Method for rapid localization of a scintillation event in a gamma camera using a maximum likelihood estimator

Claim 2

In a process for localizing a scintillation event in a gamma camera having a plurality of photomultipliers forming a camera surface, each photomultiplier generating an output signal in response to said scintillation event, said process including the steps of generating a plurality of comparative signal sets from output signals of said photomultipliers corresponding to respective scintillation events of known locations, forming a location-dependent probability function based on a comparison of outputs of said photomultipliers for a scintillation event of unknown location with said comparative signal sets, said probability function having a maximum, and defining the location of said scintillation event of unknown location as the location corresponding to said maximum of said location-dependent probability function, the improvement comprising: initially undertaking a pre-localization by defining a part of said camera surface, in which a high probability exists that the location of said scintillation event of unknown location lies, as a partial camera surface; and limiting the forming of said location-dependent probability function to locations within said partial camera surface.

Claim 3

The improvement of claim 1 wherein each of said photomultipliers has an observation surface uniquely associated therewith within said camera surface, and wherein the step of defining a part of said camera surface as a partial camera surface is further defined by defining the observation surface of a photomultiplier having the largest output signal as said partial camera surface.

Claim 4

The improvement of claim 1 wherein each of said photomultipliers has an observation surface uniquely associated therewith within said camera surface, each observation surface having a center and a diameter, and wherein the step of defining a part of said camera surface as said partial camera surface is further defined by defining a square as said partial camera surface, said square having a center coinciding with the center of the observation surface of a photomultiplier having the largest output signal, and said square having a side length which is substantially twice said diameter.

Claim 5

The improvement of claim 1 wherein the step of initially undertaking a prelocalization is further defined by the steps of: dividing said partial camera surface into a first raster having coarse divisions and having a plurality of first raster points, one of said first raster points having a location at which said location-dependent probability function has a greatest value within said first raster; defining a further partial camera surface, centered on said one of said first raster points and being smaller than said partial camera surface; dividing said further partial camera surface into a second raster having divisions finer than the divisions of said first raster and having a plurality of further raster points; and limiting the forming of said location-dependent probability function to locations within said further partial camera surface.

Claim 6

The improvement of claim 4 wherein none of said points within said first raster coincide with any of said points within said second raster.

Claim 7

The improvement of claim 4 wherein said scintillation events of known location used to generate said plurality of comparative signal sets are spaced from one another by a pixel distance, and said improvement comprising the additional step of: successively defining a plurality of further partial camera surfaces, each being smaller than an immediately preceding further partial camera surface and each being divided into a raster of increasingly finer divisions, until a distance between raster points in one of said plurality of further partial camera surfaces equals said pixel distance.

Claim 8

The improvement of claim 6 wherein none of said raster points in any of said partial camera surface and said further partial camera surfaces coincide.

Claim 9

The improvement of claim 6 wherein said raster points of said first raster are spaced from one another at distances equal to eight times said pixel distance and wherein said raster points of said second raster are spaced from one another at distances equal to six times said pixel distance.

Claim 10

The improvement of claim 1 wherein the steps of generating said plurality of comparative signal sets and forming a location-dependent probability function are respectively further defined by: generating said plurality of comparative signal sets from the standard deviations of said output signals of said photomultipliers corresponding to said respective scintillation events of known locations; and forming said location-dependent probability function as a Gaussian distribution function of said standard deviations.

Claim 11

In a process for localizing a scintillation event in a gamma camera having a plurality of photomultipliers forming a camera surface, each photomultiplier generating an output signal in response to said scintillation event, said process including the steps of generating a plurality of comparative signal sets from output signals of said photomultipliers corresponding to respective scintillation events of known locations, spaced from one another at a pixel distance, forming a location-dependent probability function based on a comparison of outputs of said photomultipliers for a scintillation event of unknown location with said comparative signal sets, said probability function having a maximum, and defining the location of said scintillation event of unknown location as the location corresponding to said maximum of said location-dependent probability function, the improvement comprising: dividing said camera surface into a plurality of overlapping partial camera surfaces; identifying a point within each partial camera surface at which said location-dependent probability function has a maximum within that partial camera surface; identifying one of said partial camera surfaces in which said location probability function has a largest value; dividing said partial camera surface in which said location-dependent probability function has a largest value into overlapping further partial camera surfaces and identifying a further partial camera surface in which said location-dependent probability function has a largest value; successively dividing said further partial camera surface in which said location-dependent probability function has a largest value into a plurality of further, increasingly smaller, overlapping partial camera surfaces and successsively identifying one of said further partial camera surfaces in which said location-dependent probability function has a largest value until a further partial surface is reached having a size equal to said pixel distance; and defining the location of said scintillation event of unknown location as being within said further partial camera surface having said size equal to said pixel distance.

Claim 12

In a process for localizing a scintillation event in a gamma camera having a plurality of photomultipliers forming a camera surface, each photomultiplier generating an output signal in response to said scintillation event, said process including the steps of generating a plurality of comparative signal sets from output signals of said photomultipliers corresponding to respective scintillation events of known locations, spaced from one another at a pixel distance, forming a location-dependent probability function based on a comparison of outputs of said photomultipliers for a scintillation event of unknown location with said comparative signal sets, said probability function having a maximum, and defining the location of said scintillation event of unknown location as the location corresponding to said maximum of said location-dependent probability function, the improvement comprising: dividing said camera surface into a plurality of overlapping partial camera surfaces; identifying a point within each partial camera surface at which said location-dependent probability function has a maximum within that partial camera surface; identifying one of said partial camera surfaces in which said location probability function has a largest value; dividing said partial camera surface in which said location-dependent probability function has a largest value into overlapping further partial camera surfaces and identifying a further partial camera surface in which said location-dependent probability function has a largest value; successively dividing said further partial camera surface in which said location-dependent probability function has a largest value into a plurality of further, increasingly smaller, overlapping partial camera surfaces and successsively identifying one of said further partial camera surfaces in which said location-dependent probability function has a largest value until a further partial surface is reached having a size equal to said pixel distance; and defining the location of said scintillation event of unknown location as being within said further partial camera surface having said size equal to said pixel distance.

Claim 13

The improvement of claim 10 wherein each of said partial camera surfaces and said further partial camera surfaces are rectangular, and wherein the steps of defining said overlapping further partial camera surfaces are further defined by dividing each immediately preceding partial camera surface or further partial camera surface into four overlapping further partial camera surfaces and thereby defining four points respectively corresponding to the intersections of the sides of said four overlapping partial camera surfaces; and determining said location-dependent probability function at each of said intersection points.

Claim 14

The improvement of claim 11 wherein the step of dividing said partial camera surface of said further partial camera surfaces into four overlapping partial camera surfaces is further defined by dividing said partial camera surface or said further partial camera surfaces into four overlapping partial camera surfaces by division according to the golden section.

Claim 15

The improvement of claim 10 wherein the steps of generating said plurality of comparative signal sets and forming a location-dependent probability function are respectively further defined by: generating said plurality of comparative signal sets from the standard deviations of said output signals of said photomultipliers corresponding to said respective scintillation events of known locations spaced from one another at said pixel distance; and forming said location-dependent probability function as a Gaussian distribution function of said standard deviations.