Method, system and computer readable medium for automatic segmentation of a medical image
A method for automatic segmentation of a medical image is provided. The method comprises registering a reference image associated with an object to the medical image, determining a transformation function on the basis of the registration, applying the transformation function to a probability map associated with the object; carrying out a probability thresholding on the transformed probability map by selecting a first area of the medical image in which the probability of the object is within a probability range, carrying out an intensity thresholding on the medical image by selecting a second area of the medical image in which the intensity is within an intensity range, selecting a common part of the first and second areas and carrying out on the common part a morphological opening resulting in separate sub-areas, selecting the largest sub-area as a seed, and segmenting the medical image on the basis of the seed.
1. A method for automatic segmentation of a medical image, the medical image comprising an object image to be segmented, the method comprising:
registering a reference image associated with the object to the medical image;
determining a transformation function on the basis of the registration;
applying the transformation function to a probability map associated with the object;
carrying out a probability thresholding on the transformed probability map by selecting a first area of the medical image in which the probability of the object is within a probability range;
carrying out an intensity thresholding on the medical image by selecting a second area of the medical image in which the intensity is within an intensity range;
selecting a common part of the first and second areas and carrying out on the common part a morphological opening resulting in separate sub-areas of the common part;
selecting the largest sub-area as a seed; and
segmenting the medical image on the basis of the selected seed.
2. The method according to claim 1 , further comprising applying the transformation function to at least one further probability map associated with a further object;
carrying out a probability thresholding on the transformed further probability map; and
deselecting areas from the first area in which the probability of the further object is within a further probability range.
3. The method according to claim 1 , further comprising applying a morphological dilatation on bone image parts before carrying out an intensity thresholding on the medical image.
4. The method according to claim 1 , further comprising validating the segmentation on the basis of at least one thresholding criterion selected from the group consisting of volume, enclosing block volume, ratio of volume/enclosing block volume, enclosing sphere volume, ratio of volume/enclosing sphere volume, enclosing sphere volumes of slices, and axial distortion.
5. The method according to claim 1 , wherein segmenting the medical image on the basis of the selected seed comprises growing a region from the seed with a growing method.
6. A non-transitory computer readable medium comprising a computer readable program for automatic segmentation of a medical image, wherein the computer readable program when executed on a computer causes the computer to perform the steps of the method according to claim 1 .
7. A system for automatic segmentation of a medical image, the medical image comprising an object image to be segmented, the system comprising:
a computer comprising a computer memory;
a registering unit stored in the computer memory configured to register a reference image associated with the object to the medical image, and to determine a transformation function on the basis of the registration;
a transforming unit stored in the computer memory configured to apply the transformation function to a probability map associated with the object;
a probability thresholding unit stored in the computer memory configured to carry out probability thresholding on the transformed probability map by selecting a first area of the medical image in which the probability of the object is within a probability range;
an intensity thresholding unit stored in the computer memory configured to carry out an intensity thresholding on the medical image by selecting a second area of the medical image in which the intensity is within an intensity range,
a morphological opening unit stored in the computer memory configured to carry out a morphological opening on a common part of the first and second areas, resulting in separate sub-areas of the common part,
a selecting unit stored in the computer memory configured to select the largest sub-area as a seed, and
a segmenting unit stored in the computer memory configured to segment the medical image on the basis of the selected seed.
8. The system according to claim 7 , further comprising a further registering unit stored in the computer memory configured to apply the transformation function to at least one further probability map associated with a further object, to carry out a probability thresholding on the further probability map, and to deselect areas from the first area in which the probability of the further object is within a further probability range.
9. The system according to claim 7 , further comprising a morphological dilatation unit stored in the computer memory configured to apply a morphological dilatation on bone image parts.
10. The system according to claim 7 , further comprising a validating unit stored in the computer memory configured to validate the segmentation on the basis of at least one thresholding criterion selected from the group consisting of volume, enclosing block volume, ratio of volume/enclosing block volume, enclosing sphere volume, ratio of volume/enclosing sphere volume, enclosing sphere volumes of slices, and axial distortion.
11. The system according to claim 7 , wherein the segmenting unit is further configured to grow a region from the seed with a growing method.