Image processing method

Abstract

The present invention relates to a method of determining cartilage properties in a target joint of a patient. The method includes: receiving data representing a joint gap width of a target joint; receiving data representing a relationship between the joint gap width and a cartilage property in the target joint, wherein the data representing the relationship between the joint gap width and the cartilage property in the target joint is based on a measured cartilage in the target joint population; and determining a cartilage property in the target joint based on the data representing the joint gap width of the target joint and the data representing a relationship between the joint gap width and the cartilage property in the target joint.

Description

Technical Field

[0001] This invention relates to methods and apparatus for predicting the state of musculoskeletal joints. More specifically, this invention relates to methods and apparatus for determining the cartilage properties in target joints (such as ankle, knee, hip, shoulder, spinal, or elbow joints). Background Technology

[0002] With an aging population, degenerative joint diseases have become very common, with approximately 8 million people suffering from osteoarthritis in the UK and approximately 27 million in the US.

[0003] The causes of osteoarthritis and some other degenerative joint diseases are not fully understood; however, many factors are believed to contribute to the likelihood of disease development in certain joints, such as the knee. These factors include a high body mass index, traumatic injuries to ligaments and menisci, valgus alignment of the knee, and how the joint is used in daily activities.

[0004] All tissues in the knee—including bone, cartilage, meniscus, ligaments, joint capsule, and synovial membrane—can be affected by degenerative joint disease. In individual patients, some or all of these tissues may be damaged, and at any given time, each patient's knee may contain any combination of damaged tissues.

[0005] The diagnosis of osteoarthritis is usually based on radiographic scores of bone changes (osteophytes and bone damage) and narrowing of the joint space (indicating loss of cartilage on the surface of the joint bones). For example, the Kellgren-Lawrence classification system divides osteoarthritis into one of five grades, each based on a combination of joint space narrowing, presence or absence of osteophytes, bone damage, and the severity of each.

[0006] Various medical and surgical interventions exist for degenerative joint diseases. For example, modern orthopedic surgery allows for routine surgical repair of joints, sometimes referred to as arthroplasty. Arthroplasty may involve reshaping the joint by modifying existing joint surfaces and bone to provide a more functional joint. In recent decades, surgical replacement of joints or joint surfaces with prostheses has become the most common and successful form of arthroplasty and has become a standard treatment for osteoarthritis. The decision to perform these medical and surgical interventions is usually based at least in part on a diagnosis based on radiographic scoring of bone changes and joint space narrowing.

[0007] However, improvements are still needed in providing clinicians with information on the status of target joints for the diagnosis and treatment of degenerative joint diseases such as osteoarthritis. Summary of the Invention

[0008] This article describes a method for determining the cartilage properties in a target joint of a patient. The method includes: receiving data representing the joint space width of the target joint; receiving data representing the relationship between the joint space width and the cartilage loss properties in the target joint, wherein the data representing the relationship between the joint space width and the cartilage loss properties in the target joint is based on cartilage loss measured in a population of target joints; and determining the cartilage properties in the target joint based on the data representing the joint space width of the target joint and the data representing the relationship between the joint space width and the cartilage loss properties in the target joint.

[0009] By determining the cartilage properties in a target joint according to the methods described herein, improvements can still be made in providing useful diagnostic information related to cartilage and patient outcomes, without resorting to typically complex, expensive, and time-consuming methods for imaging cartilage properties at a given time point.

[0010] In some implementations, the nature of cartilage loss refers to the probability of cartilage loss or the quality of cartilage structure. By determining the cartilage nature (i.e., the probability of cartilage loss), it is possible not only to identify signs that the target joint has lost more cartilage than normal, but also to determine that the target joint is likely to continue losing cartilage at a higher-than-normal rate. By determining the cartilage nature (i.e., the quality of cartilage structure), the target joint can be evaluated accordingly.

[0011] In some implementations, the data representing the joint space width of the target joint includes one or more normalized values ​​associated with the joint space width. By using one or more normalized values, it is easier to quantify variations and deviations relative to the mean.

[0012] In some implementations, each patient associated with one or more healthy joints does not exhibit symptoms associated with a disease affecting cartilage. For example, patients can be selected based on one or more ages, whether the patient's target joint is painful, or any other property associated with the disease. By selecting only patients associated with healthy joints, a normal distribution of the observed properties is determined. This distribution allows for the identification of values ​​that deviate from the normal distribution, accurately categorizing them as indicative of a specific property (such as cartilage loss).

[0013] In some implementations, one or more normalized values ​​are determined based on the patient's sex. By normalizing based on the patient's sex, significant differences in mean joint space width and / or cartilage thickness between men and women can be taken into account. A surprising finding is that sex-based normalization allows for the consideration of all significant variations. That is, it has been determined that to provide a single value representing cartilage thickness, only sex-based normalization is required, and other patient-specific properties need not be considered in the normalization process.

[0014] In some implementations, one or more normalized values ​​are determined based on healthy joint space width data representing the healthy joint space width of one or more healthy joints.

[0015] In some implementations, one or more normalized values ​​represent deviations from one or more means associated with one or more healthy joints corresponding to the target joint. In some implementations, the means are determined based on sex. Normalized values ​​can provide a standard score that allows for comparisons between patients. For example, the standard score could be a z-score.

[0016] In some implementations, data representing the joint space width is determined based on one or more images of the target joint. In some implementations, data representing the joint space width of the target joint is determined based on image processing not used to determine cartilage. In some implementations, one or more images are generated via computed tomography and / or magnetic resonance imaging. For example, data representing the joint space width can be determined from imaging modalities where cartilage is typically not identifiable or requires additional image processing to identify cartilage. Imaging modalities may prevent cartilage from being measured accurately.

[0017] In some implementations, data representing the joint space width of the target joint is determined based on first image data, and data representing the relationship between the joint space width and the cartilage properties in the target joint is determined based on second image data, which is different from the first image data. The data representing the joint space width can be determined based on image data obtained from an imaging modality different from the imaging modality used to obtain patient data for generating the model.

[0018] In some implementations, data representing the relationship between joint space width and cartilage properties in the target joints are obtained from multiple target joints, where both the joint space width and the cartilage properties in the target joints are known. The multiple target joints can come from individuals other than the patient. Thus, data representing cartilage properties in a population can be used to predict cartilage properties in an individual.

[0019] In some implementations, the target joint includes a lateral compartment and a medial compartment. In some implementations, the data representing the joint space width includes lateral joint space width data corresponding to the lateral compartment and / or medial joint space width data corresponding to the medial compartment. Additionally or alternatively, the lateral compartment and the medial compartment may be considered together.

[0020] In some implementations, the method described herein may also include: determining whether the target joint is affected by degenerative joint disease (such as osteoarthritis) based on cartilage properties.

[0021] In some implementations, data representing the joint space width is determined based on multiple measurements between predetermined points of the target joint. In some implementations, the data representing the joint space width includes the average of the multiple measurements, the percentiles of the multiple measurements, and / or the median of the multiple measurements. Percentiles can be determined based on a portion of the bone adjacent to each other in the imaging state of the target joint.

[0022] In some implementations, the target joint is the ankle, knee, hip, shoulder, spinal, or elbow joint. In some implementations, data representing the joint space width of the target joint is associated with one or more of the tibia, fibula, and patella of the knee joint. Furthermore, lateral and medial compartments may be considered.

[0023] In some implementations, the cartilage properties in the target joint can be used to select and / or form a prosthesis for the patient. Additionally or alternatively, the cartilage properties in the target joint can be used to determine the surgical intervention and / or surgical plan for the patient.

[0024] In some implementations, data representing the joint clearance width is generated based on the target joint under compression and / or load-bearing conditions. For example, the joint can be imaged when it is under compression and / or load-bearing conditions. By determining data representing the joint clearance width of the target joint under compression, any mechanical or positional factors that might affect the accurate representation of the joint clearance width can be mitigated.

[0025] This document also describes a computer program comprising the computer-readable instructions as described above. This document also describes a computer-readable medium carrying the computer program as described above. This document further describes a computer device for determining the cartilage properties in a target joint of a patient, the device comprising: a memory storing processor-readable instructions; and a processor configured to read and execute the instructions stored in the memory, the processor-readable instructions including the instructions as described above.

[0026] Although specific implementations have been described above, it should be understood that the present invention can be implemented in ways other than those described. The above description is intended to be illustrative and not restrictive. Therefore, it will be apparent to those skilled in the art that modifications can be made to the described invention without departing from its spirit. Attached Figure Description

[0027] Figure 1 A system for generating anatomical data is shown, which provides the probability of cartilage loss in a patient's target joint.

[0028] Figure 2A computer system that can implement the techniques described herein is shown.

[0029] Figure 3 A flowchart of a method for generating anatomical data that provides the probability of cartilage loss in a patient's target joint is shown.

[0030] Figure 4a The knee joint, its corresponding compartments, and a patch used to measure the width of the joint space are shown.

[0031] Figure 4b A mask used for shape modeling of the tibia is shown.

[0032] Figure 5a Two normal distributions of the width of the medial joint space are shown, one for male patients and the other for female patients.

[0033] Figure 5b Two normal distributions of the lateral joint space width are shown, one for male patients and the other for female patients.

[0034] Figure 6 A system for generating relational data representing the relationship between joint space width and the likelihood of cartilage loss is shown.

[0035] Figure 7 A flowchart of a method for determining relational data is shown.

[0036] Figure 8 A scatter plot of the width of the medial and lateral joint spaces and its corresponding representation of the probability of cartilage loss are shown.

[0037] Figure 9 The first relationship between the range of values ​​indicating the joint space width and the corresponding likelihood of cartilage loss is shown.

[0038] Figure 10 A second relationship is shown between the range of values ​​indicating the joint space width and the corresponding likelihood of cartilage loss.

[0039] Figure 11 A knee joint, its corresponding compartments, and the corresponding cartilage loss probability of each compartment are shown.

[0040] The reference numerals that appear repeatedly in multiple figures are intended to identify the same features in various implementations. Detailed Implementation

[0041] See Figure 1Computer 102 is configured to receive patient data 104, which represents the joint space width associated with a target musculoskeletal joint of the patient. For example, the target musculoskeletal joint of the patient could be the knee or ankle joint. It will be readily appreciated that the methods and systems described herein can be applied to any suitable joint. Although the term "patient" is used herein, it should be understood that these techniques can be applied to data associated with any individual. Patient data 104 represents the joint space between the bones of the patient's target joint. Patient data 104 is obtained by parametrically analyzing the patient's target musculoskeletal joint based on three-dimensional image data in which features of the target joint have been identified. For example, the three-dimensional image data could be X-ray computed tomography image data, and the parametric can be generated by fitting a statistical model (such as an active appearance model or an active shape model) to the three-dimensional image data, which is created based on a training set of images of target joints of the same type as the target joint of the subject corresponding to patient data 104. The model can be any statistical model of variation within the training images. For example, a statistical model can be generated by processing the image training set to generate an average model of the training set and a range of variation relative to the average model. Parametric representations are used to represent features of a target joint in image data, and therefore can be used to identify features of the represented target joint. A suitable model fitting technique is described in International Patent Publication No. WO2011 / 098752, the contents of which are incorporated herein by reference.

[0042] As described in further detail below, patient data can provide a normalized joint space width for the patient. Patient data can be normalized for the patient's gender. Additionally or alternatively, patient data can be normalized based on the joint space width associated with a healthy joint of the target joint, and the patient data can provide, for example, a z-score or other scores representing the relationship between the patient's joint space width and that of a typical healthy joint. Patient data can be associated with multiple measurements obtained from the target joint, and can, for example, provide the average of the multiple measurements and / or the average of specific percentiles of these measurements.

[0043] Computer 102 is further configured to receive relational data 106. Relational data 106 is data representing the relationship between joint space width and properties associated with the cartilage of the target joint. Although properties representing the likelihood of cartilage loss in the target joint will be mentioned below, it should be understood that other cartilage properties can be modeled using appropriate relational data that models the relationship between joint space width and cartilage properties. Relational data can be generated based on analysis of the patient's target joint to determine the joint space width in the patient's target joint and data (or another property as described above) indicating cartilage loss in the patient's target joint. As described in further detail below, relational data can define the relationship between specific values ​​of patient data 104 representing the joint space of the patient's target joint and the likelihood of cartilage loss associated with patient data 104.

[0044] Data indicative of cartilage loss can be generated by examining the target joint using any known method to determine the corresponding joint space width and cartilage loss. In one instance, data indicative of cartilage loss in the target joint can be determined based on cartilage thickness measurements obtained for each target joint of a patient. For example, if the cartilage thickness of a particular target joint is determined to be less than the 95th percentile cartilage thickness of a population (including one or more corresponding target joints), the specific target joint of the patient used to generate relational data 106 can be classified as indicative of cartilage loss. In other words, cartilage loss of the target joint can be determined by comparing the cartilage thickness of the target joint with cartilage thickness data of the population. Relational data 106 can be generated based on joint space width data and cartilage data obtained from a population with healthy joints (e.g., joints of individuals not showing any signs of disease (such as OA)). Cartilage thickness can be a normalized value as described herein (e.g., a value normalized based on the sex of the patient associated with each joint). The inventors recognize that the cartilage thickness of a population with healthy joints follows an unexpectedly normal distribution when normalized only for sex. Given this normal distribution, if cartilage thickness falls outside the 95th percentile, finding cartilage thickness provides useful information about the likelihood of cartilage loss in a patient. The likelihood of cartilage loss can indicate past and / or future cartilage loss.

[0045] Patient data 104 is processed based on relational data 106 to generate anatomical data 108. Anatomical data 108 provides the probability of cartilage loss in the patient's target joint. Anatomical data 108 can be a discrete or continuous numerical value representing the probability of cartilage loss, or it can be a label representing the patient data 104 corresponding to a category representing a classification indicating the probability of cartilage loss.

[0046] Obtaining accurate measurements of cartilage thickness can be challenging and expensive. For example, it is often impossible to image cartilage using methods other than magnetic resonance imaging or arthroscopic computed tomography (CT), which are not widely available. However, the inventors recognized that by determining the relationship between joint space width and cartilage loss, the likelihood of cartilage loss could be determined, eliminating the need for direct imaging of the cartilage.

[0047] Patient data 104 may be three-dimensional image data of the patient's knee joint, and relational data 106 may be data representing the relationship between the joint space width of the patient's knee joint and the probability of cartilage loss in the patient's knee joint. Anatomical data 108 may include data representing the probability of cartilage loss in the patient's knee joint. In this example, data representing the probability of cartilage loss in the patient's knee joint can be used to determine that cartilage in the knee joint has been lost and / or that the cartilage in the knee joint will be lost relatively quickly compared to knees not indicated to have the probability of cartilage loss. Data representing the probability of cartilage loss may represent one or more of a variety of patient characteristics (e.g., the patient may have degenerative arthritis or the biomechanics of the knee joint has changed). As described in further detail below, the probability may be associated with one or more compartments of the target joint. The probability of cartilage loss in the knee joint can provide useful information when medical interventions are performed on the knee (e.g., knee surgery). In this way, knee joint-related information that may not be available in other cases can be made available in a probabilistic and data-driven manner, thereby helping to improve patient outcomes. In some implementations, a diagnosis of OA can be given by determining the probability of cartilage loss.

[0048] It should also be recognized that anatomical data¹⁰⁸ can be used to provide medical interventions. For example, by determining the likelihood of cartilage loss, appropriate medical interventions specifically targeting cartilage loss and / or any associated conditions can be identified. For instance, the likelihood of cartilage loss can be used to determine whether to perform a surgical or pharmacological intervention.

[0049] Anatomical data 108 can be used to provide medical interventions. For example, by determining the likelihood of cartilage loss in a patient's target joint, a surgical intervention can be tailored to the patient's joint itself. For example, a surgical plan suitable for guiding a surgeon or automated surgical equipment can be generated based on anatomical data 108. For example, the surgical plan can provide data indicating how surgical remodeling of the joint can restore it to a disease-free state and / or to a state where it can function effectively. For example, the likelihood of cartilage loss in each of the multiple compartments of the target joint can be determined. These likelihoods can be used to determine the state associated with each of the multiple compartments. Based on the likelihood of each corresponding compartment, a specific surgical intervention can be determined. For example, in some implementations, anatomical data 108 can be used to determine whether to perform a unicompartmental knee replacement or a total knee replacement.

[0050] Additionally or alternatively, the anatomical data 108 can be used for non-surgical interventions (such as analyzing the efficacy of medications, physical therapy, or any other medical interventions that can be used to alleviate symptoms). The symptoms may include osteoarthritis. The anatomical data 108 may include data suitable for generating prostheses for the patient's target musculoskeletal joints.

[0051] Figure 2 Computer 1 is shown, and computer 1 is described in further detail. Figure 1 Computer 102. It can be seen that computer 1 includes a CPU 1a, which is configured to read and execute instructions stored in volatile memory 1b, which takes the form of random access memory. Volatile memory 1b stores the instructions executed by CPU 1a and the data used by those instructions. For example, in use, Figure 1 The relational data 106 and patient data 104 can be stored in volatile memory 1b.

[0052] Computer 1 also includes non-volatile memory in the form of a hard disk drive 1c. Data generated from relational data 106 and patient data 104 can be stored on the hard disk drive 1c. Computer 1 also includes an I / O interface 1d, to which peripheral devices used in conjunction with computer 1 are connected. More specifically, a display 1e is configured to display output from computer 1. For example, display 1e can display a representation of patient data 104, relational data 106, and / or anatomical data 108. Input devices are also connected to I / O interface 1d. These input devices may include a keyboard 1f and a mouse 1g that allow users to interact with computer 1. Network interface 1h allows computer 1 to connect to a suitable computer network to receive and transfer data to other computing devices. CPU 1a, volatile memory 1b, hard disk drive 1c, I / O interface 1d, and network interface 1h are connected together via bus 1i.

[0053] Figure 3 The process performed to generate anatomical data 108 is shown. This process can be performed on computer 1.

[0054] In step 302, patient data 104 is received. As described herein, patient data 104 represents the joint space width of a target joint in the patient. In one instance, patient data 104 includes the joint space width measurement described herein. Alternatively, patient data 104 may include one or more normalized values ​​indicating the joint space width. For example, one or more normalized values ​​may be one or more z-scores.

[0055] In step 304, relational data 106 is received. As described herein, relational data 106 is data representing the relationship between the joint space width and the probability of cartilage loss in the target joint. In one instance, relational data 106 may include one or more ranges of values, each range corresponding to a different probability of cartilage loss.

[0056] In step 306, patient data 104 and relational data 106 are processed to generate anatomical data 108 associated with the patient's target musculoskeletal joint. This processing may include comparing patient data 104 with relational data 106. For example, as described further below, different values ​​or ranges of values ​​in the relational data may correspond to specific cartilage loss probabilities. For instance, a first range of values ​​in relational data 106 may correspond to a lower cartilage loss probability, while a second range of values ​​in relational data 106 may correspond to a higher cartilage loss probability. If patient data 104 (i.e., joint space width) falls within the second range and not within the first range, this could indicate a higher probability of cartilage loss in the target joint.

[0057] Relationship data 106 provides a model that allows for the prediction of cartilage loss based on a patient's joint space width data. Relationship data 106 can be determined based on analysis of image data associated with multiple patients, where both cartilage loss and joint space width can be determined as described above. The inventors were surprised to realize that, based on the joint space width of a specific patient, the relationship between joint space width in a population and cartilage loss in the same population can be used to predict cartilage loss in that specific patient. While it is generally accepted that joint space narrowing in a specific patient can indicate cartilage loss in the joint, it is generally believed that many factors contribute to joint space width and cartilage loss in complex ways that are not fully understood. Therefore, joint space width has previously been used as one of many indicators for determining the development of diseases such as osteoarthritis. However, the inventors recognized that relationship data 106, as described herein, can be determined, through which the likelihood of cartilage loss can be accurately predicted without requiring expensive imaging modalities or computationally costly image processing.

[0058] Therefore, anatomical data 108 can be used to predict cartilage loss in the target joint. Thus, anatomical data can be used to refine the determination of whether to provide medical intervention (e.g., perform total knee arthroplasty).

[0059] As described above, relationship data 106 can be determined based on the analysis of image data associated with multiple patients, where cartilage loss and joint space width can be determined. Joint space width represents the distance between the two bones of a joint. The processing of image data to determine joint space width will be described in further detail below. While identifying bones in joint images is relatively straightforward, obtaining accurate data representing cartilage is often more challenging. For example, magnetic resonance imaging (MRI) can be used to obtain image data for determining relationship data 106. Image data can be manually examined and labeled, or automated methods based on, for example, shape modeling and deep learning can be used to process the image data to determine cartilage thickness, thereby accurately identifying the cartilage. An exemplary technique for measuring cartilage thickness is described in Bowes et al.’s “Precision, Reliability, and Responsiveness of a Novel Automated Quantification Tool for Cartilage Thickness: Data from the Osteoarthritis Initiative” (The Journal of Rheumatology, February 2020, 47 (2) 282-289), the contents of which are incorporated herein by reference.

[0060] The joint space width in a target joint can be determined in any convenient manner. Generally, the joint space width associated with patient data and the joint space width associated with image data associated with multiple patients are determined in a corresponding manner to achieve consistency between patient data and relational data. A method for determining the joint space width in a target joint will now be described. Although the joint space width is described below for the knee joint, it should be understood that the technique can also be applied to other target joints.

[0061] See Figure 4a The image shows three-dimensional image data of the patella 401, femur 402, and tibia 403 of the knee joint. Patches 404-411 are each associated with a corresponding surface of the bone in the knee joint, which can be used to determine the joint space width at different locations of the knee joint. Each patch 404-411 may include multiple target points associated with a surface of the corresponding bone surface. Multiple target points can be determined by determining a segmentation of the bone surface and fitting a mask to that segmentation. Segmentation can be implemented in any convenient manner, such as using the model fitting technique described in International Patent Publication No. WO2011 / 098752. Figure 4b An exemplary mask is shown, which illustrates multiple lines on which multiple target points lie. However, it should be understood that the mask can take any convenient form and can be used with... Figure 4b The target points shown are different target points.

[0062] The distance between one or more of the multiple facets 404 to 411 can be determined. For example... Figure 4a As shown, patches 404 and 406 are associated with the lateral patellofemoral compartment, patches 405 and 407 with the medial patellofemoral compartment, patches 408 and 410 with the lateral tibiofemoral compartment, and patches 409 and 411 with the medial tibiofemoral compartment. The joint space width of one or more of the lateral patellofemoral compartment, medial patellofemoral compartment, lateral tibiofemoral compartment, and medial tibiofemoral compartment can be determined based on the three-dimensional spatial distance between the corresponding patches associated with a particular compartment. In particular, each pair of patches associated with a particular compartment may include multiple points, wherein each of the multiple points in the first patch of that pair of patches is associated with a corresponding point in the multiple points in the second patch of that pair of patches. The distance between each associated corresponding point can be determined to provide multiple distances between associated corresponding points of a particular compartment. Multiple determined distances can be combined to provide a representation of the joint space width for a particular compartment (e.g., the lateral patellofemoral compartment, etc.). In one example, when generating... Figure 4a and Figure 4bWhen the three-dimensional image data shown is presented, the knee joint may be under compression. For example, a patient may be standing and / or bearing weight, thus applying compressive forces to the knee joint. In another instance, the target joint can be flexed within a specific range of flexion. In this instance, the range of flexion could be approximately 15 degrees or greater (e.g., flexion between 15 and 20 degrees). In other words, when the target joint is under compression, the representation of the facets, distances, and joint space width can be determined. This reduces any mechanical or positional factors that might affect the accurate representation of the joint space width in a particular compartment.

[0063] Multiple distances can be combined in any convenient manner to provide a representation of the joint space width for a specific compartment. For example, an average value can be determined to represent the average of multiple points. This average value can be the mean, median, or mode. Additionally or alternatively, multiple distances can be combined by obtaining predetermined percentiles of the distance values. For example, the predetermined percentile of the distance values ​​could be the smallest 10 percentile or the smallest 20 percentile, etc. For example, predetermined percentiles of distance values ​​can be combined by averaging as described above. It has been found that improvements can be achieved in specific compartments by employing predetermined percentiles of distance values. For example, for some measurement techniques, it has been determined that cartilage loss in the patellofemoral compartment occurs only in a small portion of the facet in some patients. Therefore, by employing predetermined percentiles of distance values ​​in these compartments, joint space information can be obtained that represents the damage in the target compartment in an improved manner.

[0064] In some implementations, the determination of cartilage loss in the patellofemoral compartment can be refined by imaging the knee under predetermined conditions. For example, imaging a knee with a flexion angle of approximately 15 degrees or greater can refine the determination of cartilage in the patellofemoral compartment. When imaging a slightly flexed or non-flexed knee, the patella dislocates from the femur, making imaging the cartilage between the patella and femur more challenging. As mentioned above, it has been determined that using the smallest percentile of the measurements can alleviate this challenge. In some implementations, a specific percentile of the measurements used can be determined based on the knee imaging modality. In some implementations, a shape model can be used to process the joint to determine whether the cartilage can be identified by processing the joint to determine the position of the patella relative to the femur. In some implementations, the joint can be imaged in a standing position. The joint can be imaged with a predetermined flexion angle applied to it, for example, using a knee brace. In some implementations, predetermined conditions are applied to the joints used to generate relational and patient data to ensure consistency between the model and the input data.

[0065] The representation of joint space width for a specific joint may also include a normalized value. For example, the joint space width determined as described above in conjunction with Figure 4 may be further normalized to provide patient data 104 and / or joint space width data for generating relational data 106. In some implementations, the joint space width determined as described above is normalized relative to the patient's gender. Figure 5a and Figure 5b Charts showing the joint space widths of the medial and lateral tibiofemoral compartments in the entire group, where... Figure 5a and Figure 5b Each of the accompanying diagrams illustrates the target joints associated with males and females, respectively. From Figure 5a and Figure 5b As shown in each of the attached figures, in every joint of both the medial and lateral joints, the joint space width is greater in males than in females. The joint space width data for each joint in the patient and / or joint group used to generate relationships can be normalized by adjusting the joint space width for the patient's gender associated with the joint.

[0066] In some implementations, the data is additionally or alternatively normalized based on the healthy joint space width. For example, the joint space width measurements described above in conjunction with Figure 4 can be normalized by determining the deviation of the joint space width from the average healthy joint space width. For example, normalization based on healthy joint space width can produce a z-score of the joint space width based on the standard deviation and average joint space width determined across the entire population of healthy joints (e.g., joints where no cartilage loss was observed).

[0067] Healthy joint populations can be identified for both male and female groups, and therefore normalization based on healthy joint space width can provide normalization for both healthy joint space width and sex. In some implementations, only two healthy joint populations can be used, making normalization based solely on healthy joint space width and sex. That is, in some implementations, other patient properties can be disregarded in the normalization process. The inventors recognized that sex is the only factor significantly significant enough to require correction affecting healthy cartilage thickness. Therefore, the inventors were surprised to find that, in order to provide a model for predicting the likelihood of cartilage loss based on joint space width, it is not necessary to normalize properties other than patient sex. This improves the accuracy of predicting the likelihood of cartilage loss.

[0068] Figure 6 The generation of relational data 106 is illustrated schematically. For example... Figure 6 As shown, computer 604 (this computer can be) Figure 1 and Figure 2The computer (which may be the same computer or different computers) receives patient data 602. Patient data 602 may include joint space width data associated with multiple target joints and corresponding cartilage loss data indicating the presence of cartilage loss in each of the multiple target joints. The computer is configured to generate relational data 606 as output. Relational data 606 may be... Figure 1 Relationship data 106. Relationship data 606 represents the relationship between joint space width and cartilage loss. Patient data 602 may include, as described in this article. Figure 4a and Figure 4b The obtained joint space width data may be, for example, normalized joint space width data, and may be associated with a specific group (such as a single-sex group).

[0069] Figure 7 A flowchart of a method for generating relational data 106 is shown. In step 702, joint gap width data is received. This joint gap width data includes a representation of the joint gap width for one or more target joints in a group. The joint gap width data may include one or more joint gap width values, each corresponding to a different joint gap width (e.g., lateral joint gap width and medial joint gap width) for a specific joint among the one or more target joints. In one instance, the one or more target joints are joints of the same type. In one instance, the group is distinguished based on gender.

[0070] In step 704, cartilage loss data is received. The cartilage loss data includes a representation of cartilage loss for each joint in one or more target joints within the population. The cartilage loss representation can be a continuous or discrete numerical value, or it can be a label. For example, continuous or discrete numerical values ​​or labels can indicate that the corresponding joint space width indicates cartilage loss.

[0071] In step 706, relational data 106 is determined. Relational data 106 can be determined based on received joint space width data and received cartilage loss data. Relational data 106 can indicate the likelihood of cartilage loss based on the joint space width of a specific target joint. For example, statistical methods can be used to determine relational data 106. In another instance, relational data 106 can be determined by segmenting joint space width values ​​based on corresponding cartilage loss data, as described below. Figure 8 As described above. In another example, a machine learning algorithm can be used to determine the relational data 106, which is configured to learn the relationship between joint space width values ​​and cartilage loss. In some implementations, the relational data can represent the relationship between joint space width and cartilage loss for multiple target joints.

[0072] Figure 8Patient data 802 suitable for determining relationship data 808 is shown. For each of the multiple patients, the patient data includes the normalized medial tibiofemoral intercompartmental space width and the normalized lateral tibiofemoral joint space width. Figure 8 In the diagram, each of the x and y axes represents a z-score, which indicates the deviation of the corresponding joint space width from the mean joint space width of the healthy joint population associated with patients of the same sex as the corresponding joint space width. Because joint space widths are normalized in this way, data associated with male and female patients can be considered together. Figure 8 The color of the point associated with a specific patient indicates whether the specific patient associated with the point has been identified as having cartilage loss in the medial joint space, lateral joint space, or medial and lateral joint space, or neither.

[0073] from Figure 8 As can be seen, patients with cartilage loss in specific joints cluster in specific regions associated with specific z-scores. Based on the distribution of patients with and without cartilage loss associated with specific z-scores, the probability of a patient having cartilage loss can be determined for each z-score associated with a specific joint space width (in... Figure 8 Specifically, this refers to the likelihood that the patient has cartilage loss in the medial compartment, lateral compartment, or both. By determining the likelihood for each compartment, improved medical interventions can be made. For example, if the cartilage loss is likely only in the medial compartment, this can inform the surgeon that only the medial compartment needs surgery, rather than replacing the entire knee.

[0074] Figure 9 It shows that according to Figure 8 The data shown represents various probabilities of cartilage loss associated with a specific z-score; however, it should be recognized that the probability of a patient having cartilage loss can be provided in any convenient form. For example, in some implementations, the probability is provided as an output suitable for clinician use (e.g., Figure 10 (As shown). In Figure 10 In the equation, the z-score is related to the thermal properties representing the probability of cartilage loss. Figure 1 It begins to appear. However, it should be recognized that, although Figure 9 and Figure 10 Each provides a probability representation suitable for clinicians, but patient data provides a value between 0 and 1 for each z-score or combination of z-scores.

[0075] although Figures 8 to 10Based on two target joints, however, it should be recognized that other numbers of target joints can be modeled. For example, each of the medial and lateral tibiofemoral compartments and the medial and lateral patellofemoral compartments can be modeled individually or in combination to provide the probability of cartilage loss based on corresponding combinations of one or more joint space widths.

[0076] Figure 11 The four compartments of the knee joint are shown. Specifically, the medial patellofemoral compartment 1102, the lateral patellofemoral compartment 1104, the medial tibiofemoral compartment 1106, and the lateral tibiofemoral compartment 1108 are shown. For each compartment of the knee joint, a superimposed z-score is used to represent the probability of cartilage loss. For example, the superimposed z-score for the lateral tibiofemoral compartment 1108 is 0, indicating that there may be cartilage loss in this compartment. In another example, the superimposed z-score for the lateral patellofemoral compartment 1104 is -0.9, indicating that there is a high probability of cartilage loss in this compartment. A z-score of 0 indicates that the cartilage thickness in the given compartment is equivalent to the average cartilage thickness of the knee joint population. A z-score of -1 indicates that the cartilage thickness in the given compartment is one standard deviation less than the average cartilage thickness of the knee joint population. A z-score of 1 indicates that the cartilage thickness in the given compartment is one standard deviation greater than the average cartilage thickness of the knee joint population. In this case, the knee joint, including the medial patellofemoral compartment 1102, the lateral patellofemoral compartment 1104, the medial tibiofemoral compartment 1106, and the lateral tibiofemoral compartment 1108, is likely to have cartilage loss in three of its four compartments. It is readily apparent that various methods can be conceived for representing the likelihood of cartilage loss. For example, Figure 11 The superimposed z-scores on the knee joint are shown, with four corresponding z-score ranges 1110, 1112, 1114, and 1116, each range used to determine the likelihood of cartilage loss in its respective knee joint compartment. Other scoring metrics can be envisioned, such as percentages or percentile scores representing one of multiple percentile ranges, T-scores based on T-score ranges, cartilage thickness measurements based on cartilage thickness measurement ranges, or any other suitable scoring system. As described herein, this may indicate not only that the knee joint has lost more cartilage than normal, but also that the knee joint may continue to lose cartilage at a higher-than-normal rate.

[0077] To determine the likelihood of cartilage loss in a new patient, the medial and lateral joint space width data determined for that patient can be input into the model (i.e., Figure 1 Patient data (104) were used to determine the likelihood of cartilage loss. For example, in cases such as... Figures 8 to 10 In the case of modeling the medial and lateral joints, the probability can be directly determined by inputting the width of the medial and lateral joint space. As mentioned above, the probability can be a value between 0 and 1, where a value close to 0 indicates a lower probability of cartilage loss in the patient, while a value close to 1 indicates a higher probability of cartilage loss in the patient.

[0078] For example, if a patient reports joint pain to a clinician, the methods and systems described herein can be used to determine the likelihood of cartilage loss. Based on joint space width measurements, it can be determined whether cartilage loss may have occurred. As described herein, this joint space width measurement is easier to perform than directly and precisely measuring the thickness of the articular cartilage. It is possible that clinicians may implement medical interventions based on observed cartilage loss even when no loss has occurred. Cartilage loss can be detected based on observed narrowing of the joint space width. For example, narrowing of the joint space width may occur if the meniscus within the interspace no longer functions properly. In this case, there may be no cartilage loss, but the joint space width may still be reduced, leading to the erroneous assumption of cartilage loss. The methods and systems described herein overcome this problem by determining the likelihood of cartilage loss.

[0079] As mentioned above, while the likelihood of cartilage loss has been generally discussed, other properties can be modeled and determined. For example, patient data used to train a model can represent the structural quality associated with cartilage. The T2 values ​​from MRI images can be used to determine the structural quality associated with cartilage. Additionally or alternatively, this property can represent the likelihood of complete exposure at a point or set of points, indicating bone-on-bone contact at the associated imaging point. Additionally or alternatively, using a training model containing data from multiple time points, this property can represent the future likelihood of cartilage loss at a future time point. Relational data can be processed against joint space data, and the output can indicate the cartilage properties associated with the patient data used to train the model.

[0080] The methods and systems described herein are illustrated with reference to the generation and / or determination of the likelihood of cartilage loss. However, it is readily apparent that any other cartilage properties in the patient's target joint can be determined. For example, other cartilage properties may include cartilage condition and / or status.

Claims

1. A method for determining the cartilage properties in a target joint of a patient, the method comprising: Receive data representing the joint space width of the target joint; Receive data representing the relationship between joint space width and cartilage properties in the target joint, wherein the data representing the relationship between joint space width and cartilage properties in the target joint is based on tested cartilage in a population of target joints; as well as The cartilage properties in the target joint are determined based on the data representing the joint space width of the target joint and the data representing the relationship between the joint space width and the cartilage properties in the target joint.

2. The method according to claim 1, wherein, The cartilage properties refer to the likelihood of cartilage loss and / or the quality of cartilage structure.

3. The method according to claim 1 or 2, wherein, The data representing the joint space width of the target joint includes one or more normalized values ​​associated with the joint space width.

4. The method according to claim 3, wherein, The one or more normalized values ​​are determined based on the patient's gender.

5. The method according to claim 3 or 4, wherein, The one or more normalized values ​​are determined based on healthy joint space width data representing the healthy joint space width of one or more healthy joints.

6. The method according to any one of the preceding claims, wherein, Each patient with one or more healthy joints did not exhibit symptoms associated with a disease affecting cartilage.

7. The method according to any one of claims 3 to 6, wherein, The one or more normalized values ​​represent deviations from one or more average values ​​associated with one or more healthy joints corresponding to the target joint.

8. The method according to any one of the preceding claims, wherein, The data representing the joint space width is determined based on one or more images of the target joint.

9. The method according to claim 8, wherein, The data representing the joint space width of the target joint is determined based on image processing that is not used to determine cartilage.

10. The method according to claim 8 or 9, wherein, The one or more images are generated by computed tomography and / or magnetic resonance imaging.

11. The method according to claim 3 or any of its dependent claims, wherein, The data representing the joint space width of the target joint is determined based on the first image data, and the data representing the relationship between the joint space width and the cartilage properties in the target joint is determined based on the second image data, which is different from the first image data.

12. The method according to claim 3 or any of its dependent claims, wherein, The data representing the relationship between joint space width and cartilage properties in the target joint is obtained based on data obtained from multiple target joints, wherein the joint space width and cartilage loss in the target joint are known.

13. The method according to any one of the preceding claims, wherein, The target joint includes a lateral compartment and a medial compartment.

14. The method according to claim 13, wherein, The data representing the joint space width includes lateral joint space width data corresponding to the lateral compartment and medial joint space width data corresponding to the medial compartment.

15. The method according to any one of the preceding claims further comprises: The cartilage properties are used to determine whether the target joint is affected by osteoarthritis.

16. The method according to any one of the preceding claims, wherein, The data representing the joint gap width is determined based on multiple measurements between predetermined points of the target joint.

17. The method according to claim 16, wherein, The data representing the joint space width includes the average of the plurality of measurements, the percentiles of the plurality of measurements, and / or the median of the plurality of measurements.

18. The method according to any one of the preceding claims, wherein, The target joint is the ankle, knee, hip, shoulder, spinal joint, or elbow joint.

19. The method according to claim 18, wherein, The data representing the joint space width of the target joint is associated with one or more of the tibia, fibula, and patella of the knee joint.

20. The method according to any one of the preceding claims, wherein, The cartilage properties in the target joint are used to select and / or form a prosthesis for the patient.

21. The method according to any one of the preceding claims, wherein, The cartilage properties of the target joint are used to determine the surgical intervention and / or surgical plan for the patient.

22. The method according to any one of the preceding claims, wherein, The joint gap width is generated based on the target joint under compression and / or load-bearing conditions.

23. A computer program comprising computer-readable instructions configured to cause a computer to perform the method according to any one of claims 1 to 22.

24. A computer-readable medium carrying a computer program according to claim 22.

25. A computer device for determining the cartilage properties in a target joint of a patient, comprising: Memory, which stores instructions that can be read by the processor; as well as A processor configured to read and execute instructions stored in the memory; The processor-readable instructions include instructions configured to control the computer to perform the method according to any one of claims 1 to 22.

Images