Computer program products and methods for detection and tracking of rheumatoid arthritis

A method for producing an image indicative of arthritic symptomatology, said method comprising:

• a. during a model building phase creating a constrained deformable statistical template characterized by a set of parameters that spans a vector space representing a set of spatial positions of specified features of a specified joint, each vector in said vector space describes a unique variation from a population mean, wherein creating said statistical template comprises building a statistical template from a set of sample radiographs of a skeletal system that includes at least said specified joint, said set of sample radiographs chosen to substantially exclude pathological variations of a specified joint to represent variations in radiographic shape and appearance;
• b. during a runtime phase fitting, in a digitized target two-dimensional radiograph of said specified joint, a plurality of loci associated with said specified features of said specified joint to allow parameterization in terms of said statistical template;
• c. during said runtime phase parameterizing positions of said plurality of loci fitted in said digitized target radiograph to generate an instance of said statistical template by deriving values for parameters of said digitized target radiograph from said statistical template;
• d. during said runtime phase searching said digitized target radiograph by applying said statistical template to said digitized target radiograph until an optimal fit of said statistical template to at least one region of said digitized target radiograph is found, wherein searching said digitized target radiograph for said optimal fit comprises adjusting the position, orientation, shape and texture of said statistical template until the differences between said statistical template and said at least one region of said digitized target radiograph are minimized;
• and
• e. during said runtime phase producing a comparison image by comparing said at least one region of said digitized target radiograph with said optimal fit of said statistical template.

A computer program product for producing an image indicative of arthritic symptomatology, said computer program product comprising:

• a. a constrained deformable statistical template characterized by a set of parameters that spans a vector space representing a set of spatial positions of specified features of a specified joint, each vector in said vector space describes a unique variation from a population mean, wherein said statistical template comprises is built from a set of sample radiographs of a skeletal system that includes at least said specified joint, said set of sample radiographs chosen to substantially exclude pathological variations of a specified joint to represent variations in radiographic shape and appearance;
• b. code for fitting, in a digitized target two-dimensional radiograph of said specified joint, a plurality of loci associated with said specified features of said specified joint to allow parameterization in terms of said statistical template;
• c. code for parameterizing positions of said plurality of loci fitted in said digitized target radiograph to generate an instance of said statistical template by deriving values for parameters of said digitized target radiograph from said statistical template;
• d. code for searching said digitized target radiograph by applying said statistical template to said digitized target radiograph until an optimal fit of said statistical template to at least one region of said digitized target radiograph is found, wherein searching said digitized target radiograph for said optimal fit comprises adjusting the position, orientation, shape and texture of said statistical template until the differences between said statistical template and said at least one region of said digitized target radiograph are minimized;
• and
• e. code for producing a comparison image by comparing said at least one region of said digitized target radiograph with said optimal fit of said statistical template.

A method for deriving measures indicative of arthritic symptomatology, said method comprising:

• a. during a model building phase creating a constrained deformable statistical template characterized by a set of parameters that spans a vector space representing a set of spatial positions of specified features of a specified joint, each vector in said vector space describes a unique variation from a population mean, wherein creating said statistical template comprises building a statistical template from a set of sample radiographs of a skeletal system that includes at least said specified joint, said set of sample radiographs chosen to substantially exclude pathological variations of a specified joint to represent variations in radiographic shape and appearance;
• b. during a runtime phase fitting, in a digitized target two-dimensional radiograph of said specified joint, a plurality of loci associated with said specified features of said specified joint to allow parameterization in terms of said statistical template;
• c. during said runtime phase parameterizing positions of said plurality of loci fitted in said digitized target radiograph to generate an instance of said statistical template by deriving values for parameters of said digitized target radiograph from said statistical template;
• d. during said runtime phase searching said digitized target radiograph by applying said statistical template to said digitized target radiograph until an optimal fit of said statistical template to at least one region of said digitized target radiograph is found, wherein searching said digitized target radiograph for said optimal fit comprises adjusting the position, orientation, shape and texture of said statistical template until the differences between said statistical template and said at least one region of said digitized target radiograph are minimized;
• and
• e. during said runtime phase deriving at least one measure indicative of arthritic symptomatology of said specified joint by comparing said at least one region of said digitized target radiograph with said optimal fit of said statistical template.

A computer program product for deriving measures indicative of arthritis symptomatology, said computer program product comprising:

• a. a constrained deformable statistical template characterized by a set of parameters that spans a vector space representing a set of spatial positions of specified features of a specified joint, each vector in said vector space describes a unique variation from a population mean, wherein said statistical template comprises is built from a set of sample radiographs of a skeletal system that includes at least said specified joint, said set of sample radiographs chosen to substantially exclude pathological variations of a specified joint to represent variations in radiographic shape and appearance;
• b. code for fitting, in a digitized target two-dimensional radiograph of said specified joint, a plurality of loci associated with said specified features of said specified joint to allow parameterization in terms of said statistical template;
• c. code for parameterizing positions of said plurality of loci fitted in said digitized target radiograph to generate an instance of said statistical template by deriving values for parameters of said digitized target radiograph from said statistical template;
• d. code for searching said digitized target radiograph by applying said statistical template to said digitized target radiograph until an optimal fit of said statistical template to at least one region of said digitized target radiograph is found, wherein searching said digitized target radiograph for said optimal fit comprises adjusting the position, orientation, shape and texture of said statistical template until the differences between said statistical template and said at least one region of said digitized target radiograph are minimized;
• and
• e. code for deriving at least one measure indicative of arthritic symptomatology of said specified joint by comparing said at least one region of said digitized target radiograph with said optimal fit of said statistical template.

A method comprising:

• a. during a model building phase creating a constrained deformable statistical template characterized by a set of parameters that spans a vector space representing a set of spatial positions of specified features of a specified joint, each vector in said vector space describes a unique variation from a population mean, wherein creating said statistical template comprises building a statistical template from a set of sample radiographs of a skeletal system that includes at least said specified joint, said set of sample radiographs chosen to substantially exclude pathological variations of a specified joint to represent variations in radiographic shape and appearance;
• b. during a runtime phase fitting, in a first digitized target two-dimensional radiograph of said specified joint, a plurality of loci associated with said specified features of said specified joint to allow parameterization in terms of said statistical template;
• c. during said runtime phase parameterizing positions of said plurality of loci fitted in said first digitized target radiograph to generate an instance of said statistical template by deriving values for parameters of said digitized target radiograph from said statistical template;
• d. during said runtime phase searching said first digitized target radiograph by applying said statistical template to said first digitized target radiograph until a first optimal fit of said statistical template to at least one region of said first digitized target radiograph is found, wherein searching said first digitized target radiograph for said first optimal fit comprises adjusting the position, orientation, shape and texture of said statistical template until the differences between said statistical template and said at least one region of said first digitized target radiograph are minimized;
• e. during said runtime phase repeating steps (b), (c) and (d) for a second digitized target two-dimensional radiograph of said specified joint obtained at a second time to find a second optimal fit of said statistical template to at least one region of said second digitized target radiograph;
• and
• f. during said runtime phase deriving at least one measure indicative of the progression of arthritic symptomatology of said specified joint by comparing said first optimal fit of said statistical template and said second optimal fit of said statistical template.