AU2007209792A1 - Assessing the condition of a joint and preventing damage - Google Patents

Description

AUSTRALIA

Patents Act COMPLETE SPECIFICATION

(ORIGINAL)

Class Int. Class Application Number: Lodged: Complete Specification Lodged: Accepted: Published: Priority Related Art: Name of Applicant: The Board of Trustees of the Leland Stanford Junior University Actual Inventor(s): Eugene J Alexander, Thomas Andriacchi, Philipp Lang, Sandy A Napel Address for Service and Correspondence: PHILLIPS ORMONDE FITZPATRICK Patent and Trade Mark Attorneys 367 Collins Street Melbourne 3000 AUSTRALIA Invention Title: ASSESSING THE CONDITION OF A JOINT AND PREVENTING DAMAGE Our Ref: 809205 POF Code: 476364/227921 The following statement is a full description of this invention, including the best method of performing it known to applicant(s): -1- 6006q ASSESSING THE CONDITION OF A JOINT AND PREVENTING DAMAGE This application is a divisional of Australia patent application no. 2004203105 which is a divisional of Australian patent no. 772012 the contents of both of which are herein incorporated by reference.

This invention was supported in part by a National Institute of Health Grant No. PAR- 97-014, and the government may have rights in this invention.

This application claims the benefit of U.S. Provisional Application Serial No.

60/112,989, filed December 16, 1998.

W ISASKIA~atent Spec\20O4203105div-13 807doc BACKGROUND OF THE INVENTION FIELD OF INVENTION This invention relates to assessing the condition of a joint and the use of the assessment in aiding in prevention of damage to the joint or treatment of diseased cartilage in the joint.

BACKGROUND

Osteoarthritis is the most common condition to affect human joints as well as a frequent cause of locomotor pain and disability. More particularly, osteoarthritis (OA) of the knee occurs in a substantial portion of the population over the age of fifty.

W \SASKIA\Patenl Spec\2004203 1O5v-13 8 07 doc In spite of its societal impact and prevalence, however, there is a paucity of information on the factors that cause osteoarthritis to progress more rapidly in some individuals and not in others. Previously considered a "wear and tear" degenerative disease with little opportunity for therapeutic intervention, osteoarthritis is now increasingly viewed as a dynamic process with potential for new pharmacologic and surgical treatment modalites such as cartilage transplantation, osteochondral allo- or autografting, osteotomies and tibial corticotomies with angular distraction.

However, the appropriate deployment and selection of treatment interventions for OA is dependent on the development of better methods for the assessment of the condition of a patient's joint and the degeneration process.

W:\SASKIA\Palent Spec\2004203105dn-13 8 07 Ooc There is, therefore, a need for improved methods for examining the factors that influence as well as quantification of the progression of the disease.

Magnetic resonance imaging (MRI) is an accurate non-invasive imaging technique for visualization of articular cartilage in osteoarthritis, particularly in knees. However, current MRI techniques cannot provide information on the relationship between the location of the cartilage loss and variations in the load bearing areas during the walking cycle. This information is important since it has been shown that dynamic loads during walking are related to the progression of knee OA. Thus, the ability to locate cartilage defects or areas of cartilage thinning relative to the load bearing areas of the knee could be valuable in evaluating factors influencing the progression of osteoarthritis.

W: SASKIA\Patent Spec\200420310Sdiv-13 8 07 doc SUMMARY OF THE INVENTION This invention relates to assessing the condition of a joint of a mammal, particularly a human subject, using the assessment to treat and monitor the subject as needed for cartilage degeneration problems. While the numerous aspects of the invention are useful for joints generally, they are particularly suited for dealing with the human knee.

Some aspects related the static images and degeneration patterns of a cartilage, while others relate to the interaction of such images and patterns to provide a better means of assessing the condition of a cartilage.

W \SASKIAPatenI Spec\2004203105dv 13 B 07 doc According to an aspect of the present invention there is provided a method for providing a biochemically-based map of joint cartilage of a mammal, wherein the joint includes cartilage and associated bones on either side of the joint, which method includes: measuring a detectable biochemical component throughout the cartilage; determining the relative amounts of the biochemical component throughout the cartilage; mapping the amounts of the biochemical component in three dimensions through the cartilage; and determining the areas of abnormal joint cartilage by identifying the areas having altered amounts of the biochemical component present.

W \SASKIA\Palent Spec\2004203105d-13 8 07 doc Once a map is obtained, it can be used in assessing the condition of a cartilage at an initial time and over a time period. Thus, the biochemical map may be used in the method aspects of the invention in a manner similar to the cartilage thickness map.

Another aspect of the invention is a biochemical map of a cartilage that comprises a three-dimensional representation of the distribution of the amount of the biochemical component throughout the cartilage.

Other aspects of the invention may be apparent upon further reading the specification and claims of the patent application.

W \SASKIA\Palent Spec2004203105div-13 8 07 doc WO 00/35346 PCT/US99/30265 F' DESCRIPTION OF THE DRAWINGS tb3 In the accompanying drawings: 5 Figure 1 shows an overview schematic representation of some aspects of the invention of this application.

(Ni Figure 2 shows a DEFT pulse sequence.

O Figure 3 shows the signal levels for cartilage and synovial fluid with RARE and DEFT pulse sequences, both TE 14 miliseconds.

Figure 4 shows the mean contrast to noise ratio (CNR) of cartilage to joint fluid for various MRI pulse sequences.

Figure 5 shows the mean contrast for cartilage and joint fluid for various MRI pulse sequences.

Figure 6 shows a DEFT acquisition using non-selective refocusing pulses to maximize the SNR efficiency and a partial K- Echo-Plainer acquisition gradients in order to minimize the required scan time for 3D volume.

Figure 7 shows four sample images acquired with a DEFT pulse sequence combined with a partial K- Echo-Plainer acquisition in order to provide efficient 3D coverage.

Figures 8A and 8B show a 3-point Dixon GRE image of the articular cartilage of medial fermorotibial compartment in a normal 35-year old volunteer. Figure 13A has the subject in supine position and Figure 13B has the subject in an upright position.

Figures 9A -9C show patient position and application of imaging coil and tracker coil for kinetic MR imaging of the knee. Patient is in upright weight-bearing position for active flexion and extension study of the knee.

Figure 9B is a 2D cartilage thickness map demonstrating abrupt decrease in cartilage thickness in the area of the defect (arrows). The A thickness between the neighboring pixels can be use to define the borders of the cartilage defect. Note defused cartilage thinning in the area enclosed by the asterisks 9 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 Figures O10A-1OC show a 3D surface registration of femoral condyles based on T1 weighted Spin-Echo MR images. Figure 6A is a baseline with a knee and neutral position. 6B is a follow-up with knee and external rotation with a 3D view that is the identical to the one used in 6A but the difference in knee rotation is apparent. In Figure 6C, transformation and re-registration of Scan B into the object coordinate system of Scan A shows the anatomic match to A is excellent.

Figure 11 A shows a 2D cartilage thickness map where a proton density fast spin-echo MR image demonstrates a focal cartilage defect in the posterior lateral fermoral condyle (black arrows). White arrows indicate endpoints of the thickness map.

Figure 12 shows the anatomic coordinate system in the femur and in the tibia.

Figure 13 shows calculation of the anatomic coordinate system from palpable bony landmarks.

Figure 14 shows additional marker names and locations for MR to optical cross registration.

Figure 15 shows the marker names and locations for the standard point-cluster technique protocol.

Figure 16 shows the error in the tibial location estimate for the rigid body model and the intrical deformation correction technique.

Figure 17 shows the error in tibial orientation estimate for the rigid body model and the interval deformation correction technique.

Figure 18A 181 show functional joint imaging.

Figure 19 shows the super imposition of the tibiofemoral contact line onto the 3D cartilage thickness map.

Figure 20 shows the determination of the natural line of curvature as the cutting plain is rotated about the transepicondyear reference, the cartilage-plain intersection results in a curve.

Figure 21 shows the determination of the tibiofemoral contact line through the SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCTIUS99/30265 proximity detection and approach algorithm.

Figures 22A and 22B show a 2D MRI (3D SPGR) and 3D cartilage thickness map.

Figures 23A E show the matching of 3D thickness maps generated from MR images obtained with a knee neutral position and external rotation.

SPECIFIC DESCRIPTION Overview Figure 1 is a schematic overview of some of the various aspects of the invention.

While a complete description of the many aspects of the invention is found in the specification and claims, the schematic overview gives some of the broad aspects of the invention.

This invention relates to assessing the condition of a joint in a mammal. One aspect is a method for such an assessment. The assessment can be done using internal images, or maps, of the cartilage alone or in combination with a movement pattern of the joint.

If used alone, a map obtained at an initial time is compared with a map obtained at a later time to provide a view of the change in cartilage over time. Another aspect is a method is comparing the movement pattern for a joint of a subject being studied with the cartilage degeneration pattern of the subject, then determining the relationship between the movement pattern and the degeneration pattern. If, in determining the relationship between the two patterns, one finds that the movement pattern has caused the degeneration pattern or will continue to adversely affect the degeneration pattern, therapy can be prescribed to minimize the adverse effects, such as further degeneration or inflammation.

In overview, some of the systems and methods of this invention are illustrated by the flow chart in the attached Figure 1. Figure 1 is based on the full range of processes, preferably applied to a knee and surrounding cartilage.

In Figure 1, the first step 10 represents obtaining an image of the cartilage itself. This is typically achieved using MRI techniques to take an image of the entire knee and 11 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 then, optionally, manipulating "subtracting out" or "extracting") the noncartilage images as shown in step 12. Non-cartilage images typically come from bone and fluid. Preferably, the MRI is taken using external markers to provide reference points to the MRI image (step 11).

If the cartilage is imaged with a 2D MRI acquisition technique, the resulting stack of 2D images so obtained can be combined into a 3D image, as indicated in step 14. A preferred alternative is to use 3D MRI acquisition techniques to acquire a 3D image directly. In either case, the same "non-cartilage image extraction techniques referred to in step 12 can be used.

With a full 3D image captured, various "maps" or displays of the cartilage can be constructed to give a cartilage degeneration pattern. This is represented by step 16.

One such display can, for example, be a color-coding of a displayed image to reflect the thickness for the cartilage. This will allow easy visual identification of actual or potential defects in the cartilage.

Together with or independently of the cartilage imaging, and as represented by parallel step 20, a 3D image of the knee joint is taken, again preferably using MRI.

Many of the same techniques as applied in steps 10 to 14 are used to do this.

However, as illustrated by sub-step 22, it is useful to define and register a skinexternal frame of reference around the joint. This is achieved by placing fiduciary markers on the skin around the outside of the knee (step 22) prior to taking the image.

In addition to an image extraction technique (as described above in step 12), an image is manipulated to enhance the image of the position of the markers (step 24). The resulting manipulated image is used to give a 3D image of the joint and associated bones (step 26).

With the markers in place, and as shown by step 30, an additional set of markers is placed on the skin along the outside of the leg, and an external image of the limb is obtained. Using at least two cameras, images are then taken of the subject in a static state. In addition, images are also taken of the subject while moving. This is shown collectively by step 32. The images obtained are then processed to relate the 12 SUBSTITUTE SHEET(IULE 26) movement of the skin relative to the bone. In addition, certain calculations are performed, for example, the center of mass is calculated. These manipulations are shown in Step 34. Further, as the fiduciary markers are still in place during the video image capture, a correlation between the fiduciary and the additional set of markers can be made. This is shown in step 36.

Once this marker-to-marker correlation is made, the static 3D image of the joint (with associated fiduciary markers) and the movement images of the leg bones (also with fiduciary markers in place) can be combined. The fiduciary markers, therefore, serve as baseline references. The combination (step 40) of 3D cartilage image (from step 14), 3D knee joint image (step 26), and the moving leg co-ordinates (step 34) will, after appropriate corrections, result in a displayable, 3D motion image of the joint moving as per step 46.

The moving images, showing the contact areas of the knee joint can be used in conjunction with the various "maps" or displays generated at step 16 to provide a visual indication of potential or actual cartilage defects and help in determining their relation between movement and degeneration patterns. This is shown in step 48.

Furthermore, as the various images are supported by actual mathematical quantification, real measurements (such as cartilage thickness) can be taken and compared with later or earlier measurements and/or imaging. This allows the tracking of the progression of a defect, or conversely, continued tracking of healthy cartilage.

This aids a health worker in providing therapy for the patients. The method allows monitoring and evaluation of remedial actions as well as possible treatment prescriptions.

Thus, there is disclosed, for example, a method to examine the relationship between articular cartilage morphology and the functional load bearing areas of a knee joint measured during movement. The method includes enhanced imaging techniques to reconstruct the volumetric and biochemical parameters of the articular cartilage in three dimensions; and a method for in vivo kinematic measurements of the knee. The kinematic measurement permits direct in vivo measurements of complete six-degrees of freedom motion of the femur or the tibia or associated bones during WO 00/35346 PCT/US99/30265 Snormal activities. This permits the study of load bearing of articular cartilage during 0 movement. In particular, this method can aid in locating cartilage defects relative to the changing load bearing areas of the knee joint during daily activities. While the various aspects of the invention are useful in mammals generally, they are particularly useful for human patients.

C Obtaining the Cartilage Degeneration Pattern Imaging Articular Cartilage In general, the joint of a patient is that place of union, more or less movable, between two or more bones. A joint comprises cartilage and other elements such as the accompanying bones on either side of the joint, fluid, and other anatomical elements.

Joints are classified into three general morphological types: fibrous, cartilaginous, and synovial. This invention is particularly useful for assessing synovial joints, particularly the knee.

In obtaining an image of the cartilage of a joint in a mammal, a number of internal imaging techniques known in the art are useful for electronically generating a cartilage image. These include magnetic resonance imaging (MRI), computed tomography scanning (CT, also known as computerized axial tomography or CAT), and ultrasound imaging techniques. Others may be apparent to one of skill in the art.

MRI techniques are preferred.

MRI, with its superior soft tissue contrast, is the best technique available for assessing tissue and its defects, for example articular cartilage and cartilage lesions, to obtain a cartilage degeneration can provide morphologic information about the area of damage. Specifically, changes such as fissuring, partial or full thickness cartilage loss, and signal changes within residual cartilage can be detected.

The reason MR imaging techniques are particularly suitable for cartilage is because they can provide accurate assessment of cartilage thickness, demonstrate internal cartilage signal changes, evaluate the subchondral bone for signal abnormalities, and demonstrate morphologic changes of the cartilage surface.

14 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCTfUS99/30265 MRI provides several important advantages over other techniques in this invention.

tb One advantage is good contrast between cartilage, bone, joint fluid, ligaments, and muscle in order to facilitate the delineation and segmentation of the data sets. Another 5 is the coverage of the entire region of interest in a single scan within acceptable acquisition times. For a brief discussion of the basic MRI principles and techniques, N see MiRI Basic Principles and Applications, Second Edition, Mark A. Brown and Richard C. Semelka, Wiley-Liss, Inc. (1999).

10 MRI employs pulse sequences that allow for better contrast of different parts of the Sarea being imaged. Different pulse sequences are better fitted for visualization of different anatomic areas, for example, hyaline cartilage or joint fluid. More than one pulse sequence can be employed at the same time. A brief discussion of different types of pulse sequences is provided below.

High Resolution 3D MRI Pulse Sequences Routine MRI pulse sequences available for imaging tissue, such as cartilage, include conventional T1 and T2-weighted spin-echo imaging, gradient recalled echo (GRE) imaging, magnetization transfer contrast (MTC) imaging, fast spin-echo (FSE) imaging, contrast enhanced imaging, rapid acquisition relaxation enhancement, (RARE) imaging, gradient echo acquisition in the steady state, (GRASS), and driven equilibrium Fourier transform (DEFT) imaging. As these imaging techniques are well known to one of skill in the art, e.g. someone having an advanced degree in imaging technology, each is discussed only generally hereinafter. While each technique is useful for obtaining a cartilage degeneration pattern, some are better than others.

Conventional Ti and T2-Weighted Spin-Echo Imaging Conventional T1 and T2-weighted MRI depicts articular cartilage, and can demonstrate defects and gross morphologic changes. Ti-weighted images show excellent intra-substance anatomic detail of hyaline cartilage. However, T -weighted imaging does not show significant contrast between joint effusions and the cartilage surface, making surface irregularities difficult to detect. T2-weighted imaging SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 demonstrates joint effusions and thus surface cartilage abnormalities, but since some components of cartilage have relatively short T2 relaxation times, these are not as well depicted as other preferred imaging.

Gradient-Recalled Echo Imaging Gradient-recalled echo imaging has 3D capability and ability to provide high resolution images with relatively short scan times. Fat suppressed 3D spoiled gradient echo (FS-3D-SPGR) imaging has been shown to be more sensitive than standard MR imaging for the detection of hyaline cartilage defects in the knee.

Magnetization Transfer Contrast Imaging Cartilage, as well as other ordered tissues, demonstrate the effects of magnetization transfer. Magnetization transfer imaging can be used to separate articular cartilage from adjacent joint fluid and inflamed synovium.

Fast Spin-Echo Imaging Fast spin-echo imaging is another useful pulse sequence to evaluate articular cartilage.

Incidental magnetization transfer contrast contributes to the signal characteristics of articular cartilage on fast spin-echo images and can enhance the contrast between cartilage and joint fluid. Sensitivity and specificity of fast spin-echo imaging have been reported to be 87% and 94% in a study with arthroscopic correlation.

Contrast Enhanced Imaging The use of gadolinium for imaging of articular cartilage has been applied in several different forms. Direct magnetic resonance (MR) arthrography, wherein a dilute solution containing gadolinium is injected directly into the joint, improves contrast between cartilage and the arthrographic fluid. Indirect MR arthrography, with a less invasive intravenous injection, can also been applied. Gadolinium enhanced imaging has the potential to monitor glycosaminoglycan content within the cartilage, which may have implications for longitudinal evaluations of injured cartilage.

16 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 Driven Equilibrium Fourier Transform Another 3D imaging method that has been developed is based on the driven equilibrium fourier transform (DEFT) pulse sequence Patent No. 5,671,741), and is specifically designed for cartilage imaging. DEFT provides an effective tradeoff between T2/T1 weighting and spin density contrast that delineates the structures of interest in the knee. Contrast-to-noise ratio between cartilage and joint fluid is greater with DEFT than with spoiled gradient echo (SPGR). DEFT is an alternative approach to SPGR. DEFT contrast is very well suited to imaging articular cartilage. Synovial fluid is high in signal intensity, and articular cartilage intermediate in signal intensity. Bone is dark, and lipids are suppressed using a fat saturation pulse. Hence, cartilage is easily distinguished from all of the adjacent tissues based on signal intensity alone, which will greatly aid segmentation and subsequent volume calculations.

The basic DEFT pulse sequence is shown in Fig. 2. A conventional spin echo pulse sequence was followed by an additional refocusing pulse to form another echo, and then a reversed, negated, excitation pulse to return any residual magnetization to the +z axis. This preserved the magnetization of longer T2 species, such as synovial fluid.

Typical MRI parameters for cartilage are a T1-relaxation time of 900 Milliseconds (ms) and a T2-relaxation time of 40 ms, while synovial fluid has a TI-relaxation time of 3000 ms and a T2-relaxation time of 200 ms. In addition, synovial fluid has a greater proton density than cartilage. The signal levels of cartilage and synovial fluid were plotted in Fig. 3 for a RARE pulse sequence and for DEFT, and show that DEFT maintains excellent contrast for any relaxation time It achieves this contrast while maintaining a signal-to-noise ratio (SNR) efficiency (SNR/ (Tacquisition)) that is equal to or better than other methods with much lower contrast, such as TI-weighted

GRASS.

DEFT was compared with a fast spin-echo (FSE), a gradient-echo (GRE), and a spoiled gradient-echo (SPGR) sequence with parameters similar to the ones published by Disler et al. The patella was scanned in 10 normal volunteer knees using a whole-body system (GE Signa) with a 3 inch surface coil. All images were acquired 17 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 with field of view (FOV) 10x10 cm, matrix 256x256 elements, slice thickness 4 mm using fat-saturation. DEFT (400/15 [TR/TE in msec], 2 NEX (number of excitations), FSE (3500/15, echo train length [ETL] 8, 2 NEX (number of excitations), FSE (3500/15, ETL 4, 2 NEX), GRE (400/20, 30, 2 NEX), and SPGR (50/15, 30* [flip angle], 2 NEX) images were obtained. Contrast-to-noise ratios (CNR) between cartilage and joint fluid were calculated as: CNR I (SIjoint Fluid SIcartiage) SIBackground Noise I Eq. 1 Contrast between cartilage and joint fluid was calculated as: C I [(SIjoint Fluid SIcartilage) SIJoint Fluid X 100 Eq. 2.

In the equations SI is signal intensity. DEFT demonstrated greater contrast-to-noise ratio and contrast between cartilage and joint fluid than SPGR, GRE, and FSE sequences (Figs. 4 Cartilage had intermediate signal intensity with DEFT, while joint fluid was high in signal intensity. The difference in CNR between DEFT and SPGR was statistically significant (p<0.001). Cartilage morphology, i.e. cartilage layers, were consistently best delineated with the DEFT sequence. At the resolution used in this study, FSE sequences suffered from image blurring. Blurring was improved with ETL 4 when compared to ETL8; nonetheless, even with ETL 4, cartilage morphology seen on FSE images was inferior to the DEFT sequence. In light of these results, DEFT imaging is a preferred MRI technique.

Another Application ofDEFT DEFT was combined with a partial k-space echo-planar data acquisition. This pulse sequence is illustrated in Fig. 6 above. A slab selective pulse in z defines the imaging volume, which is then resolved with phase-encoding gradients in the y and z axes, and an oscillating EPI gradient in the x axis.

Example images acquired with this approach are shown in Fig. 7. This case was optimized for resolution, in order to image the patellar cartilage. The EPI readout 18 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 acquired 5 echoes for each DEFT sequence. Partial k-space acquisition collected only of the data along the x-axis. Correction for the missing data was performed using a homodyne reconstruction. The image matrix was 192x192x32, with a resolution of 0.5x0.5x2.5 mm, resulting in a 10x10x8 cm FOV. The echo time TE was 22 ms, and the TR was 400 ms. Fat was suppressed with a fat presaturation pulse.

The total scan time for this acquisition was 5 minutes.

Additional image studies that can be performed using this approach may require greater spatial coverage, but one can permit slightly less spatial resolution, and a longer scan time similar to the one used with the 3D SPGR approach. If one relaxes the resolution to 0.75x0.75x1.5 mm, and doubles the z slab thickness and z phase encodes, the result will be a FOV of 15x15x16 cm, and a total scan time of approximately 15 minutes, which exactly fits the desired scan protocol. Similar to the 3D SPGR acquisition, one can acquire a first 3D DEFT scan in the sagittal plane with fat saturation. The 3D DEFT acquisition can then be repeated without fat saturation using the identical parameters and slice coordinates used during the previous acquisition with fat saturation. The resultant non-fat-saturated 3D DEFT images can be used for 3D rendering of the femoral and tibial bone contours.

In summary, Driven Equilibrium Fourier Transform is a pulse sequence preferred for cartilage imaging that provides higher contrast-to-noise ratios and contrast between cartilage and joint fluid than SPGR, GRE, and FSE sequences. Cartilage morphology is better delineated with DEFT sequences than with SPGR, GRE, and FSE images.

The combination of high anatomic detail and high cartilage-joint fluid CNR and contrast may render this sequence particularly useful for longitudinal studies of cartilage in patients with osteoarthritis.

Fully Weight-Bearing MRI of the Knee Joint Using a Vertically Open MRI Unit Vertically open MRI units can provide the opportunity to study the knee joint in true weight-bearing position with the patient placed in upright position. A vertically open MRI unit (GE Signa SP 0.5T) exists at Stanford University. The vertically open MRI unit can be used to develop techniques for kinematic joint imaging under weightbearing conditions as a means of diagnosing joint instability.

19 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 F,1 Upright Cartilage Imaging A cartilage sensitive MRI scans was obtained in the open magnet using a three-point 5 Dixon gradient-echo sequence in supine and upright positions. A representative example is demonstrated in Fig. 8.

SA Representative Example of MR Imaging is described below: 10 A MR image can be performed using a whole body magnet operating at a field 0 strength of 1.5 T (GE Signa for example, equipped with the GE SR-120 high speed gradients [2.2 Gauss/cm in 184psec risetimes]). Prior to MR imaging, external markers filled with Gd-DTPA (Magnevist®, Berlex Inc., Wayne, doped water (T1 relaxation time approximately 1.0 sec) can be applied to the skin around the knee joint and optionally at the same positions used for gait analysis in a biomotion laboratory (discussed below). The external markers can be included in the field of view of all imaging studies. Patients can be placed in the scanner in supine position.

After an axial scout sequence, coronal and sagittal T1-weighted images of the femur can be acquired using the body coil (spin-echo, TR=500msec, TE=15msec, 1 excitation (NEX), matrix 256x128 elements, field of view (FOV) 48 cm, slice thickness 7 mm, interslice spacing 1 mm). The scanner table can then be moved to obtain coronal and sagittal images of the knee joint and tibia using the same sequence parameters. These T1-weighted scans can be employed to identify axes through the femur and tibia which can be used later for defining the geometry of the knee joint.

The knee can then be placed in the knee coil with the joint space located in the center of the coil. The knee can be secured in the coil with padding. Additionally, the foot and ankle region can be secured in neutral position to the scanner table using adhesive tape in order to minimize motion artifacts. A rapid scout scan can be acquired in the axial plane using a gradient echo sequence (GRASS, 2D Fourier Transform (2DFT), TR=50msec, TE=10msec, flip angle 400, 1 excitation (NEX), matrix 256x128 elements, field of view (FOV) 24 cm, slice thickness 7 mm, interslice spacing 3 mm).

This scout scan can be used to demonstrate the position of the knee joint space in the coil and to prescribe all subsequent high resolution imaging sequences centered over the joint space. Additionally, using the graphic, image based sequence prescription SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 mode provided with the scanner software, the scout scan can help to ensure that all external markers around the knee joint are included in the field of view of the high b1 resolution cartilage sensitive MR sequences.

S 5 There are several issues to consider in obtaining a good image. One issue is good contrast between cartilage, bone, joint fluid, ligaments, and muscle in order to Ci facilitate the delineation and segmentation of the data sets. Another is the coverage of both condyles of the knee in a single scan within acceptable acquisition times. In 0 addition, if there are external markers, these must be visualized. One way to address r- 10 these issues is to use a three-dimensional spoiled gradient-echo sequence in the O sagittal plane with the following parameters (SPGR, 3DFT, fat-saturated, TE=5msec, flip angle 400, 1 excitation (NEX), matrix 256x160 elements, rectangular FOV 16x12 cm, slice thickness 1.3 mm, 128 slices, acquisition time approximately 15 min). Using these parameters, one can obtain complete coverage across the knee joint and the external markers both in mediolateral and anteroposterior direction while achieving good spatial resolution and contrast-to-noise ratios between cartilage, bone and joint fluid (Figs. 8 and The fat-saturated 3D SPGR sequences can be used for rendering the cartilage in three dimensions (see description below). The 3D SPGR sequence can then be repeated in the sagittal plane without fat saturation using the identical parameters and slice coordinates used during the previous acquisition with fat saturation. The resultant non-fat-saturated 3D SPGR images demonstrate good contrast between low signal intensity cortical bone and high signal intensity bone marrow thereby facilitating 3D rendering of the femoral and tibial bone contours. It is to be understood that this approach is representative only and should not be viewed as limiting in any way.

Magnetic Resonance Imaging Vertically Open Magnet Upright weight-bearing MR imaging of the knee can be performed using a 0.5 T vertically open MR unit (GE Signa SP, General Electric, Milwaukee, WI) and a MR tracking system. Prior to MR imaging, external markers filled with Gd-DTPA (Magnevist®, Berlex Inc., Wayne, doped water (T1 relaxation time approximately 1.0 sec) can be applied to the skin around the knee joint at the same positions optionally used for gait analysis in the biomotion laboratory (discussed 21 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 below). The subject can be placed in upright position inside the magnet (see Fig. 9).

The knee joint can be perpendicular to the main magnetic field. A flexible transmit/receive coil can be positioned around the knee and secured to the knee with elastic straps. The external MR tracking coil can be taped to the skin overlying the tibial tuberosity. For coverage across the entire knee, a 2DFT fast spin echo pulse sequence can be acquired in the sagittal plane (FSE, TR=4000 msec, TE=25 msec, bandwidth 7.8 kHz, echo train length 8, 3 excitations, slice thickness 4 mm, interslice spacing 0.5 mm, matrix 256 x 192 elements, field of view 24 cm). For rapid scan acquisition with scan plane tracking, a fast single slice gradient-echo pulse sequence can be acquired in the sagittal plane or in the axial plane (GRASS, TR=14 msec, msec, flip angle 40 degrees, bandwidth 32 kHz, 1 excitation, slice thickness 4 mm, matrix 256 x 128 elements, field of view 20 cm, temporal resolution 2 sec/image). A field of view of 20 cm can be chosen in order to achieve sufficient anatomic coverage in superoinferior direction across the knee joint. Using this technique, one can assess tibiofemoral motion during knee flexion and extension.

Volumes of Interest (VOI) The invention allows a health practitioner to determine cartilage loss in a reproducible fashion and thus follow the progression of a cartilage defect over time.

In one embodiment of the invention, one can use a 2D or a 3D surface detection technique to extract the surface of the joint, e.g. the femoral condyles, on both baseline and follow-up scans. For example, a TI-weighted spin-echo sequence can be used for surfaces extraction of the femoral condyles. The TI-weighted spin-echo sequence provides high contrast between low signal intensity cortical bone and high signal intensity fatty marrow. For detection of the surface of the femoral condyles, a step-by-step problem solving procedure, an algorithm, can convolve a data set with a 3D kernel to locate the maximum gradient location. The maximum gradient location corresponds to the zero crossing of a spatial location. When the kernel is designed properly, then there will be only one zero crossing in the mask. Thus, that zero crossing is the surface. This operation is preferably three-dimensional rather than two-dimensional. The surface of the joint, e.g. the femoral condyles, on the baseline scan can be registered in an object coordinate system A. The surface of the 22 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 joint, e.g. the femoral condyles, on the follow-up scan can be registered in an object coordinate system B. Once these surfaces have been defined, a transformation B to B' can be performed that best matches B' with A. Such transformations can, for example, be performed using a Levenberg Marquardt technique. Alternatively, the transformations and matching can be applied to the cartilage only. The same transformation can be applied to the cartilage sensitive images on the follow-up scan in order to match the cartilage surfaces.

Using the 3D surface registration of the joint on the baseline scan and resultant object coordinate system A, one can place volumes of interest over the area of a cartilage defect seen on the cartilage sensitive images. For example, in the knee joint, the size of the targeted volumes of interest can be selected to exceed that of the cartilage defect in anteroposterior and mediolateral direction, e.g. by 0.5 to 1 cm. If the defect is located high on the femoral condyle or in the trochlear region, the targeted VOI can be chosen so that its size exceeds that of the cartilage defect in superoinferior and mediolateral direction. The third dimension of the targeted VOI (parallel to the surface normal of the cartilage) can be fixed, for example at 1 cm. VOI size and placement can be manual or automatic on the baseline study. Once the targeted VOI has been placed on the image using visual or automated computer control, the 3D coordinates of the targeted VOI relative to the 3D contour of the joint and object coordinate system A can be registered and saved. On follow-up studies, e.g. scans inadvertently obtained with slightly different patient position, the 3D surface of the joint is registered to match the orientation of the baseline scan and the targeted VOI is then automatically placed on the joint using object coordinate system B' and the coordinates saved on the baseline study. Cartilage volume within the targeted VOI on baseline and follow-up studies can, for example, be determined using standard thresholding and seed growing techniques.

Reference markers When obtaining the MR images for use in this invention, whether the MR is of cartilage or of bone, external reference markers are placed on the skin around the joint of the subject being imaged. The external marker is designed not only to show up in the MRI, but also to show up if an external image of the joint is obtained. The 23 SUBSTITUTE SHEET (RULE 26) importance and value of such unique reference markers will be discussed in more detail hereinafter.

Thus, one embodiment is a skin reference marker that can be used in the assessment of the condition of a joint of a human. Multiple skin reference markers can be placed upon one or more limbs of a patient prior to internal imaging and external imaging.

Each skin reference marker comprises a material detectable by an imaging technique, a container for the material in which the container preferably has multiple surfaces, a means for affixing the container to the skin an adhesive placed on at least one surface of the container in an amount sufficient to adhere the container to the skin of a human), and a reflective material (preferably retro-reflective) placed on another surface of the container located away from the adhesive. Several imaging techniques can be used that are able to detect the marker. For example, magnetic resonance imaging is preferred, but, ultrasound, or X-ray are also useful. In the case of X-ray, further manipulations must be performed in which multiple X-ray images are assimilated by a computer into a 2 dimensional cross-sectional image called a Computed Tomography (CT) Scan. The material detectable by an imaging can be either in a liquid form or a solid form. The material can be any imaging contrast agent or solution, e.g. a paramagnetic material. The material can be a lanthanide, such as one belonging to the yttrium group of rare earth metals. More specifically, the material can be gadolinium. The shape of the container can be any shape allowing it to be placed on the skin of a human. For example, it can be cubical, spherical, elliptical, discoid or cylindrical. The size of the container can be any size, but optimally a size allowing it to be recorded by an imaging machine. The longest dimension of the container can be up to 5.0 cm, but preferably is about 0.25 to 2.0 cm. The reflective or retro-reflective material can be any material that is able to reflect light directly back to the source of the light so that the position of the reference marker is captured by the opto-electrical recording means, e.g. a video camera. 3M Corporation makes several retro-reflective materials.

Manipulating Non-Cartilage Images W SASKIAPatent Spec200420310l5dv- 13 8 07 doc WO 00/35346 PCT/US99/30265 Once a magnetic resonance image is obtained, it can be manipulated to improve the image by reducing unwanted, non-cartilage images.

Segmentation To prepare the data set for 3D rendering, the cartilage can be image by image using a signal-intensity-based threshold combined with a seed growing technique. The femoral, tibial, and patellar cartilage can be segmented separately based on the fatsaturated 3D SPGR or 3D DEFT sequence. Manual disarticulation is performed by outlining the cartilage contour in areas where the signal intensity of the articular cartilage is similar to that of adjacent structures. The contours of the femoral, tibial, and patellar bone are segmented separately using the non-fat-saturated 3D SPGR or 3D DEFT sequence. Segmentation software allows for manual editing of cartilage thickness maps and cartilage defects detected using the above embodiments. In this fashion, the operator can correct erroneous detection of cartilage defects in areas where the cartilage may be naturally thinner, e.g. at the inside of the condyles near the notch. Such software includes seed-growing algorithms and active-contour algorithms that are run on standard PC's. A sharp interface is present between the high signal intensity bone marrow and the low signal intensity cortical bone thereby facilitating seed growing. Fat-saturated and non-fat-saturated 3D sequences can be acquired with the same field of view, slice thickness and slice positions, thereby enabling super imposition and cross registration of any resultant 3D renderings of the femoral, tibial, and patellar cartilage and bone. External reference markers aid in registering the 3D data in the same object coordinate system. 3D maps of cartilage thickness can be generated using several different techniques. One representative, but not limiting, approach uses a 3D surface detection technique which is based on a 2D edge detector (Wang-Binford) that has been extended to 3D. This surface detection technique can generate surface points and their corresponding surface normal. To smooth the contour, the program samples 25 percent of the surface points and fits a cubic spline to the sample points. The program can compute the curvature along sample spline points and find two sample points that have the maximum curvature and are separated by about half the number of voxels on the contour. These points partition the spline into two subcontours. For each subcontour, the program can compute the average distance between the points and the center of the mass. The SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCTIUS99/30265 Sprogram can designate the subcontour with the smaller average distance as the inner cartilage surface and the other subcontour as the outer cartilage surface (OCS). The Sintersect between the inner cartilage surface (ICS) (located at the subchondral bone interface) and the outer cartilage surface with the surface normal can be used to 5 compute the 3D thickness of the articular cartilage on a pixel-by-pixel basis.

Creating A Three Dimensional (3D) Image of the Cartilage Three Dimensional Geometric Model Generation N After the 3D image of cartilage and the 3D image of joint with bones (as discussed hereinafter), are obtained, for example, the set of segmented two dimensional MR images can be transformed to a voxel representation using a computer program developed in the AVS Express (Advanced Visual Systems, Inc., Waltham, MA).

Every voxel has a value of zero if it is not within an object of interest or a value ranging from one to 4095, depending on the signal intensity as recorded by the 1.5 T MR. An isosurface can then be calculated that corresponds to the boundary elements of the volume of interest. A tesselation of this isosurface is calculated, along with the outward pointing normal of each polygon of the tesselation. These polygons are written to a file in a standard graphics format (Virtual Reality Modeling Language Version 1.0: VRML output language).

Visualization Software One choice for the software program used to assess the cartilage degeneration pattern, the bones of the joint, and the motion pattern of the patient is a user controllable 3D visual analysis tool. The program reads in a scene, which scene consists of the various 3D geometric representations or "actors" (for example, VRML files of the tibia, tibia cartilage, femur, femoral cartilage), the static relationship transformations between these actors, and, if available, sequence of transformations describing how these actors move with respect to each other as the patient performs some activity, such as walking, jogging, etc.

The program allows the user, through the use of the mouse and/or keyboard, the ability to observe the scene from arbitrary angles; to start and stop the animation derived from the motion profiles and to observe the contact line and any cartilage 26 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 lesions while the animation is running. Additionally, the user may derive quantitative information on the scene through selecting points with the mouse.

The software program is written in the CTT computer language and is compiled to run on both Silicon Graphics Workstations and Windows/Intel personal computers.

Cartilage thickness maps Cartilage thickness is determined by several methods. One example is detecting the locations of the bone- cartilage and the cartilage- joint fluid interface along the surface normal using the same edge detector described below, and subtracting them.

This procedure can be repeated for each pixel located along the bone cartilage interface. The x, y, and z position of each pixel located along the bone cartilage interface is registered on a 3D map or multiple 2D maps and thickness values are translated into color values. In this fashion, the anatomic location of each pixel at the bone cartilage interface can be displayed simultaneously with the thickness of the cartilage in this location.

The edge detector produces accurate surface points and their corresponding surface normal. The detector is applied to the baseline and the follow-up data set. For the baseline data set, both the surface points and surface normals are used to form locally supporting planes (for each voxel). These planes can form an approximated surface for the baseline skeletal site. As for the follow-up data set, the surface points can be matched in the registration procedure onto the surface of the baseline data set.

One can use a newly developed 3D surface detection technique to extract the surface of the skeletal site on both the baseline scan and the follow-up scan. Once these surfaces are detected, one can use the Levenberg Marquardt procedure to find the transformation that best matches these two surfaces.

A preferred approach for calculating the cartilage thickness is based on a 3D Euclidian distance transformation (EDT). After thresholding, the voxels on the edge of the cartilage structure can be extracted using a slice by slice 8-neighbor search, resulting in a binary volume with the voxels on the cartilage surface having a value of 1 and all others being 0. To classify these surface points as part of the ICS or OCS, a semi-automatic approach, which requires the user to enter a point that lies outside the 27 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 S cartilage structure and faces the ICS, is useful. From this point, rays are cast in all directions of the volume using a modified Bresenham's line drawing algorithm. If a b) ray hits a voxel with a value of 1, this point is classified as part of the ICS. After a complete sweep of the volume, for initialization of the EDT the ICS voxels are given 5 a value of 0, whereas all other voxels are set to 1.

For computation of the EDT, the following representative algorithm is useful. It can t"decompose the calculation into a series of 3 one-dimensional transformations and can N use the square of the actual distances, which accelerates the process by avoiding the 0 10 determination of square roots.

First, for a binary input picture F {f,jk} (1 5 i L, 1 j M, 1 I k 5 N) a new picture G={gijk} is derived using equations (cc, P, and y denote the voxel dimensions). Here F is a set of all voxels initially and G is a set of all voxels at the later time.

gk min{(a(i x)) 2 ;fj, 0;1 5 x i L} x [Eq. 3] Thus, each point can be assigned the square of the distance to the closest feature point in the same row in i-direction. Second, G can be converted into H={hijk) using equation min{g, y)) 2 ;1 y M} y [Eq. 4] The algorithm can search each column in the j-direction. According to the Pythagorean theorem, the sum of the square distance between a point (ij,k) and a point in the same column, (3(j y)) 2 and the square distance between (i,y,k) and a particular feature point, giyk, equals the square distance between the point (ij,k) and that feature point. The minimum of these sums is the square distance between (ij,k) and the closest feature point in the two-dimensional i-j-plane.

28 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCTIUS99/30265 The third dimension can be added by equation which is the same transformation as described in the equation for the k-direction sj nin{h, (y(k z 5 N} [Eq. After completion of the EDT, the thickness of the cartilage for a given point on the OCS equals the square root of Sabc. The x, y, and z position of each pixel located along the bone cartilage interface can be registered on a 3D map and thickness values are translated into color values. In this fashion, the anatomic location of each pixel at the bone cartilage interface can be displayed simultaneous with the thickness of the cartilage in this location.

Displaying the Degeneration Pattern In an approach the cartilage thickness maps obtained using the algorithm described above display only a visual assessment of cartilage thickness along the articular surface. In another approach, in order to derive a true quantitative assessment of the location, size, and depth of a focal cartilage defect, one can use an iterative approach comparing cartilage thickness of neighboring pixels located along the bone cartilage interface.

For example, assuming an image resolution of 0.5 x 0.5 x 1.0 mm and an average thickness of the articular cartilage in the femoral condyles ranging between 2 to 3 mm, a 25% decrement in cartilage thickness will be the smallest change that can be observed with most current imaging sequences. Therefore, for example, pixels along the bone cartilage interface that demonstrate a decrease exceeding the smallest change observable on a given MRI pulse sequence, in this example 25% or greater, in overlying cartilage thickness when compared to cartilage thickness at the neighboring bone cartilage interface pixels, can be used to define the margins of a focal cartilage defect. Other criteria can be employed to define a cartilage defect based on comparisons of neighboring pixels. For example, a fixed value can be used. If the difference in cartilage thickness between neighboring pixels exceeds the fixed value, 29 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 CN e.g. 1 mm, the pixel where this difference is observed can be used to define the margin of the cartilage defect. This comparison can be performed for each pixel located along the bone cartilage interface for the entire data set. This comparison is preferably performed in three dimensions. Pixels that demonstrate a decrease in cartilage thickness exceeding defined criteria but that are completely surrounded by other pixels fulfilling the same criteria may not be considered to be part of the margin of the cartilage defect, but will typically be considered to lie inside the cartilage defect.

O 10 The invention provides for means for calculating the area covered by the cartilage CK defect Acartilage defect and the mean thickness of the cartilage in the region of the defect Dcarilage defect as well as the mean thickness of a defined area of surrounding normal cartilage. The thickness of the cartilage previously lost in the defect is then estimated as: Dcartilage loss Dnormal cartilage Dcartilage defect [Eq.6].

Since the area A of the cartilage defect is known, the volume of cartilage loss can be computed as: Vcartilage loss Acanilage defect X D cartilage loss [Eq. 7].

In case image noise represents a problem for defining the margin of the cartilage defect using this iterative approach, one can pursue an alternative approach using a 3D surface registration technique to match the anatomic orientation of the cartilage thickness map of the baseline and follow-up scan (described above). The re-registered cartilage thickness map of the follow-up scan can then be subtracted from the baseline scan. This will yield the thickness difference, i.e. cartilage loss, as a function of x, y, and z. This difference can be thresholded with respect to the precision of the 3D surface registration technique by comparing 3D surface registration to registration using a custom-made fiducial frame. This can yield a volume of cartilage defect length vectors. The area of the cartilage defect can be calculated by adding up the SUBSTITUTE SHEET (RULE 26) area between all the voxels on the bone-cartilage interface that survived the thresholded subtraction. The volume of the cartilage defect can be calculated by numerically integrating the volume of thickness difference vectors.

Turning now to Figures 22A and 22B, one can see a 2D MRI (3D SPGR) and 3D cartilage thickness map. In A, the 2D MRI demonstrates a full thickness cartilage defect in the posterior lateral femorl condyle (arrows). Figure 22B shows a 3D cartilage thickness map generated using a 3D Euclidian distance transformation. The thickness of the articular cartilage is color encoded and displayed on a pixel-by-pixel basis along the 3D surface of the articular cartilage. The cartilage defect is black reflecting a thickness of zero (arrows) medial, L: lateral, S: superior, I: inferior).

W \SASKIAPaterU Spec\2004203105d-v13 8 07 doc In Figures 23A-23E, one can see the matching of 3D thickness maps generated from MR images obtained with the knee in neutral position and in external rotation. A.

Sagittal baseline MR image (3D SPGR) with the knee in neutral position. B. Sagittal follow-up MR image of the same volunteer obtained two weeks later with the knee in degree external rotation (note the artificially widened appearance of the femur resulting from the rotation). C. 3D thickness map generated based on baseline MRI in neutral position. D. 3D thickness map generated based on follow-up MRI in external rotation (note segmentation error between condyles in trochlear region). E.

Transformation of D into the object coordinate system of C. Despite extreme differences in joint orientation between baseline and follow-up MRI scans and despite segmentation errors, the thickness distribution on the matched follow-up scan demonstrates great similarity with that seen on the baseline scan in neutral position (in W: SASKIA\Patent Spec\2004203105div-13 8 07 doc N Display of Biochemical Information In addition to providing a 2D or 3D representation of the morphological properties of 5 cartilage, the invention provides for techniques to represent a biochemical components of articular cartilage.

A biochemical component includes, but is not limited to, glycosaminoglycan, water, sodium, or hyaluronic acid. Biochemical data can be generated with other magnetic S 10 resonance based techniques including the use of paramagnetic and other contrast media and sodium rather than proton MR imaging. Thus, one aspect of this invention is a method for providing a biochemically-based map of joint cartilage. The method comprises measuring a detectable biochemical component throughout the cartilage, determining the relative amounts of the biochemical component throughout the cartilage; mapping the amounts of the biochemical component through the cartilage; and determining the areas of cartilage deficit by identifying the areas having an altered amount of the biochemical component present.

W SASKIA\Patent Sec2004203105div-13 8 07 doc Once a map is obtained, it can be used in assessing the condition of a cartilage at an initial time and over a time period. Thus, the biochemical map may be used in the method aspects of the invention in a manner similar to the cartilage thickness map.

For example, one aspect is a method of estimating the loss of cartilage in a joint. The method comprises obtaining a biochemical map of the cartilage at an initial time and analyzing the biochemical content of a region thought to contain degenerated cartilage so mapped at the initial time, obtaining a biochemical map of the cartilage at a later time, and time analyzing the biochemical content of the region thought to contain degenerated cartilage so mapped at the later time, and determining the change in biochemical content of the cartilage between the later and initial times.

W:\SASKIA\Patent Spec2004203105d- 13.8 07 c Preferably, this aspect is directed to a volume of interest in the cartilage, a region of the cartilage that includes a cartilage defect. Such a defect may be the result of a disease of the cartilage osteoarthritis) or the result of degeneration due to overuse or age. This aspect allows a health practitioner to evaluate and treat such defects. The volume of interest may include only the region of cartilage that has the defect, but preferably will also include contiguous parts of the cartilage surrounding the cartilage defect.

As discussed herein before, another aspect is a method for assessing the condition of cartilage in a joint using the biochemical map. The method comprises electronically transferring an electronically-generated biochemically based image of a cartilage of the joint from a transferring device to a receiving device located distant from the transferring device; receiving the transferred image at the distant location; converting the transferred image to a degeneration pattern of the cartilage; and transmitting the degeneration pattern to a site for analysis.

W \SASKIA\Palent Spec2004203105dv- 13 8 07 doc r 0 Another aspect is a method for determining the change of biochemical content in a C region of a cartilage defect of a cartilage in joint of a mammal. Thile method ;comprises deterining the biochemical content (BCN) of the normal cartilage near the cartilage defect; obtaining the biochemical content of the cartilage defect (BCD) of the region; and subtracting BCD firom BCN to give the value of the cartilage change, BCD. The method is useful for situations wherein the region of cartilage defect is limited to the defective cartilage and preferably wherein the region of the cartilage defect includes a portion of the cartilage contiguous to the defect.

Alternatively, for step the normal content of the defect area could be estimated. It may be estimated from measurements of cartilage of other subjects having similar characteristics such as gender, age, body type, height, weight, and other factors. It may be estimated from measurements of a similar 'normal" cartilage from another corresponding joint if the right knee has the defect, measure the normal left knee). It may have been measured at an initial time T, when the cartilage was normal to provide a baseline. Other means of determining the normal content may be available to one of skill in the art. Once BCN is obtained and BCD is obtained the two are subtracted to give the A. By determining the change of content at an initial T 1 and again at a later time T 2 one can determine the change in volume loss over time.

Once the biochemically-based map is provided, morphological maps of articular cartilage obtained with MR imaging are superimposed, merged or fused with the biochemical map or data. Several different techniques can be applied in order to superimpose, merge, or fuse morphological data with biochemical data. For example, 2D or 3D morphological data of articular cartilage can be acquired with the same object coordinates as the biochemical data. Morphological data and biochemical data can then be easily displayed simultaneously using different colors, opacities, and or gray scales. Alternatively, 2D or 3D morphological data or articular cartilage can be acquired with different object coordinates as the biochemical data. In this case, a 3D surface registration can be applied in order to superimpose, merge, or fuse the morphological data and the biochemical data. As an alternative to 3D object coordinates, anatomic landmarks can be used to register the morphological data and subsequently the biochemical data in a 3D object coordinate system. 3D object coordinate systems can then be matched by matching the landmarks obtained from the morphological data with those obtained from the biochemical data.

Thus, another aspect is a method for assessing the condition of a subject's cartilage in a joint, the method comprises obtaining a three dimensional biochemical representation of the cartilage, obtaining a morphological representation of the cartilage, and merging the two representations, and simultaneously displaying the merged representations on a medium. The merged representations are then used to assess the condition of a cartilage, estimate the loss of cartilage in a joint, detennining the volume of cartilage loss in a region of cartilage defect, or estimating the change of a region of cartilage at a particular point in time or over a period of time. One can see that similar steps would be followed as spelled out for the use of a thickness map or biochemical map.

Simultaneous display of morphological data with biochemical data provides a useful tool to assess longitudinal changes in morphology or articular cartilage and biochemical composition of articular cartilage, for example during treatment with chondroprotective and chondroregenerative agents.

Part of the unique aspect of this technology is that it lends itself to assessment of a patient from a distant position after an image is taken of the joint under evaluation.

Thus one aspect is a method for assessing the condition of cartilage in a joint from a distant location. The method comprises electronically transferring an electronically-generated image of a cartilage of the joint from a transferring device to a receiving device located distant from the transferrihg device; receiving the transferred image at the distant location; converting the transferred image to a degeneration pattern of the cartilage; and transmitting the degeneration pattern to a site for analysis.

The degeneration pattern includes a measure of cartilage thickness or regional cartilage volume.

The electronically generated image of the cartilage preferably is an MR image and the degeneration pattern is displayed as a three-dimensional image as a thickness pattern, a biochemical content pattern or a merged thickness biochemical pattern. The electronically generated image is transmitted via Dicom, using the international standards for transmission of such images.

Another aspect is a kit for aiding in assessing the condition of cartilage in a joint of a mammal, which kit comprises a software program, which that when installed and executed on a computer reads a cartilage degeneration pattern presented in a standard graphics format and produces a computer readout showing a cartilage thickness map of the degenerated cartilage.

The software can be installed in a PC, a Silicon Graphics, Inc. (SGI) computer or a Macintosh computer. Preferably, the software calculates the thickness or regional volume of a region of degeneration of the cartilage which does not include the entire volume of the articular cartilage.

THE MOVEMENT PATTERN To acquire a movement pattern of a joint, one obtains an internal image of the bones in a joint, preferably using MRI techniques, and obtains an external image of the bones in motion. The images are correlated, preferably through the use of external marker sets, to give a pattern that shows a static or moving condition. The correlated images are then displayed and the relation between the movement and degeneration patterns is determined.

WO 00/35346 PCT/US99/30265 Obtaining An Internal Image of Joint with Bones To obtain an internal image of a joint with the associated bones, one preferably uses MRI techniques that provide an image of the bones on either side of the joint. Here, it is important to use the imaging technique that gives the best image of the bones and how they interact. Because the internal image of the bones will be combined with the image of the bones obtained by external measurements, it is particularly useful, and therefore preferred, to use external reference markers that can be similarly-positioned to the markers used in obtaining the external measurements. The external markers are placed at any landmarks about the joint of interest. At least three markers are used for each limb being imaged. Preferably the markers will be made of a material that not only will be detected by MRI imaging techniques, but also will be detected by external imaging techniques. The markers will be associated with a means to affix them to the skin and preferably have an adhesive portion for adhering to the skin and a detectable entity that will show up on the MRI image.

The preferred MRI imaging technique useful for obtaining an internal image is a spoiled 3D gradient echo or a 3D DEFT sequence. A further discussion may be found hereinbefore or in the 2 n d Edition of Brown and Semelka's book entitled "MRI Basic Principles and Applications." Once an MR image is obtained the image is manipulated to enhance the image of the bones. Procedures similar to those discussed hereinbefore for cartilage may be used, but modified for application to bone images.

Creating Three-Dimensional (3D) Image of Joint/Bones Three-Dimensional Geometric Model Generation After the 3D image of a joint with bones, the set of segmented two dimensional MR images are transformed to a voxel representation inside AVS Express (Advanced Visual Systems, Inc., Waltham, MA). Every voxel has a value of zero if it is not within an object of interest or a value ranging from one to 4095, depending on the signal intensity as recorded by the 1.5 T MR. An isosurface can then be calculated that corresponds to the boundary elements of the region of interest. A tesselation of 39 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 this isosurface can be calculated, along with the outward pointing normal of each polygon of the tesselation. These polygons can then be written to a file in a standard graphics format (Virtual Reality Modeling Language Version As discussed hereinbefore, the use of reference markers on the skin around the joint and the bones provides an image that can later be matched to the reference markers for the cartilage image and the bone images obtained from external measurements.

Alternatively, a semi-automated, 3D surface-based registration technique that does not require the use of an external frame or fiducial markers can be used. This 3D surface-based registration technique can be used to match the anatomic orientation of a skeletal structure on a baseline and a follow-up CT or MRI scan. We extended a robust and accurate 2D edge detector (Wang-Binford) to 3D. This detector is described hereinbefore.

A registration technique for the femoral condylcs and the tibial plateau is shown in Figure 10. It shows an example where 3D surfaces of the femoral condyles were extracted from two differently oriented TI-weighted spin-echo MRI scans (baseline A and follow-up B, respectively) obtained in the same patient in neutral position and in 40 degree external rotation The 3D surfaces were used to derive a coordinate transformation relating the two scans. Fig. 10C demonstrates the use of the derived transformation to re-register scan B in the object coordinate system of scan A. Such a transformation relating two T1-weighted scans can then be used to register DEFT cartilage-sensitive scans that are acquired in the same respective orientations as the A and B Tl-weighted scans.

We performed the registration using a Sun Sparc 20 workstation with 128MBytes of memory. The surface detection algorithm extracted approximately 12,000 surface patches from each data set. The surface extraction and registration routines took about 1 hour in total.

Since the algorithm for 3D surface registration of the femoral condyles also computes the surface normals for the medial and lateral femoral condyles on a pixel-by-pixel basis, it can form the basis for developing maps of cartilage thickness. Fig. 11 shows SUBSTITUTE SHEET (RULE 26) an example of a 2D map of cartilage thickness derived from the surface nonnals of the lateral femoral condyle. Figure 11 A shows a proton density fast spin-echo MR image that demonstrates a focal cartilage defect in the posterior lateral femoral condyle (black arrows). White arrows indicate endpoints of thickness map.

Figure 11B is a 2D cartilage thickness map that demonstrates abrupt decrease in cartilage thickness in the area of the defect (arrows). The A thickness between neighboring pixels can be used to define the borders of the cartilage defect. Note diffuse cartilage thinning in area enclosed by the astericks Cartilage sensitive images can be used instead of T1-weighted or T2-weighted scans and the surface match can be performed based on the cartilage contour.

Alternatively, anatomic landmarks present on both baseline and follow-up scans can be used to match the data obtained during the baseline and those obtained during the follow-up scan. Another alternative for matching the baseline and the follow-up scan includes the use of external or internal fiducial markers that can been detected with MR imaging. In that case, a transformation is performed that matches the position of the markers on the follow-up scan with the position of the markers on the baseline scan or vice versa.

Obtaining An External Image of Joint/Bones Before merging or superimposing morphological maps of articular cartilage obtained by MR imaging with biomechanical data, one must obtain the biomechanical data.

Such biomechanical data include, but are not limited to, estimations of static loading alignment in standing or weight-bearing position and lying or non-weight-bearing position, as well as during joint motion, the movement of load-bearing pathway on the cartilage in the knee joint during gait. Biomechanical data may be generated using theoretical computations, based on data stored in a database that can be accessed by calling up and screening for certain characteristics. Alternatively, gait analysis may be performed for an individual and data obtained during gait analysis may be merged or fused with morphological MRI data. Morphological data and biomechanical data can then be easily displayed simultaneously using different colors, WO 00/35346 PCT/US99/30265 opacities, and or gray scales. Additionally, the load-bearing pathway, for example around a cartilage defect, can be plotted or superimposed onto morphological maps.

Preferably, reference markers or fiducial markers can be applied to the external surface on the skin overlying the joint. These markers adhere to the skin are typically made of materials that can be detected with MRI and that can be used to register joint motion during biomechanical analysis, e.g. gait analysis. These markers can then be used to correlate the morphological with the biomechanical data.

Simultaneous display of morphological data with biomechanical data provides a useful tool to assess the load pathway applied to articular cartilage and inside and around cartilage defects. Estimation of load pathway applied in and around a cartilage defect can be used to assess a cartilage defect and to guide the choice of therapy, e.g. treatment with chondroprotective or chondroregenerative agents, osteochondral allografting, cartilage transplantation, femoral or tibial osteotomy, or joint replacement surgery.

Recording Static Joint/Bones and Joint/Bones in Movement In obtaining an external image of the bones on either side of a joint, one must record a static image as well as a moving image of the subject joint and bones. For analysis of the knee joint, gait analysis techniques have been shown to be very effective in generating accurate, reproducible data on the six degree of freedom motion of the knee. The motion of the knee joint can be quantified in terms of flexion, rotation and displacement. Fidelity in the dynamic visualizations of subject specific MR generated knee geometry and subsequent contact surface determination call for a high degree of accuracy for the motion capture portion of the studies.

Gait Analysis Activities In performing a gait analysis, a subject is tested standing still, laying down, walking or running on a level surface, flexing a leg in a standing position, ascending and descending stairs, flexing the leg in a seated position, and the like. The level walking measurements can include, but is not limited to, six stride cycles for each side over a 42 SUBSTITUTE SHEET (RULE WO 00/35346 PCT/US99/30265 range of walking speeds. The subject can be instructed to walk at a comfortable speed (normal), slower than normal and faster than normal. Typically, this protocol produces gait measurements over a range of walking speeds. The standing and laying portions of the protocol can be used in the cross registration to the MR data. The instrumentation preferably includes, at least a two camera, video-based optoelectronic system for 3-D motion analysis, a multi-component force plate for measurement of foot-ground reaction force and a computer system for acquisition, processing and analysis of data.

Anatomic Coordinate Systems Currently, the anatomic coordinate systems are defined through bony landmarks which can be identified through palpation. To describe the motion of the underlying bones in terms of the global coordinate system a subset of the markers in a point cluster technique (discussed hereinafter) are referenced to bony landmarks on the femur and tibia. Techniques described previously by Hopenfeld and Benedetti are used to locate these bony landmarks. The anatomic coordinate systems used are similar to that previously described by LaFortune with the exception of the origin of the femoral coordinate system. For the thigh segment, a coordinate system is located in the femoral condyles. The femoral condyles medial(M)-lateral(L) axis (Fig. 12) runs through the trans-epicondylar line (a line drawn between the medial-lateral femoral epicondyles). The midpoint of this axis is the origin. The inferior(l)superior(S) axis runs parallel to the long axis of the femur, passing through the midpoint of the trans-epicondylar line. The anterior(A)-posterior(P) axis is the cross product of the medial-lateral and inferior-superior axes. The final position of the inferior-superior axis is made orthogonal to the anterior-posterior and medial-lateral axis through a cross product operation (Fig. 13). For the shank segment, the tibial coordinate system begins with the medial-lateral axis running through the most medial and lateral edges of the plateau. The inferior-superior axis is perpendicular to the medial-lateral axis passing through the tibial eminence. The anterior-posterior axis is the cross product of the medial-lateral and inferior-superior axes.

43 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 Placement of Markers Prior to Activity In assessing a joint, the lower extremity is idealized as 3 segments with six degree-offreedom joints at the knee and ankle. For the mobile activities described above, at least 3 markers per segment are used. Figure 14 shows 21 passive retro-reflective markers located on the leg: some at bony prominences (greater trochanter, lateral malleolus, lateral epicondyle, lateral tibial pleateu), some clustered on the thigh and shank (Fal-3,11-3, Fpl-3; Tal-3, Tll-13). Additionally, two markers are placed on the foot at the lateral aspect of the calcaneus and base of the fifth metatarsal and one on the pelvis at theiliac crest). During the static activities (standing still, laying down) 7 additional markers are placed: medial malleolus, medial epicondyle, medial tibial plateau, medial and lateral superior patella, medial and lateral inferior patella. The eight markers nearest to the knee joint are filled with Gadolinium, and will be replaced at these same locations prior to the MR images (Fig. 15). The locations are marked with a non-toxic marker-pen.

Reference Database The reference database is a compendium of demographic and motion analysis data for all subjects whose data has been processed by a central processing site 4500 over the past 10 years). This database contains fields describing each of the subject's name, age, height, weight, injury types, orthopedic medical history, other anatomic measurements (thigh length, shank length, shoe size, etc.). The database also contains the results of any and all gait analysis run on these patients. This includes, for all activities tested (walk, run, jog, etc.), a number of peak valves (peak knee flexing, peak hip adduction movement; toe-out, angle, etc). along with the motion trajectories of the limb segments while the subjects are performing different activities.

In order to obtain a typical motion profile, the sex, age, height, weight, limb length, and type of activity desired is entered as an avery into the database. The database searches for a set of subjects most closely watching the input avery. From this set of data, a typical motion pattern is distilled and a data set is output. This data set includes, over a time interval, the motion characteristics: hip knee ankle flexion extension angles, knee hip ankle adduction abduction angles, movement, stride 44 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 length, cadence, etc. This data can then be used to drive an animation of the motion of the desired joint.

Process Image of Joint/Bones Calculation of Limb Segment Parameters Each limb segment (thigh, shank and foot) can idealized as a rigid body with a local coordinate system defined to coincide with a set of anatomical axes (the assumption of rigidity is dropped in calculating the location of the femur and tibia). The intersegmental moments and forces are calculated from the estimated position of the bones, the ground reaction force measurements, and the limb segment mass/inertia properties. The moment at the knee can be resolved into a coordinate system fixed in a tibial reference system with axes defining flexion-extension, abduction-adduction, and internal-external rotation.

This approach provides results in a range of patients in a highly reproducible manner.

Typically the magnitudes of the moments are dependent on walking speed. To control for the influence of walking speed, the walking speed closest to 1 meter/second is used. This speed is within the normal range for the type of patients for which the invention is particularly useful. In addition to the gait trial collected at 1 meter/second, self-selected speeds can also be evaluated to give a good correlation between gait-quantitative estimates of joint load lines and other measures when using self-selected speeds. In order to test patients under their typical daily conditions, medications should not be modified prior to gait analyses.

Point Cluster Technique The Point Cluster Technique (PCT) movement analysis protocol is an extensible and accurate approach to bone motion estimation. Basically, a number of retro-reflective markers retro-reflective material from 3M, Corp.) are attached to each limb segment under observation. Multiple video cameras acquire data with the subject standing still and during activities of interest. An over-abundance of markers on each SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCTIUS99/30265 limb segment is used to define a cluster coordinate system, which is tied to an anatomically relevant coordinate system calculated with the subject at rest.

The standard PCT transformations are described below. In short, each marker is assigned a unit mass and the inertia tensor, center of mass, principal axes and 'i principal moments of inertia are calculated. By treating the center of mass and principal axes as a transformation, local coordinates are calculated. Another set of coordinate systems are established; limb segment specific anatomic landmarks are I identified through palpation and a clinically relevant coordinate system defined. For the femur and tibia, these anatomic coordinate systems are shown in Fig. 12. The C transformation from the reference cluster coordinate system to the anatomic coordinate system is determined with the subject at rest by vector operations. During an activity, the transformation from the global coordinate system to the cluster coordinate system is calculated at each time step. To place the anatomic coordinate in the global system during the activity, the reference coordinate system to anatomic system transformation is applied, followed by the inverse global coordinate system to cluster coordinate system transformation for each time step.

In the Point Cluster Technique (PCT) a cluster of N markers can be placed on a limb segment of the subject. The location vector of each marker in the laboratory coordinate system is denoted as G(i,t) for marker i, (i at time t, to t tr.

A unit weight factor is assigned to each marker for the purpose of calculating the center of mass, inertia tensor, principal axes and principal moments of inertia of the cluster of markers. The cluster center of mass and principal axes form an orthogonal coordinate system described as the cluster system. The local coordinates of each of the markers relative to this coordinate system are calculated. Then G(i,t) C(t) E(t) L(i,t) Te 9 L(i,t) i I N where G(t) is a matrix of all marker coordinate vectors, C(t) is the center of mass of E(t) is the matrix of eigenvectors of the inertia tensor of and L(i,t) are the local coordinates of marker i.

46 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 These markers are observed by opto-electronic means while the subject performs activities and while standing completely still in a reference position. With the subject in this same reference position, a subset of the markers is observed relative to the underlying bones by other techniques, which might include x-rays, CT scan, or palpation.

The measured marker locations are defined with respect to the unobservable location and orientation of the bone by G(i,t) P(t) O(t) R(i,t) Th R(i,t) i= I N where P(t) is the location and O(t) is the orientation of a coordinate system embedded in the bone and also unobservable, are the trajectories of the markers relative to the underlying rigid body coordinate system at time t. The bone and cluster systems are each orthogonal systems, related by the rigid body transformation Tbc(t): L(i,t) Tbc R(i,t) substituting and eliminating R(i,t) yields Tb Tc Tcb (t) To maintain physical consistency, Tcb(t)= Tbc(t)-' must be the inertia tensor eigendecomposition transformation of Once R(i,t) are specified, Tcb(t) and subsequently Tb(t) are calculable.

Point Cluster to Anatomic Coordinate System Transformation From these equations one can also relate the global coordinate system with respect to a limb segment system. As an example of how these systems are used to describe joint motion, one can consider the tibio-femoral joint. The motion that is of interest is how the femoral condyles move with respect to the tibial plateau. This is done by first defining a set of coordinate axes in the femoral condyles and the tibial plateau.

47 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 A coordinate system is located in both the femoral condyles and the tibial plateau.

The femoral condyles medial-lateral (ML) axis runs through the trans-epicondylar line (TEL), a line drawn between the ML femoral epincondyles. The midpoint of this axis is the origin. The inferior-superior (IS) runs parallel to the long axis of the femur, passing through the midpoint of the TEL. The anterior-posterior (AP) is the cross product of the ML and IS axes. The tibial coordinate system begins with the ML axis running through the most medial and lateral edges of the plateau. The IS axis is perpendicular to the ML axis passing through the tibial eminence. The AP axis is the cross product of the ML and IS axes. These are known as the anatomic coordinate system (A(t)thigh, A(t)shank).

Relating the cluster system to the anatomic coordinate system is done by use of another transformation matrix. This is done by relating the thigh cluster to a cluster of markers, a sub cluster, that is related to the femoral condyles and femur (cluster to anatomic transformation).

R(t)thigh U(t)thigh A(t)thigh The tibia has a similar transformation matrix.

R(t)shank U(t)shank A(t)shank Therefore, from a cluster of markers in the global system, motion of the femur with respect to the tibia can be determined by: TS(t) A(t)thigh G(t)thgh R(t)shank A(t)shank Here TS(t) is the motion of the thigh with respect to the shank.

Angles are calculated by a projection angle system, an axis from the femoral anatomic system and one from the tibia are projected onto a plane in the tibial coordinate 48 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 system. For example, flexion/extension can be determined by projecting the IS axis of the femur and tibia onto the sagittal plane (AP-IS plane) of the tibia.

Validation of the Point Cluster Technique The point cluster technique was evaluated as a method for measuring in vivo limb segment movement from skin placed marker clusters. An Ilizarov device is an external fixture where 5 mm diameter pins are placed directly into the bone on either side of a bony defect. The rigid external struts affixed to these pins form a rigid system fixed in the underlying bone. Two subjects were tested with Ilizarov fixation devices. One subject had the Ilizarov device placed on the femur and second subject had the device placed on the tibia. Each subject was instrumented with point clusters placed on the thigh and shank segment. In addition, markers were placed on the Ilizarov device to establish a system fixed in the underlying bone.

The relative angular movement and translational displacement between the system affixed in the bone and the point cluster coordinate system were calculated while ascending a 20-cm step (Step Test). Angular changes between the three orthogonal axes fixed in the bone versus three axes in the point cluster were calculated. The average difference over the trials for three axes were 0.95 1.26, 2.33 +1.63, and 0.58 0.58 degrees. Similarly, the average error for the distance between coordinate systems was 0.28 0.14 cm. The second subject with the Ilizarov device placed on the femur could not perform the Step-Test, but was able to perform a weight-bearing flexion test where his knee flexed to approximately 200 from a standing position. The average change between the coordinate origin was 0.28 0.14 cm. The changes in axis orientation were 1.92 0.42, 1.11 .69 and 1.24 0.16 degrees.

The simultaneously acquired motion for a coordinate system embedded in bone (Ilizarov system) and a set of skin-based markers was compared. At every time instant the location and orientation of the Ilizarov system, the rigid body model skin marker system, and the interval deformation technique skin marker system were determined. The change in the transformation from the Ilizarov system to one of the skin marker systems over time is a measure of the deformation unaccounted for in the skin marker system.

49 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 The interval deformation technique produced a substantial improvement in the estimate of the location and orientation of the underlying bone. For perfectly modeled motion there would be no relative motion between the Ilizarov system and the skin marker system over the time interval. The change in the transformation from the llizarov system to the skin marker systems are shown in Figures 14 and 15, for location and orientation respectively, for both a rigid body model and the interval deformation technique. For this single data set, the location error was reduced from 7.1 cm to 2.3 cm and the orientation error from 107 degrees to 24 degrees, with the error summed over the entire time interval. The subject performed a 10 cm step-up; the marker deformation was modeled as a single Gaussian function.

Deformation Correction There are a number of algorithmic alternatives available to minimize the effects of skin motion, soft tissue deformation, or muscle activation that deform the externally applied markers relative to the underlying bone. The Point Cluster Technique decreases the effects of marker movement relative to the underlying bone through averaging. If more correction is required, one of a number of deformation correction techniques may be added. In order of increasing computational complexity and deformation correction ability, these are rigid body linear least square error correction, global optimization correction, anatomic artifact correlation correction and interval deformation correction.

An overview of the Interval Deformation Correction Technique is given below. In short, the technique provides a maximum likelihood estimate of the bone pose, assuming that each marker on a limb segment deforms relative to the underlying bone in some functional form. The technique parameterizes these functional forms and then performs a multi-objective non-linear optimization with constraints to calculate these parameters. This is an extremely computationally intensive technique, with the current instantiation of the algorithm requiring 6- 8 hours per limb segment of running time on 266 MHz Pentium 2 computer.

SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 Interval Deformation Technique Since Tc can be calculated directly from the global coordinates of the markers, the remainder of this development only examines the determination of R(i,t) and subsequently Tcb(t). For this reduced problem, the input data is the local coordinates in the cluster system L(i,t) for all i, To t tr. It can be assumed that each marker has some parameterized trajectory, d(aij, relative to the underlying bone at each time step, with independent and identically distributed noises v(ij,t) Rj(i,t)= d(aij, t) v(ij,t) j 1 3 i 1 N or, equivalently F(ai, t) v(i,t) i 1 N where aij is a vector of parameters for marker i, ordinate j; ai is a vector of parameters combining all of the parameters for all of the ordinates of marker i. Then the estimate of the data, can be given by M(i,t) Tbe R(i,t) Without further restrictions the problem is indeterminate, as the locations of the markers in the bone system R(i,t) are never observable with the opto-clectronic system. The indeterminate problem can be converted to a chi-squared estimate problem through a series of steps. An observation of the truly unobservables at the time boundaries is inferred; that is, it is assumed that Tcb (t to) and Tcb (t tf) are observed. The value of Tcb can be selected depending on the activity being studied.

For example, consider the step up activity, where the subject starts and stops in the reference position. For this activity the body is not deforming outside the estimation interval; that is, the markers are not moving with respect to the bone: Tcb (t t o Tcb (t to) and Tcb (t tr) Tcb (tr).

51 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 It can now be assumed that the noise functions v(i, j, t) are normal distributions with individual standard deviations o(i, j, the probability P(ij,t) of the data for ordinate j, marker i, time t being a realization of the stochastic process is given by: 1 (L(ij,t) -M(i,j,t) P(i,j,t) oc exp 2 oc(i,j,t) Provided the noise functions v(i, j, t) are independent of each other, the probability of the entire data set being a realization is a product of each of the individual probabilities: Ncjt 3 exp (1 (L(i,j,t)-M(ijt)) 2 P(i, j, t) oc I Inn n exp Ji r= 2 o(i, t) Maximizing this probability can be equivalent to minimizing the negative of its logarithm, yielding the familiar chi-square criteria. As an intermediate step the following error matrices can be defined: X(a, t) (L(it)M i 1 N j j,t) 3 X(a) X(a,t) and seek a which in some sense minimizes a matrix whose elements represent the error over the entire time interval for each ordinate of each marker. If the normal noise distribution assumption is true, then this minimization results in the maximum likelihood estimate of the parameterization, and by inference maximum likelihood estimate of the transformation from the bone system to the cluster system. If the normal noise assumption is not true, the chi-squared estimate is still appropriate for parameter estimation; the results cannot be interpreted as a maximum likelihood estimate, but, for example, confidence regions on the estimate or the formal covariance matrix of the fit can be determined.

52 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT7US99/30265 Obtaining the parameter set a is a computationally complex operation. The approach taken was to define a scalar to represent this entire error matrix, f(a)=Z X(a)i j 2=I I-i and seek a that minimizes f(a).

The limits on marker motion previously discussed can now be converted into deformation constraints, which allow the formulation of the problem as a general nonlinear programming problem. The constraints arise from two sources; human limb segments do not deform outside a small range, and the locations of the markers are chosen with specific properties in mind. For computational purposes, the deformation constraints are selected to be: 1. The axes of the cluster system moves by less than 15 degrees relative to the bone system.

2. The center of mass of the cluster system moves by less than 3 cm relative to the bone system.

3. The markers move by less than 4 cm relative to the bone system.

4. Each of the principal moments of inertia of the cluster system change by less than 25 percent from the reference values.

The Point Cluster Technique marker set was designed to ensure that the cluster of points is non-coplanar and possess no axes of rotational symmetry. These properties ensure a local coordinate system that is well defined and unambiguous over the entire time interval. The constraints are then: The ratio of the smallest principal moment of inertia of the cluster system to the largest is more than 5 percent; the magnitude of the smallest principal moment of inertia of the cluster system is greater than some small positive value.

6. The principal moments of each axis are different from each other by at least percent.

53 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 The general problem can then be formulated: Minimize f(a) a RD Subject to: gi(a) 0 i= 1 me gi(a) 0 i m, 1 m a a au where D is the total number of parameters; me, the number of equality constraints, is 0; and m, the total number of constraints, is The approach taken to verify the operation of the algorithm implementation began with generating a set of 50 synthetic data sets with known characteristics. The program was then applied to all of the data sets. The program results were then compared to the known, generated deformation. Error results were calculated for both the interval deformation technique descried herein and for the standard rigid body model formulation.

The 50 trial data sets were processed through the algorithm. The results over all of the trial sets are summarized in Table I, where the center of mass and direction cosine error of the interval deformation technique and the rigid body model are compared.

After processing by the interval deformation algorithm the center of mass error has been reduced to 29% and the direction cosine error has been reduced to 19% of the rigid body model error. In a t-test for paired samples, both of these decreases were significant at p 0.001.

Validation of the Interval Deformation Correction Technique A subject fitted with an Ilizarov external fixation was observed with the optoelectronic system. The Point Cluster Marker set was affixed to the subject's shank (6 markers), along with a set of four markers rigidly attached to the Ilizarov device, which is rigidly connected to the tibia with bone pins. These four markers 54 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 Sdefine a true bone embedded coordinate system. Data were acquired by GaitLink t) software (Computerized Functional Testing Corporation) controlling four Qualisys cameras operating at a video frequency of 120 Hz. Three dimensional coordinates were calculated using the modified direct linear transform.

The subject was a 46 year old male (height 1.75 m, weight 84.1 kg) fitted with a tibial Ilizarov external fixation device. The device was rigidly attached to the tibia with nine bone pins, located in three sets (top, middle, and bottom) of three (medial, c1 anterior, and lateral). The clinical purpose of the device was tibial lengthening; the S 10 test on the subject was performed two days prior to final removal of the device. The Csubject exhibited a limited range of motion and was tested performing a 10 cm stepup onto a platform.

The simultaneously acquired motion for a coordinate system embedded in bone (Ilizarov system) and a set of skin-based markers was compared. At every time instant the location and orientation of the Ilizarov system, the rigid body model skin marker system, and the interval deformation technique skin marker system was determined. The change in the transformation from the Ilizarov system to one of the skin marker systems over time is a measure of the deformation unaccounted for in the skin marker system.

The interval deformation technique produced a substantial improvement in the estimate of the location and orientation of the underlying bone. For perfectly modeled motion there would be no relative motion between the Ilizarov system and the skin marker system over the time interval. The change in the transformation from the Ilizarov system to the skin marker systems are shown in Figs. 14 and 15 for location and orientation respectively, for both a rigid body model and the interval deformation technique. For this single data set, the location error was reduced from 7.1 cm to 2.3 cm and the orientation error from 107 degrees to 24 degrees, with the error summed over the entire time interval. The subject performed a 10 cm step-up; the marker deformation was modeled as a single Gaussian function.

SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 CORRELATING RESULTS FROM GAIT ANALYSIS AND GEOMETRICAL REPRESENTATIONS OF THE BONE In correlating the load pattern obtained from a gait analysis using, e.g. the PCT, with the geometrical representation of the bone from the segmented MRI data, one is guided by the general process as described below. The process allows for dynamic visualization animations) of high-resolution geometrical representations derived from MRI scans (or other imaging techniques). The motion of the subject specific anatomic elements is generally driven by data acquired from the motion (gait) lab.

Fidelity of these animations requires calculation and application of a sequence of rigid body transformations, some of which are directly calculable and some of which are the result of optimizations (the correction for skin marker deformation from rigidity does not use the rigid body assumption, but generates a correction that is applied as a rigid body transform).

The process comprises: a) acquiring data from MRI (or other imaging techniques), and PCT gait protocols; b) directly calculating a set of transformations from the data; c) calculating a set of transformations from optimizations, as needed; d) generating a 3D geometric representation of the anatomic element from the MR data; and e) applying the transformations of and to the 3D geometric representation.

Each of these steps are described in detail below.

56 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 Acquiring the Data from MRI (or other imaging techniques) and PCT Gait Protocols In the Point Cluster Technique (PCT) protocol, a patient can have a number of retroreflective markers attached to each limb segment under observation. Multiple video cameras acquire data with the subject standing still and during activities of interest.

In addition, in order to correspond activities in the gait lab with the MRI scans, another reference data set (subject standing still, prescribed posture) can be acquired using 8 additional markers clustered about the knee. These markers are filled with gadalinium and covered with a retro-reflective material to allow for correlation between the MRI image and the video data.

Directly Calculating a Set of Transformations from the Data The transformations are described in detail in Andriacchi et al., J. Biomech. Eng., 1998. In short, each marker is assigned a unit mass and the inertia tensor, center of mass, principal axes and principal moments of inertia can be calculated. By treating the center of mass and principal axes as a transformation, local coordinates are calculated. Another set of coordinate systems can also be required for this technique; limb segment specific anatomic landmarks are identified through palpation and a clinically relevant coordinate system can be defined. The required transformations are summarized in Table below.

Calculating a Set of Transformations from Optimizations There arc three required transformations: Optimization 1. One calculates the linear least square error rigid body transformation from the MRI common local coordinate system to the VID common local coordinate system.

Optimization 2. For each limb segment, one calculates the linear least square rigid body transformation from the MRI limb segment anatomic coordinate system to the 57 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 video limb segment anatomic coordinate system (obtained from the gait analysis), using a subset of common markers appropriate for each segment.

Optimization 3. One calculates a correction for the deviation of the limb segment from rigidity during each time step of the activity, using the PCT with either the mass redistribution (Andriacchi et al., J. Biomech Eng., 1998) or interval deformation algorithms (Alexander et al., Proceedings of the 3 rd Annual Gait and Clinical Movement Analysis Meeting, San Diego, CA, 1998).

Generating a 3D Geometric Representation of the Anatomic Element from the MR data The MR slices are segmented for the multiple anatomic and fiducial elements. The slices are combined to a voxel representation. An isosurface can be calculated from the boundary voxel elements. A tessellation of the isosurface can be calculated, along with the outward pointing normal for each surface element. This data can then be stored in a standard 3D graphic format, the Virtual Reality Modeling Language

(VRML).

Appling the Transformation Sequence to the Geometric Representation The transformation sequence is provided below in Table 1. This transformation sequence can be applied to each of the anatomic elements over each time step of the activity, starting with sequence 6.

58 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 TABLE 1 SEQ FROM SYSTEM TO SYSTEM XFORM 1 MR Global MR Local ED1 2 MR Local Common Local OPTI 3 Common Local MR Anatomic ANA2 4 MR Anatomic VID Anatomic OPT2 VID Anatomic VID Ref ANA3 6 VID Ref VID Deformed(t) ED3 7 VID Deformed(t) VID Bone(t) OPT3 8 VID Bone(t) VID Global(t) ED4 Correlating Marker Sets As pointed out at numerous places in the specification, the use of external reference markers that are detectable by both MRI and optical techniques is an important and useful tool in the method of this invention. The use of the reference markers forms the basis for an aspect of this invention that is a method for correlating cartilage image data, bone image data, and/or opto-electrical image data for the assessment of the condition of a joint of a human. This method comprises, obtaining the cartilage image data of the joint with a set of skin reference markers placed externally near the joint, obtaining the bone image data of the joint with a set of skin reference markers placed externally near the joint, obtaining the external bone image data opto-electrical image data of the joint with a set of skin reference markers placed externally near the joint. Using the skin reference markers, one can then correlate the cartilage image, 59 SUBSTITUTE SHEET (RULE 26) 7- WO 00/35346 PCT/US9930265 bone image and opto-electrical image with each other, due to the fact that each skin reference marker is detectable in the cartilage, bone and opto-electrical data. The cartilage image data and the bone image data can be obtained by magnetic resonance imaging, ultrasound or X-ray. MRI is preferred. In the case of X-ray, further manipulations must be performed in which multiple X-ray images are assimilated by a computer into a 2 dimensional cross-sectional image called a Computed Tomography (CT) Scan. The opto-electrical image data can be obtained by any means, for example, a video camera or a movie camera. Multiple skin reference markers can be placed on one or more limbs of the patient prior to imaging. The skin reference markers are described hereinbefore.

By a sequence of calculations a set of transformations that will take the subject specific geometric representation of anatomic elements determined from the MR image set to the optical reference coordinate system. From the optical reference coordinate system, the standard Point Cluster Technique transformation sequence is applied to generate dynamic visualizations of these anatomic elements during activities previously recorded in the motion lab. Fidelity of these dynamic visualizations (and subsequent contact surface determination) requires the calculation and application of a sequence of rigid body transformations. Some of these are directly calculable and some are the result of optimizations (the correction for skin marker deformation from rigidity does not use the rigid body assumption, but generates a correction that is applied as a rigid body transform).

The first required transformation is from the MR global coordinate system to the MR center of mass principal axis coordinate system. This is done by calculating the center of mass of each of the individual markers, resulting in a set of eight three dimensional points. Each of these points is assigned a unit mass, and the center of mass, inertia tensor, and principal axes are calculated. The same procedure is performed on these markers as determined by the optical system, providing a transformation from the optical global system to a center of mass principal axis system.

SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 C If the relative orientation of the tibia and femur as determined by the MR system and tb the optical system are identical, it is only necessary to apply the optical reference system to the anatomic system transformation of the MR local data. If this is not the case, an optimization calculation is performed to determine the rotation and translation of, for example, the femur with respect to the tibia. One then can calculate C the linear least square rigid body transformation from the MR limb segment anatomic coordinate system to the video limb segment anatomic coordinate system prior to applying the Point Cluster Transformations.

S 10 For visualization or contact surface determination, one can examine the relative motion of one segment to the other, for example the motion of the femur relative to a fixed tibial frame. This can be accomplished by applying the global to tibial anatomic system transform to all of the elements. An example of this type of visualization is given in Figure 18. The Figure shows what can be referred to as functional joint imaging. Figure 18A is a photograph demonstrating the position of the external markers positioned around the knee joint. The markers are filled with dilute Gdsolution. B is Sagittal 3D SPGR image through the medial femorotibial compartment.

Two of the external markers are seen anteriorly as rounded structures with high signal intensity. C is 3D reconstruction of femoral and tibial bones (light grey), external markers (dark grey), femoral cartilage (red), and tibial cartilage (blue) based on the original SPGR MR images. D-I show a functional joint imaging sequence at selected phases of leg extension from a seated position, D-F, anterior projection. The vectors represent the relative location and orientation of the femur with respect to the tibia.

G-I is a lateral projection. These dynamic visualizations can be used to demonstrate tibiofemoral contact areas during various phases if gait or other physical activities.

Superimposition of cartilage thickness map onto subject specific anatomic model and determination of distance of cartilage defect from load bearing line Superimposing the cartilage thickness maps onto the subject specific geometric models can follow the same approach taken to bring the MR generated geometries into the optical reference system. Since the thickness maps and the geometric models are initially in the same coordinate system; one possible approach is to perform a simple surface mapping of the thickness map onto the geometric model. Another 61 SUBSTITUTE SHEET (RULE 26) SWO 00/35346 PCTIUS99/30265 Salternative approach is to convert the thickness map directly into a geometric representation (Fig. 19).

Once the thickness map is embedded in the femoral geometry, one can define a scalar metric that characterizes the location of any cartilage lesions relative to the point of contact line. One approach is a simple 3D distance along the surface from the center of the cartilage lesion to the point of closest approach of the contact line. Another metric that could be useful would be to multiply the area of the lesion by the C"l adduction moment at that time instant, then divide by the distance from lesion center to point of closest approach. This could result in a metric that increases with lesion Cl area, adduction moment, and closeness of approach.

46 Display Correlated Images Determination ofAnatomic and Natural Reference Lines There are two alternative approaches one can consider for determining a reference line on the cartilage surfaces. The first approach is based on anatomic planes; the second is a natural approach building on the three dimensional cartilage thickness map.

The location of the pathway of loading relative to the femoral and tibial anatomy and geometry can be assessed by defining sagittal planes bisecting the medial femoral condyle, the lateral femoral condyle, the medial tibial plateau, and the lateral tibial plateau. For the medial femoral condyle, the operator can manually delete surface points located along the trochlea. Then, a sagittal plane parallel to the sagittal midfemoral plane is defined through the most medial aspect of the medial femoral condyle followed by a sagittal plane parallel to the sagittal midfemoral plane through the most lateral aspect of the medial femoral condyle. The sagittal plane that is located halfway between these two planes is defined as the "midcondylar sagittal plane". The intersection between the midcondylar sagittal plane and the external cartilage surface yields the "anatomic midcondylar cartilage line". The location of the pathway of loading can be assessed relative to the anatomic midcondylar cartilage line 62 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 of the medial femoral condyle. The identical procedure can be repeated for the lateral femoral condyle.

The following method can be used for the medial tibial plateau: A plane parallel to the sagittal tibial plateau plane is defined through the most medial point of the medial tibial plateau. A parallel plane located halfway between this plane and the sagittal tibial plateau plane yields the "midsagittal plane of the medial tibial plateau." The intersection of the midsagittal plane of the medial tibial plateau and the external cartilage surface yields the "anatomic midtibial plateau cartilage line" of the medial tibial plateau. The identical procedure can be repeated for the lateral tibial plateau.

In the second approach, one can calculate a "natural" line of curvature for each femoral cartilage component (Fig. 20). Intuitively, if one could roll the femoral condyles along a hard, flat surface, the line of contact with the flat surface would be the natural line of curvature. One can compare the actual tibiofemoral contact line to this reference line. Since one can't physically remove the femur and roll it around, one can apply some geometric calculations to estimate this reference line. One can begin with the trans-epicondylar reference line previously described. One can then generate a plane coincident with this line oriented in an arbitrary initial position. The intersection of this plane and the external surface of the cartilage will produce a curve.

One can then take the point furthest from the trans-epicondylar reference line as the natural contact point for this plane location. The next step is to rotate the plane by some increment, for example by one degree, and repeat the procedure. The operator can identify the rotation angles where the plane is intersecting the distinct medial lateral compartments of the cartilage, and two points can be chosen, one from the medial femoral condyle and one from the lateral femoral condyle. If cartilage defects are present, in which case a compartment will not intersect in a curve but in a set of points, one can fit a spline through the points, then take the peak point of the spline as the contact point.

This can be repeated for the entire extent of the cartilage, resulting in a set of points that branch at the intercondylar notch. One can treat these points as two lines, and fit 63 SUBSTITUTE SHEET (RULE 26) WO 00/35346 PCT/US99/30265 them with two splines. These can be the "natural" lines of curvature for each compartment.

Load Bearing Line Determination The calculations in this section begin with the relative motion of the subject specific femoral anatomy with respect to the subject specific tibial anatomy, and end with a line describing the point of closest approach between the femur and tibia during some activity of daily living. A number of approaches to this problem have been described in the literature; Crosset, Dennis, Stiehl, and Johnson have all described techniques which might be applicable. One can implement a proximity detection and approach algorithm (PDAA) as it was specifically designed to work with the Point Cluster Technique (albeit with prosthetic knee joint components).

Physically, the tibial and femoral cartilage components deform under load, leading in general to a contact patch between opposing surfaces. As the geometric models are rigid, they will not deform under this load, but will instead intersect in a nonrealizable manner. The PDAA has been designed to incrementally displace and rotate one of the surfaces until a realizable contact is achieved. It is understood that this is not a true point contact line, but rather a reproducible representation of contact location (Fig. 21).

The MR generated subject specific geometries can be used to detect rigid body contact proximity when the subject is in full extension. The femoral component can then be incrementally displaced until simultaneous medial and lateral condyle contact occur. This is a first order approximation to the location of the contact point; slip velocity calculations can then be used to determine the final estimate of the contact point. The next time step in the activity can now be examined, using the previous time step solution as a starting point for the calculation. The full extension time step can be chosen to match with the static reference posture; should it be necessary, one can add in other reference postures.

Once the contact points have been determined for all time steps of the activity, one can map the locations of these points onto the femoral cartilage. A coordinate system 64 SUBSTITUTE SHEET (RULE 26) can be defined on the surface of the femoral cartilage, choosing as a reference line the point of contact the femoral component would have had were it rolled along a flat plane. This allows one to determine a contact line relative to the subject specific anatomy.

Provide Therapy A 2D or 3D surface registration technique can be used as an aid to providing therapy to match the anatomic orientation of the cartilage thickness map of a baseline and follow-up scan of a patient. The re-registered cartilage thickness map of the followup scan can then be subtracted from the baseline scan. This will yield the thickness difference, i.e. cartilage loss, as a function of x, y, and z. This can also be expressed as percentage difference. The thickness difference can be thresholded with respect to the precision of the 2D or 3D surface registration technique. This will yield a volume of cartilage defect length vectors. The area of the cartilage defect can be calculated by adding up the area between all the voxels on the bone-cartilage interface that survived the thresholded subtraction. The volume of the cartilage defect can be calculated by numerically integrating the volume of thickness difference vectors.

Techniques are provided to assess biomechanical loading conditions of articular cartilage in vivo using magnetic resonance imaging and to use the assessment as an aid in providing therapy to a patient. In one embodiment, biomechanical loading conditions can be assessed in normal articular cartilage in various anatomic region. In the knee joint, these anatomic regions include the posterior, central, and anterior medial femoral condyle, the posterior, central, and anterior medial tibial plateau, the posterior, central, and anterior lateral femoral condyle, the posterior, central, and anterior lateral tibial plateau, the medial and lateral aspect of the trochlea, and the medial and lateral facet and the median ridge of the patella. Since biomechanical loading conditions are assessed in vivo based on the anatomic features of each individual patients, a risk profile can be established for each individual based on the biomechanical stresses applied to cartilage. In this fashion, patients who are at risk for developing early cartilage loss and osteoarthritis can be identified. For example, patients with a valgus or varus deformity of the knee joint will demonstrate higher biomechanical stresses applied to the articular cartilage in the medial femorotibial or lateral femorotibial or patellofemoral compartments than patients with normal joint anatomy. Similarly, patients with disturbances of joint congruity will demonstrate higher biomechanical stress applied to certain regions of the articular cartilage. Such disturbances of joint congruity are often difficult to detect using standard clinical and imaging assessment. The amount of stress applied to the articular cartilage can be used to determine the patient's individual prognosis for developing cartilage loss and osteoarthritis.

Estimation of biomechanical forces applied to normal cartilage can be used to determine a patient's risk for developing cartilage loss and osteoarthritis. Estimation of forces applied in and around a cartilage defect can be used to determine the prognosis of a cartilage defect and to guide the choice of therapy, e.g. treatment with chondroprotective or chondroregenerative agents, osteochondral allografting, cartilage transplantation, femoral or tibial osteotomy, or joint replacement surgery.

All publications and patent applications mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication or patent application was specifically and individually indicated to be incorporated by reference.

The invention now being fully described, it will be apparent to one of ordinary skill in the art that many changes and modifications can be made thereto without departing from the spirit or scope of the appended claims.

Claims (24)

1. A method for providing a biochemically-based map of joint cartilage of a mammal, wherein the joint includes cartilage and associated bones on either side of the joint, which method includes: measuring a detectable biochemical component throughout the cartilage; determining the relative amounts of the biochemical component throughout the cartilage; mapping the amounts of the biochemical component in three dimensions through the cartilage; and determining the areas of abnormal joint cartilage by identifying the areas having altered amounts of the biochemical component present.

2. The method of claim 1 wherein the biochemical component is glycosaminoglycan, sodium, water or, hyaluronic acid.

3. The method of claim 1 or 2 wherein the joint is a knee joint.

4. The method of claims 1 to 3 wherein the mammal is a human.

The method of any one of claims 1 to 4 wherein steps through are carried out at an initial time (TI) and are carried out again at a later time (T 2

6. The method of claim 5 wherein the areas of abnormal cartilage at Ti and the areas of abnormal cartilage at T 2 are analyzed for a degree of change.

7. The method of any one of claims 1 to 6 wherein measuring of the biochemical component is done using a magnetic resonance imaging (MRI) technique.

8. The method of claim 7 wherein the MRI technique provides a biochemical description of the cartilage. W SASKIAPaten Spec\2004203105d,. 13 807 doc

9. The method of claim 7 wherein the MRI technique provides a volumetric description of the cartilage.

The method of claim 7 wherein the MRI technique includes placing external markers on the skin overlaying the bone on either side of the joint.

11. The method of claim 10 wherein the MRI technique results in a three dimensional image of the cartilage.

12. The method of claim 10 wherein the MRI technique first obtains a series of two-dimensional views of the joint, which are then mathematically integrated to give a three-dimensional image.

13. The method of any one of claims 7 to 12 wherein the MRI technique employs a gradient echo, spin echo, fast-spin echo, driven equilibrium fournier transform, or spoiled gradient echo technique.

14. A biochemical map of a cartilage in a joint of a mammal, wherein the joint comprises cartilage and associated bones on either side of the joint, which map comprises a three-dimensional representation of the distribution of the amount of the biochemical component throughout the cartilage.

The map of claim 14 wherein the biochemical component is glycosaminoglycan, water, sodium, or hyaluronic acid.

16. The map of claim 14 or 15 in combination with a movement pattern of the joint.

17. The map of any one of claims 14 to 16 wherein the joint is a knee joint.

18. The map of any one of claims 14 to 17 wherein the mammal is a human. W \SASKIA\Paile Spec2004203105(d-13 8 07 doc I*

19. The map of any one of claims 14 to 18 wherein the biochemical map of a cartilage in a joint is obtained from a magnetic resonance imaging (MRI) technique.

20. The map of claim 19 wherein the MRI technique includes placing external markers on the skin overlaying the bone on either side of the joint.

21. The map of claim 20 wherein the MRI technique results in a three-dimensional image of the cartilage.

22. The map of claim 21 wherein the MRI technique first obtains a series of two- dimensional views of the joint, which are then mathematically integrated to give a three-dimensional image.

23. The map of any one of claim 10 to 22 wherein the MRI technique employs a gradient echo, spin echo, fast-spin echo, driven equilibrium foumier transform, or spoiled gradient echo technique.

24. The map of any one of claim 16 to 23 wherein the movement pattern and the biochemical map are merged to show how the movement pattern of the joint interacts with the biochemical map of the cartilage. W: SASKIA'Patet Spe\2004203105div-13 8 07 doc