Pietro Cerveri

Abstract: In this paper, we propose a method to estimate the parameters of a double hinge model of the trapeziometacarpal joint (TMC) by MRI-based motion analysis. The model includes two non-orthogonal and non-intersecting rotation axes accounting for flexion-extension (F-E) and adduction-abduction (A-A). We evaluated the quality of the estimated model parameters in the prediction of the relative motion of the first metacarpal bone with respect to the trapezium. As a result, we obtained that: (a) the estimated location and orientation of the F-E and A-A axes were in agreement with previous in vitro studies, (b) the motion of the first metacarpal predicted by the 2 degrees of freedom (2DoF) model exhibits a maximum surface distance error in the range of about 2 mm and (c) four thumb postures at the boundary of the TMC range of motion are sufficient to provide a good estimation of the 2DoF TMC kinematic model and good reproducibility ( approximately 1.7 mm) of the real thumb motion at TMC level.

Abstract: Hip joint centre (HJC) localization is used in several biomedical applications, such as movement analysis and computer-assisted orthopaedic surgery. The purpose of this study was to validate in vitro a new algorithm (MC-pivoting) for HJC computation and to compare its performances with the state-of-the-art (least square approach-LSA). The MC-pivoting algorithm iteratively searches for the 3D coordinates of the point belonging to the femoral bone that, during the circumduction of the femur around the hip joint (pivoting), runs the minimum length trajectory. The algorithm was initialized with a point distribution that can be considered close to a Monte Carlo simulation sampling all around the LSA estimate. The performances of the MC-pivoting algorithm, compared with LSA, were evaluated with tests on cadavers. Dynamic reference frames were applied on both the femur and the pelvis and were tracked by an optical localizer. Results proved the algorithm accuracy (1.7mm+/-1.6, 2.3-median value+/-quartiles), reliability (smaller upper quartiles of the errors distribution with respect to LSA) and robustness (reduction of the errors also in case of large pelvis displacements).

Abstract: Computer-assisted total knee replacement (TKR) has been shown to improve radiographic alignment. Continuous feedback from the navigation system allows accurate adjustment of the bone cuts, thus reducing errors. The aim of this study was to determine the impact of experience both with computer navigation and knee replacement surgery on the frequency of errors in intraoperative bone cuts and implant alignment. Three homogeneous patient groups undergoing computer assisted TKR were included in the study. Each group was treated by one of three surgeons with varying experience in computer-aided and knee replacement surgery. Surgeon A had extensive experience in knee replacement and computer-assisted surgery. Surgeon B was an experienced knee replacement surgeon. A general orthopaedic surgeon with limited knee replacement surgery experience performed all surgeries in group C. The cutting errors and the number of re-cuts were determined intraoperatively. The complications and mean surgical time were collected for each group. The postoperative frontal femoral component angle, frontal tibial component angle, hip-knee-ankle angle and component slopes were evaluated. The results showed that the number of cutting errors were lowest for TKR performed by the surgeon with experience in navigation. This difference was statistically significant when compared to the general orthopaedic surgeon. A statistically significant superior result was achieved in final mechanical axis alignment for the surgeon experienced in computer-guided surgery compared to the other two groups (179.3 degrees compared to 178.9 degrees and 178.1 degrees ). However, the total number of outliers was similar, with no statistically significant differences among the three surgeons. Experience with navigation significantly reduced the surgical time.

Abstract: INTRODUCTION: The authors performed a matched paired study between two groups: bi-unicompartmental (Bi-UKR) versus total knee replacements (TKR) for the treatment of isolated bicompartmental tibio-femoral knee arthritis with an asymptomatic patello-femoral joint. The Authors believe that Bi-UKR could achieve comparable outcomes than TKR, but with a real less invasive surgery and maintaining a higher joint function. MATERIALS AND METHODS: A total of 22 patients with bicompartmental tibio-femoral knee arthritis, who underwent Bi-UKR between January 1999 and March 2003, were included in the study (group A). In all the knees the arthritic changes were graded according to the classification of Alback. All patients had an asymptomatic patello-femoral joint. All patients had a varus deformity lower than 8 degrees , a body-mass index lower than 34, no clinical evidence of ACL laxity or flexion deformity and a preoperative range of motion of a least 110 degrees . At a minimum follow-up of 48 months, every single patient in group A was matched with a patient who had undergone a computer assisted TKR between August 1999 and September 2002 (group B). In the Bi-UKR group, in two cases we registered intraoperatively the avulsion of the treated tibial spines, requiring intra-operative internal fixation and without adverse effects on the final outcome. Statistical analysis of the results was performed. RESULTS: At a minimum follow-up of 48 months there were no statistical significant differences in the surgical time while the hospital stay was statistically longer in TKR group. No statistically significant difference was seen for the Knee Society, Functional and GIUM scores between the two groups. Statistically significant better WOMAC Function and Stiffness indexes were registered for the Bi-UKR group. TKR implants were statistically better aligned with all the implants positioned within 4 degrees of an ideal hip-knee-ankle (HKA) angle of 180 degrees . CONCLUSIONS: The results of this 48 months follow-up study suggest that Bi-UKR is a viable option for bicompartmental tibio-femoral arthritis at least as well as TKR but maintaining a higher level of function.

Abstract: This article analyzes a realistic kinematic model of the trapezio-metacarpal (TM) joint in the human thumb that involves two non-orthogonal and non-intersecting rotation axes. The estimation of the model parameters, i.e. the position and orientation of the two axes with respect to an anatomical coordinate system, was carried out by processing the motion of nine retroreflective markers, externally attached to the hand surface, surveyed by a video motion capture system. In order to compute the model parameters, prototypical circumduction movements were processed within an evolutionary optimization approach. Quality and reproducibility in assessing the parameters were demonstrated across multiple testing sessions on 10 healthy subjects (both left and right thumbs), involving the complete removal of all markers and then retesting. Maximum errors of less than 5 mm in the axis position and less than 6 degrees in the orientation were found, respectively. The inter-subject mean distance between the two axes was 4.16 and 4.71 mm for right and left TM joints, respectively. The inter-subject mean relative orientation between the two axes was about 106 and 113 degrees for right and left TM joints, respectively. Generalization properties of the model were evaluated quantitatively on opposition movements in terms of distance between measured and predicted marker positions (maximum error less than 5 mm). The performance of the proposed model compared favorably with the one (maximum error in the range of 7-8 mm) obtained by applying a universal joint model (orthogonal and intersecting axes). The ability of in vivo estimating the parameters of the proposed kinematic model represents a significant improvement for the biomechanical analysis of the hand motion.

Abstract: Modulation of the activity of the subthalamic nucleus (STN) using deep brain stimulation (DBS) in patients with advanced Parkinson's disease is the most common procedure performed today by functional neurosurgeons. The STN contours cannot be entirely identified on common 1.5 T images; in particular, the ventromedial border of the STN often blends with the substantia nigra. 3 T magnetic resonance imaging (MRI) provides better resolution and can improve the identification of the STN borders. In this work, we have directly identified the STN using 3 T MR imaging to validate the accuracy of a computer-aided atlas-based procedure for automatic STN identification. Coordinates of the STN were obtained from the Talairach and Tournoux atlas and transformed into the coordinates of the Montreal Neurological Institute (MNI) standard brain volume, creating a mask representation of the STN. 3 T volumetric T1 and T2 weighted (T1w and T2w, respectively) acquisitions were obtained for ten patients. The MNI standard brain volume was registered onto each patient MRI, using a new approach based on global affine, region-of-interest affine, and local nonrigid registrations. The estimated deformation field was then applied to the STN atlas-based mask, providing its location on the patient MRI. The entire procedure required on average about 20 min. Because STN is easily identifiable on 3 T T2w-MRIs, it was manually delineated; the coordinates of the center of mass of the manually and automatically identified structures were compared. Additionally, volumetric overlapping indices were calculated and the spatial relationship between the midcommissural point and the STN center of mass was investigated. All indices indicated, on average, good agreement between manually and automatically identified structures; displacement of the centers of mass of the manually and automatically identified structures was less than or equal to 2.35 mm, and more than 80% of the manually identified volume was covered by the automatic localization, on average. Bland-Altman analysis indicated that the automatic STN identification was within the limits of agreement with the manual localization on 3 T MRIs. Automatic atlas-based STN localization provides an accurate and user-friendly tool and can enhance target identification when 1.5 T scanners with limited capability to identify the STN boundaries are used.

Abstract: Functional magnetic resonance imaging (fMRI) is used to distinguish areas of the brain responsible for different tasks and functions. It is possible, for example, by using fMRI images, to identify particular regions in the brain which can be considered as "functional organs at risk" (fOARs), i.e., regions which would cause significant patient morbidity if compromised. The aim of this study is to propose and validate a method to exploit functional information for the identification of fOARs in CyberKnife (Accuray, Inc., Sunnyvale, CA) radiosurgery treatment planning; in particular, given the high spatial accuracy offered by the CyberKnife system, local nonrigid registration is used to reach accurate image matching. Five patients affected by arteriovenous malformations (AVMs) and scheduled to undergo radiosurgery were scanned prior to treatment using computed tomography (CT), three-dimensional (3D) rotational angiography (3DRA), T2 weighted and blood oxygenation level dependent echo planar imaging MRI. Tasks were chosen on the basis of lesion location by considering those areas which could be potentially close to treatment targets. Functional data were superimposed on 3DRA and CT used for treatment planning. The procedure for the localization of fMRI areas was validated by direct cortical stimulation on 38 AVM and tumor patients undergoing conventional surgery. Treatment plans studied with and without considering fOARs were significantly different, in particular with respect to both maximum dose and dose volume histograms; consideration of the fOARs allowed quality indices of treatment plans to remain almost constant or to improve in four out of five cases compared to plans with no consideration of fOARs. In conclusion, the presented method provides an accurate tool for the integration of functional information into AVM radiosurgery, which might help to minimize undesirable side effects and to make radiosurgery less invasive.

Abstract: This paper describes methods and experimental studies concerned with quantitative reconstruction of finger movements in real-time, by means of multi-camera system and 24 surface markers. The approach utilizes a kinematic model of the articulated hand which consists in a hierarchical chain of rigid body segments characterized by 22 functional degrees of freedom and the global roto-translation. This work is focused on the experimental evaluation of a kinematical hand model for biomechanical analysis purposes.From a static posture, a completely automatic calibration procedure, based on anthropometric measures and geometric constraints, computes axes, and centers of rotations which are then utilized as the base of an interactive real-time animation of the hand model. The motion tracking, based on automatic marker labeling and predictive filter, is empowered by introducing constraints from functional finger postures. The validation is performed on four normal subjects through different right-handed motor tasks involving voluntary flexion-extension of the thumb, voluntary abduction-adduction of the thumb, grasping, and finger pointing. Performances are tested in terms of repeatability of angular profiles, model-based ability to predict marker trajectories and tracking success during real-time motion estimation. Results show intra-subject repeatability of the model calibration both to different postures and to re-marking in the range of 0.5 and 2 mm, respectively. Kinematic estimation proves satisfactory in terms of prediction capability (index finger: maximum RMSE 2.02 mm; thumb: maximum RMSE 3.25 mm) and motion reproducibility (R (2) coefficients--index finger: 0.96, thumb: 0.94). During fast grasping sequence (60 Hz), the percentage of residual marker occlusions is less than 1% and processing and visualization frequency of 50 Hz confirms the real-time capability of the motion estimation system.

Abstract: Treatment targets in functional neurosurgery usually consist of selected structures within the thalamus and basal ganglia, which can be stimulated in order to affect specific brain pathways. Chronic electrical stimulation of these structures is a widely used approach for selected patients with advanced movement disorders. An alternative therapeutic solution consists of producing a lesion in the target nucleus, for example by means of radiosurgery, a noninvasive procedure, and this prevents the use of intraoperative microelectrode recording as a method for accurate target definition. The need to have accurate noninvasive localization of the target motivated our previous work on atlas-based identification; the aim of this present work is to provide additional validation of this approach based on the identification of the red nuclei (RN), which are located near the subthalamic nucleus (STN). Coordinates of RN were obtained from the Talairach and Tournoux (TT) atlas and transformed into the coordinates of the Montreal Neurological Institute (MNI) atlas, creating a mask representation of RN. The MNI atlas volume was nonrigidly registered onto the patient magnetic resonance imaging (MRI). This deformation field was then applied to the RN mask, providing its location on the patient MRI. Because RN are easily identifiable on 1.5 T T2-MRI images, they were manually delineated; the coordinates of the centers of mass of the manually and automatically identified structures were compared. Additionally, volumetric overlapping indices were calculated. Ten patients were examined by this technique. All indices indicated a high level of agreement between manually and automatically identified structures. These results not only confirm the accuracy of the method but also allow fine tuning of the automatic identification method to be performed.

Abstract: Functional disorders of the brain, such as Parkinson's disease, dystonia, epilepsy, and neuropathic pain, may exhibit poor response to medical therapy. In such cases, surgical intervention may become necessary. Modern surgical approaches to such disorders include radio-frequency lesioning and deep brain stimulation (DBS). The subthalamic nucleus (STN) is one of the most useful stereotactic targets available: STN DBS is known to induce substantial improvement in patients with end-stage Parkinson's disease. Other targets include the Globus Pallidus pars interna (GPi) for dystonia and Parkinson's disease, and the centromedian nucleus of the thalamus (CMN) for neuropathic pain. Radiosurgery is an attractive noninvasive alternative to treat some functional brain disorders. The main technical limitation to radiosurgery is that the target can be selected only on the basis of magnetic resonance anatomy without electrophysiological confirmation. The aim of this work is to provide a method for the correct atlas-based identification of the target to be used in functional neurosurgery treatment planning. The coordinates of STN, CMN, and GPi were identified in the Talairach and Tournoux atlas and transformed to the corresponding regions of the Montreal Neurological Institute (MNI) electronic atlas. Binary masks describing the target nuclei were created. The MNI electronic atlas was deformed onto the patient magnetic resonance imaging-T1 scan by applying an affine transformation followed by a local nonrigid registration. The first transformation was based on normalized cross correlation and the second on optimization of a two-part objective function consisting of similarity criteria and weighted regularization. The obtained deformation field was then applied to the target masks. The minimum distance between the surface of an implanted electrode and the surface of the deformed mask was calculated. The validation of the method consisted of comparing the electrode-mask distance to the clinical outcome of the treatments in ten cases of bilateral DBS implants. Electrode placement may have an effect within a radius of stimulation equal to 2 mm, therefore the registration process is considered successful if error is less than 2 mm. The registrations of the MNI atlas onto the patient space succeeded in all cases. The comparison of the distance to the clinical outcome revealed good agreement: where the distance was high (at least in one implant), the clinical outcome was poor; where there was a close correlation between the structures, clinical outcome revealed an improvement of the pathological condition. In conclusion, the proposed method seems to provide a useful tool for the identification of the target nuclei for functional radiosurgery. Also, the method is applicable to other types of functional treatment.

Abstract: The estimation of the skeletal motion obtained from marker-based motion capture systems is known to be affected by significant bias caused by skin movement artifacts, which affects joint center and rotation axis estimation. Among different techniques proposed in the literature, that based on rigid body model, still the most used by commercial motion capture systems, can smooth only part of the above effects without eliminating their main components. In order to sensibly improve the accuracy of the motion estimation, a novel technique, named local motion estimation (LME), is proposed. This rests on a recently described approach that, using virtual humans and extended Kalman filters, estimates the kinematical variables directly from 2D measurements without requiring the 3D marker reconstruction. In this paper, we show how such method can be extended to include the computation of the local marker displacement due to skin artifacts. The 3D marker coordinates, expressed in the corresponding local reference coordinate frames, are inserted into the state vector of the filter and their dynamics is automatically estimated, with adequate accuracy, without assuming any particular deformation function. Simulated experiments of lower limb motion, involving systematic mislocations (5, 10, 20 mm) and random errors of the marker coordinates and joint center locations (+/-5, +/-10, +/-15 mm), have shown that artifact motion can be substantially decoupled from the global skeletal motion with an effective increase of the accuracy wrt standard techniques. In particular, the comparison between the nominal kinematical variables and the one recovered from markers attached to the skin surface proved LME to be sensibly superior (50% in the worse condition) to the methods imposing marker-bone rigidity. In conclusion, while requiring further validation on real movement data, we argue that the proposed method can constitute an appropriate approach toward the improvement of the human motion estimation.

Abstract: In the field of 3D reconstruction of human motion from video, model-based techniques have been proposed to increase the estimation accuracy and the degree of automation. The feasibility of this approach is strictly connected with the adopted biomechanical model. Particularly, the representation of the kinematic chain and the assessment of the corresponding parameters play a relevant role for the success of the motion assessment. In this paper, the focus is on the determination of the kinematic parameters of a general hand skeleton model using surface measurements. A novel method that integrates nonrigid sphere fitting and evolutionary optimization is proposed to estimate the centers and the functional axes of rotation of the skeletal joints. The reliability of the technique is tested using real movement data and simulated motions with known ground truth 3D measurement noise and different ranges of motion (RoM). With respect to standard nonrigid sphere fitting techniques, the proposed method performs 10-50% better in the best condition (very low noise and wide RoM) and over 100% better with physiological artifacts and RoM. Repeatability in the range of a couple of millimeters, on the localization of the centers of rotation, and in the range of one degree, on the axis directions is obtained from real data experiments.

Abstract: Preprocessing, binning and dataset subsampling are investigated with regard to simultaneous maximisation of the speed, accuracy and robustness of CT-3D rotational angiography (3DRA) registration. Clinical diagnosis and treatment can both take advantage of this integration, because 3DRA allows the shape of vessel structures to be evaluated three-dimensionally with respect to standard 2D projective angiography. The method for optimising preprocessing, binning and subsampling consisted of independent variation of the corresponding parameters to maximise robustness and speed while maintaining subvoxel accuracy; the latter was computed as the sum of the mean squared errors initially present in the registrations with the errors relative to both binning and subsampling. The results suggest the choice of 256 bins, steps between 14 mm (coarse optimisation) and 2.5 mm (fine optimisation) and bone segmentation by threshold, for binning, subsampling and preprocessing, respectively. The application of this parameter set-up to 50 CT-3DRA registrations resulted in a saving, on average, of 40% of the time with respect to the method previously used, while registration error was maintained within 2 mm (1.97 mm, 90% confidence interval) and robustness was increased, so that no manual initial realignment was needed in 48 registrations. Validation by the registration of images acquired for a head phantom showed subvoxel residual errors. In conclusion, the proposed procedure can be considered a satisfactory strategy to optimise CT-3DRA registration.

Abstract: The inclusion of organ deformation and movement in radiosurgery treatment planning is of increasing importance as research and clinical applications begin to take into consideration the effects of physiological processes, like breathing, on the shape and position of lesions. In this scenario, the challenge is to localize the target in toto (not only by means of marker sampling) and to calculate the dose distribution as the sum of all the contributions from the positions assumed by the target during the respiratory cycle. The aim of this work is to investigate the use of nonrigid registration for target tracking and dynamic treatment planning, i.e., treatment planning based not on one single CT scan but on multiple CT scans representative of the respiration. Twenty patients were CT scanned at end-inhale and end-exhale. An expert radiation oncologist identified the PTV in both examinations. The two CT data sets per patient were nonrigidly registered using a free-form deformation algorithm based on B-splines. The optimized objective function consisted of a weighted sum of a similarity criterion (Mutual Information) and a regularization factor which constrains the transformation to be locally rigid. Once the transformation was obtained and the registration validated, its parameters were applied to the target only. Finally, the deformed target was compared to the PTV delineated by the radiation oncologist in the other study. The results of this procedure show an agreement between the center of mass as well as volume of the target identified automatically by deformable registration and manually by the radiation oncologist. Moreover, obtained displacements were in agreement with body structure constraints and considerations usually accepted in radiation therapy practice. No significant influence of initial target volume on displacements was found. In conclusion, the proposed method seems to offer the possibility of using nonrigid registrations in radiosurgery treatment planning, even if more cases need to be investigated in order to give a statistical consistency to parameter setup and proposed considerations.

Abstract: In this paper a novel technique is proposed and validated for radiosurgery treatment planning of arteriovenous malformations (AVMs). The technique was developed for frameless radiosurgery by means of the CyberKnife, a nonisocentric, linac-based system which allows highly conformed isodose surfaces to be obtained, while also being valid for other treatment strategies. The technique is based on registration between computed tomography (CT) and three-dimensional rotational angiography (3DRA). Tests were initially performed on the effectiveness of the correction method for distortion offered by the angiographic system. These results determined the registration technique that was ultimately chosen. For CT-3DRA registration, a twelve-parameter affine transformation was selected, based on a mutual information maximization algorithm. The robustness of the algorithm was tested by attempting to register data sets increasingly distant from each other, both in translation and rotation. Registration accuracy was estimated by means of the "full circle consistency test." A registration quality index (expressed in millimeters) based on these results was also defined. A hybrid subtraction between CT and 3DRA is proposed in order to improve 3D reconstruction. Preprocessing improved the ability of the algorithm to find an acceptable solution to the registration process. The robustness tests showed that data sets must be manually prealigned within approximately 15 mm and 20 degrees with respect to all three directions simultaneously. Results of the consistency test showed agreement between the quality index and registration accuracy stated by visual inspection in 20 good and 10 artificially worsened registration processes. The quality index showed values smaller than the maximum voxel size (mean 0.8 mm compared to 2 mm) for all successful registrations, while it resulted in much greater values (mean 20 mm) for unsuccessful registrations. Once registered, the two data sets can be used for CyberKnife treatment planning. Target delineation is performed on 3DRA while dose calculation and DRR generation are performed on CT. In conclusion, a method was developed for using 3DRA images for AVM frameless radiosurgery treatment planning. The method proved to be feasible, robust, and accurate for clinical use. 3DRA can be performed at different times or locations compared to standard, frame based stereotactic angiography. Unlike two-dimensional angiography, 3DRA allows examination of the shape of the AVM and of the surrounding target from any arbitrary point of view during treatment planning. The method can be applied to any case of intermodality registration, is operator-independent, and allows estimation of registration quality. Further research is desirable to improve time resolution in order to distinguish between feeding and draining vessels.

Abstract: A kinematical model of the lower spine was designed and used to obtain a robust estimation of the vertebral rotations during torso movements from skin-surface markers recorded by video-cameras. Markers were placed in correspondence of the anatomical landmarks of the pelvic bone and vertebral spinous and transverse processes, and acquired during flexion, lateral bending and axial motions. In the model calibration stage, a motion-based approach was used to compute the rotation axes and centres of the functional segmental units. Markers were mirrored into virtual points located on the model surface, expressed in the local reference system of coordinates. The spine motion assessment was solved into the domain of extended Kalman filters: at each frame of the acquisition, the model pose was updated by minimizing the distances between the measured 2D marker projections on the cameras and the corresponding back-projections of virtual points located on the model surface. The novelty of the proposed technique rests on the fact that the varying location of the rotation centres of the functional segmental units can be tracked directly during motion computation. In addition, we show how the effects of skin artefacts on orientation data can be taken into account. As a result, the kinematical estimation of simulated motions shows that orientation artefacts were reduced by a factor of at least 50%. Preliminary experiments on real motion confirmed the reliability of the proposed method with results in agreement with classical studies in literature.

Abstract: A novel technique based on evolutionary optimization is proposed here to compute the average rotation centres (RCs) of ball joints linked into kinematic chains using 3D trajectories of the markers attached to the external surface of the corresponding articulated structures. The chain is hierarchically solved by iteratively minimizing the variance of the marker distances from the actual RC through an evolutional strategy method (ESM) from proximal to distal joints. In particular, the technique is compared to the non-rigid sphere-fitting method, recently proposed in literature and implemented through a closed-form solution (CFS), in conditions of random and systematic noise superimposed to the marker coordinates. Results from simulated motions showed that, in case of small range of motion (5 degrees , 10 degrees ) the performance of CFS is really unreliable whereas ESM provided satisfactory accuracy. Error propagation along the kinematic chain was found to be negligible. Also in the case of systematic errors, ESM provides an accuracy that is sensibly better than that of the CFS. As a case study, ESM was applied to the in vivo computation of the RCs of the vertebrae in the lower spine region using a specific marker protocol. A set of spine movements by a normal adult male, recorded by an optoelectronic motion capture system, were processed with the developed method. The variability of the estimated average RCs was small (few millimeters) in agreement with the literature data from cadaveric studies and X-ray imaging.

Abstract: In the field of sports biomechanics and rehabilitation engineering, the possibility of computing, in real time, the angular displacements and derivatives of human joints, from a video of motion sequences, represents an appealing goal. In particular, applications of biofeedback protocols in rehabilitation can benefit from this capability. The focus of the investigation was concerned with the application of biomechanical models, comprising of a kinematic chain and surface envelopes, and state-space filters, to the computation, in real time and with high accuracy, of the angular data and derivatives. By minimising the distances, measured with TV cameras, between the 2D marker projections and the corresponding back-projected markers located on the mannequin, the configuration of the biomechanical model was automatically updated. The use of state-space estimation allowed the computation of smooth derivatives of the orientation data. Owing to the non-linearity of the functions involved, the derivatives of the observation model were obtained through a multidimensional extension of Stirling's interpolation formula. Proper algorithms were developed to cope with the model calibration, initialisation and data labelling. Extensive experiments on real and simulated motions proved the reliability (maximum angular error less than 1 degree, maximum point reconstruction less than 1 mm) of the developed system, which is robust to false matching caused by marker occlusions. Moreover, orientation artifacts due to skin motion can be reduced by a factor of 50%.

Abstract: Global polynomial (GP) methods have been widely used to correct geometric image distortion of small-size (up to 30 cm) X-ray image intensifiers (XRIIs). This work confirms that this kind of approach is suitable for 40 cm XRIIs (now increasingly used). Nonetheless, two local methods, namely 3rd-order local un-warping polynomials (LUPs) and hierarchical radial basis function (HRBF) networks are proposed as alternative solutions. Extensive experimental tests were carried out to compare these methods with classical low-order local polynomial and GP techniques, in terms of residual error (RMSE) measured at points not used for parameter estimation. Simulations showed that the LUP and HRBF methods had accuracies comparable with that attained using GP methods. In detail, the LUP method (0.353 microm) performed worse than HRBF (0.348 microm) only for small grid spacing (15 x 15 control points); the accuracy of both HRBF (0.157 microm) and LUP (0.160 microm) methods was little affected by local distortions (30 x 30 control points); weak local distortions made the GP method poorer (0.320 microm). Tests on real data showed that LUP and HRBF had accuracies comparable with that of GP for both 30 cm (GP: 0.238 microm; LUP: 0.240 microm; HRBF: 0.238 microm) and 40 cm (GP: 0.164 microm; LUP: 0.164 microm; HRBF: 0.164 microm) XRIIs. The LUP-based distortion correction was implemented in real time for image correction in digital tomography applications.

Abstract: SUMMARY: GAAS, Gene Array Analyzer Software supports multi-user efficient management and suitable analyses of large amounts of gene expression data across replicated experiments. Its management framework handles input data generated by different technologies. A multi-user environment allows each user to store his/her own data visualization scheme, analysis parameters used, values and formats of the output data. The analysis engine performs: background and spot quality evaluation, data normalization, differential gene expression analyses in single and multiple replica experiments. Results of expression profiles can be interactively navigated through graphical interfaces and stored into output databases.

Abstract: In sport science, as in clinical gait analysis, optoelectronic motion capture systems based on passive markers are widely used to recover human movement. By processing the corresponding image points, as recorded by multiple cameras, the human kinematics is resolved through multistage processing involving spatial reconstruction, trajectory tracking, joint angle determination, and derivative computation. Key problems with this approach are that marker data can be indistinct, occluded or missing from certain cameras, that phantom markers may be present, and that both 3D reconstruction and tracking may fail. In this paper, we present a novel technique, based on state space filters, that directly estimates the kinematical variables of a virtual mannequin (biomechanical model) from 2D measurements, that is, without requiring 3D reconstruction and tracking. Using Kalman filters, the configuration of the model in terms of joint angles, first and second order derivatives is automatically updated in order to minimize the distances, as measured on TV-cameras, between the 2D measured markers placed on the subject and the corresponding back-projected virtual markers located on the model. The Jacobian and Hessian matrices of the nonlinear observation function are computed through a multidimensional extension of Stirling's interpolation formula. Extensive experiments on simulated and real data confirmed the reliability of the developed system that is robust against false matching and severe marker occlusions. In addition, we show how the proposed technique can be extended to account for skin artifacts and model inaccuracy.

Abstract: In this paper we present two novel techniques, namely a local unwarping polynomial (LUP) and a hierarchical radial basis function (HRBF) network, to correct geometric distortions in XRII images. The two techniques have been implemented and compared, in terms of residual error measured at control and intermediate points, with local and global methods reported in the previous literature. In particular, LUP rests on a locally optimized 3rd degree polynomial applied within each quadrilateral cell on the rectilinear calibration grid of points. HRBF, based on a feed-forward neural network paradigm, is constituted by a set of hierarchical layers at increasing cut-off frequency, each characterized by a set of Gaussian functions. Extensive experiments have been performed both on simulated and real data. In simulation, we tested the effect of pincushion, sigmoidal and local distortions, along with the number of calibration points. Provided that a sufficient number of cells of the calibration grid is available, the obtained accuracy for both LUP and HRBF is comparable to or better than that of global polynomial technique. Tests on real data, carried out by using two different (12 in. and 16 in.) XRIIs, showed that the global polynomial accuracy (0.16+/-0.08 pixels) is slightly worse than that of LUP (0.07+/-0.05 pixels) and HRBF (0.08+/-0.04 pixels). The effects of the discontinuity at the border of the local areas and the decreased accuracy at intermediate points, typical of local techniques, have been proved to be smoothed for both LUP and HRBF.

Abstract: Video-camera systems are widely used in biomechanics and clinical fields to measure the 3D kinematic measurements of human motion. To be used, they need to be calibrated, that is the parameters which geometrically define the cameras have to be determined. It is shown here how this can be achieved by surveying a rigid bar in motion inside the working volume, and in a very short time: less than 15 s on a Pentium III. The exterior parameters are estimated through the coplanarity constraint, the camera focal lengths through the properties of epipolar geometry and the principal points with a fast evolutionary optimisation which guarantees convergence when the initial principal points cannot be adequately estimated. The method has been widely tested on simulated and real data. Results show that its accuracy is comparable with that obtained using methods based on points of known 3D coordinates (DLT): 0.37 mm RMS error over a volume with a diagonal approximately 1.5 m. A preferential absolute reference system is obtained from the same bar motion data and is used to guide an intelligent decimation of the data. Finally, the role that the principal points play in achieving a high accuracy, which is questioned in the computer vision domain, is assessed through simulations.

Abstract: In this paper a novel approach to anatomy knowledge representation is described. The focus of the present research has been on the development of a representational framework where the conceptual level has been implemented by using hierarchical and nonhierarchical conceptual networks. This has allowed handling the demand for multiple views of anatomy (systemic and topographical views). The terminological level of the knowledge representation has been implemented by using a compositional strategy which has avoided the explicit storage of the terms used to express composite concepts. Hierarchical relations and composite concept representations have required supervision of both the inheritance and concept reconstruction. For this purpose heuristic knowledge has been stored in terms of consistency rules in the knowledge base. As proof of the capability of this system, we show how the knowledge base has been used to provide symbolic access to spatial information consisting of a reduced set of images from the Visible Human Dataset.

Abstract: A novel internet-based application is presented which provides access to anatomy knowledge through symbolic modality expressed by keywords taken from controlled or non-controlled terminology. The system is based on a database where anatomical concepts have been organized into a hierarchical framework. Along with term queries that allow retrieving concepts containing or exactly matching the used keyword, the system also provides semantic access to anatomical information. Queries can be setup, which retrieve concepts relying to a particular meaning and sharing a particular relationship. Moreover, the application has the capability to refine the search of the terms by querying the UMLS knowledge server. Anatomical image data have been integrated by using Visible Human Dataset. A set of these images has been indexed according to our anatomical classification and is used inside the application. The system has been implemented through Java client-server technology and works within standard Internet browsers.

Abstract: This paper presents a new method for calibrating a video 3D stereo-photogrammetric system. The external parameters and the focal lengths of the cameras are determined from the epipolar constraint and the principal points are computed through the minimisation of a cost function carried out through an evolutionary optimisation. The method has been made more robust with a deterministic annealing procedure of the search region amplitude. Calibration is carried out by moving a rigid bar, carrying two markers on its extremities, inside the working volume. The distance between the two markers is the only measure required. Tests on real data are reported which show that the obtained accuracy is comparable to the one achieved calibrating with control points of known 3D coordinates.

Abstract: This paper compares the DLT and ILSSC approaches in the geometrical calibration of a photogrammetric stereo-system in terms of accuracy and speed. To come up with an unbiased quantitative evaluation of the accuracy of the algorithms, the concept of reliable estimate has been introduced: the statistical distribution of the accuracy is assessed over different calibration experiments performed with the same data but with different noise distribution and different test sets. Results show that in the simulations where the only error on the two-dimensional points was Gaussian, zero mean, and on real data which were corrected for distortions through polynomial or linear interpolation, the accuracy of the two methods was quite similar. DLT showed more accurate than ILSSC on simulated data with residual distortion errors and on real data which were not corrected for distortions. As far as speed is concerned, a fast triangulation algorithm is associated with ILSSC while the simultaneous solution of two pairs of DLT equations is associated to DLT. The first algorithm is much faster, requiring 113 flops per point versus 259 of DLT; the fast triangulation with DLT parameters does not achieve the same accuracy on the reconstructed three-dimensional position. Taken all together the results suggest that ILSSC can be theoretically considered the best approach to three-dimensional reconstruction, provided that distortions are corrected in advance. The statistical evaluation of the accuracy allows a fair judgement of the performances of the algorithms to be obtained, unbiased by particular distributions of measurement errors and test points.