Segmentation of the airway tree using hyperpolarized noble gases and diffusion weighted magnetic resonance imaging

1. A method for segmenting a diagnostic magnetic resonance image of the lung, the steps comprising:

a) filling the lung with a NMR responsive gas;

b) acquiring diffusion weighted image (DWI) data from the lung with an MRI system operating at the larmor frequency of the NMR responsive gas;

c) reconstructing an apparent diffusion coefficient (ADC) image of the lungs from the acquired DWI data;

d) producing a segmentation image from the ADC image in which voxels therein having an ADC value above a selected value are identified as being in the airway tree; and

e) using the segmentation image to segment the diagnostic magnetic resonance image.

2. The method as recited in claim 1 in which the NMR responsive gas is a noble gas.

3. The method as recited in claim 2 in which the noble gas is helium.

4. The method as recited in claim 1 which further includes:

f) registering the segmentation image with the diagnostic magnetic resonance image before performing step e).

5. The method as recited in claim 1 in which step d) includes growing regions in the segmentation image having corresponding voxels in the ADC image with ADC values above the selected value.

6. The method as recited in claim 1 in which the diffusion weighted image data is acquired in step b) using a two-dimensional pulse sequence that is repeated to acquire data for a series of two-dimensional slices.

7. The method as recited in claim 6 in which step c) includes reconstructing a series of two-dimensional slice images, and calculating the apparent diffusion coefficient (ADC) at each voxel in each of the series of slice images.

8. The method as recited in claim 7 in which step d) includes;

d)i) locating a voxel in one of the two-dimensional ADC images having a value that is above a first ADC threshold value;

d)ii) growing a region in a corresponding segmentation image that indicates airway tree by locating contiguous voxels in the two-dimensional ADC image that have values above a second ADC threshold value.

9. The method as recited in claim 8 in which the NMR responsive gas is helium and the first ADC threshold value is 0.4 cm 2 /s and the second ADC threshold value is 0.3 cm 2 /s.

10. The method as recited in claim 8 in which steps d)i) and d)ii) are repeated for each of the other two-dimensional ADC images having a voxel value above the first ADC threshold value.

11. The method as recited in claim 1 in which the segmentation image indicates at each of its voxels whether the voxel location contains airway or lung tissue.

12. The method as recited in claim 11 in which step e) is performed by multiplying voxel values in the segmentation image by corresponding voxel values in the diagnostic magnetic resonance image.

13. The method as recited in claim 1 in which step b) includes:

b)i) acquiring a first image data set using a pulse sequence that includes a diffusion weighted gradient; and

b)ii) acquiring a second image data set using a pulse sequence with substantially no diffusion weighting gradient.

14. The method as recited in claim 13 in which step c) includes:

c)i) reconstructing a first image from the first image data set;

c)ii) reconstructing a second image from the second image data set; and

c)iii) calculating the ADC at each voxel according to the equation

ADC

=

ln

⁡

(

S

⁢

⁢

0

S

⁡

(

b

)

)

/

b

where:

So=corresponding voxel value in the second image;

S(b)=corresponding voxel value in the first image; and

b=a constant determined by the first moment of the diffusion weighting gradient.

15. The method as recited in claim 13 which includes:

reconstructing a ventilation image from the second image data set.

16. The method as recited in claim 15 which includes:

segmenting the ventilation image using the segmentation image.