Spatial intensity correction for RF shading non-uniformities in MRI

1. A method for magnetic resonance imaging (MRI) comprising use of an MRI gantry and associated computers and hardware to:

acquire MRI map prescan data from a predetermined imaged patient volume;

decompose said acquired map prescan data to produce a transmit RF field inhomogeneity map and a receive RF field inhomogeneity map for said imaged patient volume based on a three-dimensional geometrical model of said inhomogeneity maps; and

use at least one of said transmit RF field inhomogeneity map and said receive RF field inhomogeneity map to generate intensity-corrected target MRI diagnostic scan image data representing said imaged patient volume,

wherein said geometrical model includes a plurality of variable geometric parameter values selected in an optimization process to fit in said predetermined imaged patient volume, and

wherein said geometrical model includes a plurality of spatially dependent terms, each of which each include an amplitude value that is independently fitted to said acquired map prescan data.

2. A method for magnetic resonance imaging (MRI) comprising use of an MRI gantry and associated computers and hardware to:

acquire MRI map prescan data from a predetermined imaged patient volume;

decompose said acquired map prescan data to produce a transmit RF field inhomogeneity map and a receive RF field inhomogeneity map for said imaged patient volume based on a three-dimensional geometrical model of said inhomogeneity maps; and

use at least one of said transmit RF field inhomogeneity map and said receive RF field inhomogeneity map to generate intensity-corrected target MRI diagnostic scan image data representing said imaged patient volume,

wherein said geometrical model includes a plurality of variable geometric parameter values selected in an optimization process to fit in said predetermined imaged patient volume, and

wherein said model is guided by image feature detection to locate at least one predetermined model feature within the image.

3. The method of claim 2 , wherein said image feature detection includes detecting and utilizing increased patient width at the transition from head-neck anatomy to shoulders anatomy as a landmark from which model feature positions are determined.

4. A method for magnetic resonance imaging (MRI) comprising use of an MRI gantry and associated computers and hardware to:

acquire MRI map prescan data from a predetermined imaged patient volume;

decompose said acquired map prescan data to produce a transmit RF field inhomogeneity map and a receive RF field inhomogeneity map for said imaged patient volume based on a three-dimensional geometrical model of said inhomogeneity maps; and

use at least one of said transmit RF field inhomogeneity map and said receive RF field inhomogeneity map to generate intensity-corrected target MRI diagnostic scan image data representing said imaged patient volume,

wherein spurious structure in background image noise and/or the overall level of background noise is reduced by regularizing MAP prescan data or a ratio of MAP prescan data as a function of noise.

5. An MRI system comprising:

means for acquiring MRI MAP prescan data from a predetermined imaged patient volume;

means for decomposing said acquired MAP prescan data to produce a transmit RF field inhomogeneity map and a receive RF field inhomogeneity map for said imaged patient volume based on a three-dimensional geometrical model of said inhomogeneity maps; and

means for using at least one of said transmit RF field inhomogeneity map and said receive RF field inhomogeneity map to generate intensity-corrected target MRI diagnostic scan image data representing said imaged patient volume,

wherein said three-dimensional geometrical model is defined by a closed form mathematical function that has smoothly varying spatial dependence,

wherein said geometrical model includes a plurality of variable geometric parameter values selected in an optimization process to fit in said predetermined imaged patient volume, and

wherein said geometrical model includes a plurality of spatially dependent terms which each include an amplitude value that is independently fitted to said acquired map prescan data.

6. An MRI system comprising:

means for acquiring MRI MAP prescan data from a predetermined imaged patient volume;

means for decomposing said acquired MAP prescan data to produce a transmit RF field inhomogeneity map and a receive RF field inhomogeneity map for said imaged patient volume based on a three-dimensional geometrical model of said inhomogeneity maps;

means for using at least one of said transmit RF field inhomogeneity map and said receive RF field inhomogeneity map to generate intensity-corrected target MRI diagnostic scan image data representing said imaged patient volume,

wherein said three-dimensional geometrical model is defined by a closed form mathematical function that has smoothly varying spatial dependence,

wherein said geometrical model includes a plurality of variable geometric parameter values selected in an optimization process to fit in said predetermined imaged patient volume, and

wherein said model is guided by image feature detection to locate at least one predetermined model feature within the image.

7. The MRI system of claim 6 , wherein said image feature detection includes detecting and utilizing increased patient width at the transition from head-neck anatomy to shoulders anatomy as a landmark from which model feature positions are determined.

8. An MRI system comprising:

means for acquiring MRI MAP prescan data from a predetermined imaged patient volume;

means for decomposing said acquired MAP prescan data to produce a transmit RF field inhomogeneity map and a receive RF field inhomogeneity map for said imaged patient volume based on a three-dimensional geometrical model of said inhomogeneity maps; and

means for using at least one of said transmit RF field inhomogeneity map and said receive RF field inhomogeneity map to generate intensity-corrected target MRI diagnostic scan image data representing said imaged patient volume,

wherein spurious structure in background image noise and/or the overall level of background noise is reduced by regularizing map prescan data or a ratio of map prescan data as a function of noise.