to the Blumensaat's line. Every point at the femur was connected with a point at the tibia using a fiberglass suture. The length changing of the sutures was registered while the knee was moved in the fixation device. The presentation shows the curves of the length changing. The curves are compared with each other to find out a combination of insertion points that has the least changing of length. Furthermore the maxima and minima of the curves are discussed to find appropriate flexion angles for fixation of the anteromedial and posterolateral bundle. 4271 Tu, 08:30-08:45 (P16) Landing style differences between anterior cruciate ligament deficient and healthy subjects R. van Deursen, N. Phillips. Research Centre for Clinical Kinaesiology, School of Healthcare Studies, Cardiff University, Cardiff, UK Deceleration during landing challenges knee stability, which is particularly difficult for patients with ACL deficiency (ACLD), potentially requiring adaptation of landing style. This study aimed to identify the mechanism of such an adaptation in landing technique. Thirty ACLD patients (non-reconstructed) and 30 control subjects performed a run and stop task. They landed on a force plate whilst kinematic, kinetic and electromyographic data (quadriceps, hamstrings, gastrocnemius, soleus and tibialis anterior) were recorded. Sagittal net joint moment and work were calculated. Polar coordinates of the centre of the pelvis relative to the ankle
tibia with metallic interference screw using bone patellar tendon bone and soft tendon grafts under cyclic loading to simulate forces applied in rehabilitation. Method: The experimental validated model used in this study was the 16 proximal tibia of the calves aged 20-26 weeks, harvested immediately after death and the soft tissue was removed. Bovine extensor tendons were harvested from the forelimbs of calves for the soft tissue grafts. These were prepared with the aid of commercially available graft 'sizer' cylinders to 9 mm diameter. Eight human BPTB grafts were harvested from cadaveric knees with a mean age of approximately 70 years. (Data was not available for all specimens). These grafts were prepared to 9 mm diameter. The bone blocks were typically 9 mm wide, 4 mm thick and 25 mm long. All specimens were frozen for storage at -200 while sealed in a polyethylene bas and thawed 12 hours before use. They were kept moist during use by wrapping in tissue paper. Result: Differences in graft slippage creep for 8 tested graft specimen were analysed with one tailed t-TEST. Creep value showed no statistical significance however, the graft slippage showed statistical significance. The t-TEST at 1000 cycles shows P =0.05. Graft slippage was measured at 4000 load cycles from 0 - 1 5 0 - 0 N , and the value found in soft tissue model (hamstring group) was 1.29±0.81 and that of bone tissue (patella tendon) model 0.75±0.06. Discussion: This experiment shows a significant difference in the behaviour under cyclic loading of two different fixation commonly employed for ACL reconstruction. The main finding was the less slippage of bone-to-bone fixation with metallic interference screw. This may be due to the difference between gripping soft tissues and bone. The study was limited by different tissue type.
6.1.4.1 Rehabilitation Orthopaedics 4940 Tu, 11:00-11:15 (P19) The influence o f a n t h r o p o m o r p h i c features and bilateral knee strength on one-leg hop performance W.-L. Chen, P.-Y. Lee, C.-'~ Yang. National Cheng Kung University, Tainan, Taiwan Although one-leg hop test has been widely used as a reliable test to evaluate performance after anterior cruiciate ligament (ACL) reconstruction, several studies have reported low sensitivity in detecting functional limitations of ACL-deficient knees and low correlations between hop-length and joint laxity or proprioception. On the other hand, little has been studied regarding the influence of muscle strength in ipsilateral legs on one-leg hop-length, let alone the role that muscle strength in contralateral legs might play. In addition, the anthropomorphic influence has never been discussed. Purpose: To investigate the influence of anthropomorphic features and the muscle strength of both knees on one-leg hop performance. Methods: 26 healthy subjects were recruited to participate one-leg hop and knee strength tests for both legs (n =52). The relationship between hop-length and height or weight was determined by Pearson correlation coefficients. Several partial correlation procedures controlling for height and weight were conducted to determine the relationship between hop-length and the related knee strength.
Journal o f Biomechanics 2006, Vol. 39 (Suppl 1)
5013 Mo, 16:30-16:45 (P12) Knee p a t h o l o g y classification using the instantaneous screws parameter and Support Vector Machines classifier A. Wolf 1,2, A. Degani 2. 1Dept. of Mechanical Engineering, Technion -Israel Institute of Technology, Haifa, Israel, 2Robotics Institute, Carnegie Mellon University, Pittsburgh, PA, USA Assessment of joint pathology is not always trivial and requires, in most cases, a combination of visualization data (i.e. CT, MR) combined with physical subjective tests performed by the physician. Most of the tests performed today are performed manually by the physician while manipulating the patient lower limb. In our study, we demonstrate a method in which kinematic measurements of the knee are used to automatically identify the knee pathology. The flexions of knees with different pathologies are tracked using an optical tracking system. The screw parameters which describe the transformation of the tibia with respect to the femur in each two successive observation are represented as the instantaneous screw axis of the motion given in its PlUcker line coordinate, along with its corresponding pitch. The set of screw parameters associated with a particular knee with a given pathology is then identified and clustered in R 6 to form a "signature" of the motion for the given pathology. Plastic Sawbones model and two cadaver knees with different pathologies were tracked, and the resulting screws were used to train a classifier system. The Support Vector Machines (SVM) classifier was used to cluster the different pathologies in R 6. The system was then tested successfully with new, never trained before data. The SVM classifier was able to classify the different pathologies with a high success percentage of 80 to 90 percent. We further demonstrate that the capability of the classifier to identify pathologies does not depend on the sampling frequency, i.e. sampling frequency and motion velocity do not have to be synchronized between patients.
Oral Presentations joint were used to determine control of landing angle and limb shortening during deceleration (telescopic inverted pendulum). Independent t-tests were used to compare between groups. Pearson's correlations between approach velocity and dependent variables were explored. All participants were able to carry out this deceleration task without events of giving way. Despite the significantly slower approach velocity (ACLD: 2.15±0.36 m/s vs. Control: 2.38±0.39 m/s; p < 0.01 ), ACLD patients took significantly longer to decelerate (205.3±37.5 ms vs. 157.5±29.5 ms; p <0.05) and took longer to achieve stable stance (p<0.01). Landing style differed significantly, with ACLD patients landing more upright (p<0.01). They used change in landing angle more (p<0.01) and limb shortening less (p<0.01) than control subjects. No difference in net ankle moment was observed, but net knee moment was significantly reduced in the ACLD patients (230.8±75.1 N m vs. 278.6±80.3 Nm; p <0.05). Absorption work during deceleration was significantly increased at the ankle joint (23.6±4.1 J vs. 16.2±12.1 J; p<0.05) and significantly decreased at the knee joint in the ACLD group (54.7±35.3J vs. 107.5±70.9 J; p < 0.05). Approach velocity was not significantly correlated with most dependent variables except peak knee moment. No between-group differences in EMG onset were observed. Adaptations in landing style for the ACLD group meant their landing technique relied more on ankle and less on knee control compared to healthy subjects. This knee avoidance technique was less effective to achieve deceleration but presumably protects the knee from instability. 6714 Tu, 08:45-09:00 (P16) A biomechanical study o f graft failure in early period o f graft fixation in acl s u r g e r y M. Bhattacharyya. Investigation performed at Imperial College London, UK Aim: To study mechanism of failure of graft fixation method in the proximal
6.1.3. Traumatology-Orthopaedics 5284 Tu, 08:15-08:30 (P16) Reconstruction o f the anterior cruciate ligament (ACL) in doublebundle-technique: Biomechanical evaluation o f possible insertion points S. Lorenz 1, M. Berr 1, W. Unkh~r 1, S. Eichhorn 2, A.B. Imhoff 1. 1Dept. ef Sports Orthopedic, Klinik f#r Orthop~die und Sportorthop~die, Klinikum rechts der Isar, Technical University of Munich, Germany, 2Dept. of Biomechanics, Klinik f#r Orthop~die und Sportorthop~die, Klinikum rechts der Isar, Technical University of Munich, Germany Though there is a renaissance of reconstruction of the ACL in a double-bundle technique reconstructing the anteromedial and posterolateral bundle there is no consensus about the insertion sites. Neither the exact points are described for the femoral and tibial tunnels nor the angle, in which the bundles should be fixed. The aim of this study was to find the best insertion points femorally and tibially in relation to the center of gravitiy of each bundle. Therefore the normal passive motion of 7 fresh frozen specimens was registered in a fixation device with 6 degrees of freedom. After dissection of all soft tissue including the anterior and posterior cruciate ligament the center of gravity (COG) was defined for both bundles. Around this bundles 4 points were predefined that had a distance of 3 mm in anterior, posterior, medial and lateral direction at the tibial side. At the femur points were defined that had a distance of 3mm in the same manner in relation to the Blumensaat's line. Every point at the femur was connected with a point at the tibia using a fiberglass suture. The length changing of the sutures was registered while the knee was moved in the fixation device. The presentation shows the curves of the length changing. The curves are compared with each other to find out a combination of insertion points that has the least changing of length. Furthermore the maxima and minima of the curves are discussed to find appropriate flexion angles for fixation of the anteromedial and posterolateral bundle. 4271 Tu, 08:30-08:45 (P16) Landing style differences between anterior cruciate ligament deficient and healthy subjects R. van Deursen, N. Phillips. Research Centre for Clinical Kinaesiology, School of Healthcare Studies, Cardiff University, Cardiff, UK Deceleration during landing challenges knee stability, which is particularly difficult for patients with ACL deficiency (ACLD), potentially requiring adaptation of landing style. This study aimed to identify the mechanism of such an adaptation in landing technique. Thirty ACLD patients (non-reconstructed) and 30 control subjects performed a run and stop task. They landed on a force plate whilst kinematic, kinetic and electromyographic data (quadriceps, hamstrings, gastrocnemius, soleus and tibialis anterior) were recorded. Sagittal net joint moment and work were calculated. Polar coordinates of the centre of the pelvis relative to the ankle
tibia with metallic interference screw using bone patellar tendon bone and soft tendon grafts under cyclic loading to simulate forces applied in rehabilitation. Method: The experimental validated model used in this study was the 16 proximal tibia of the calves aged 20-26 weeks, harvested immediately after death and the soft tissue was removed. Bovine extensor tendons were harvested from the forelimbs of calves for the soft tissue grafts. These were prepared with the aid of commercially available graft 'sizer' cylinders to 9 mm diameter. Eight human BPTB grafts were harvested from cadaveric knees with a mean age of approximately 70 years. (Data was not available for all specimens). These grafts were prepared to 9 mm diameter. The bone blocks were typically 9 mm wide, 4 mm thick and 25 mm long. All specimens were frozen for storage at -200 while sealed in a polyethylene bas and thawed 12 hours before use. They were kept moist during use by wrapping in tissue paper. Result: Differences in graft slippage creep for 8 tested graft specimen were analysed with one tailed t-TEST. Creep value showed no statistical significance however, the graft slippage showed statistical significance. The t-TEST at 1000 cycles shows P =0.05. Graft slippage was measured at 4000 load cycles from 0 - 1 5 0 - 0 N , and the value found in soft tissue model (hamstring group) was 1.29±0.81 and that of bone tissue (patella tendon) model 0.75±0.06. Discussion: This experiment shows a significant difference in the behaviour under cyclic loading of two different fixation commonly employed for ACL reconstruction. The main finding was the less slippage of bone-to-bone fixation with metallic interference screw. This may be due to the difference between gripping soft tissues and bone. The study was limited by different tissue type.
6.1.4.1 Rehabilitation Orthopaedics 4940 Tu, 11:00-11:15 (P19) The influence o f a n t h r o p o m o r p h i c features and bilateral knee strength on one-leg hop performance W.-L. Chen, P.-Y. Lee, C.-'~ Yang. National Cheng Kung University, Tainan, Taiwan Although one-leg hop test has been widely used as a reliable test to evaluate performance after anterior cruiciate ligament (ACL) reconstruction, several studies have reported low sensitivity in detecting functional limitations of ACL-deficient knees and low correlations between hop-length and joint laxity or proprioception. On the other hand, little has been studied regarding the influence of muscle strength in ipsilateral legs on one-leg hop-length, let alone the role that muscle strength in contralateral legs might play. In addition, the anthropomorphic influence has never been discussed. Purpose: To investigate the influence of anthropomorphic features and the muscle strength of both knees on one-leg hop performance. Methods: 26 healthy subjects were recruited to participate one-leg hop and knee strength tests for both legs (n =52). The relationship between hop-length and height or weight was determined by Pearson correlation coefficients. Several partial correlation procedures controlling for height and weight were conducted to determine the relationship between hop-length and the related knee strength.