Method and system for binary and quasi-binary atlas-based auto-contouring of volume sets in medical images
View Patent ↗The present invention is a set of algorithms or calculations that may be applied to efficiently use atlas-based segmentation using QBinary (binary and/or quasi-binary) representations of images and/or their ROIs. A QBinary representation of an image or a part of an image (for example, such as a ROD may be a binary image (for example, such as a code with 1 bit) and or a quasi-binary image (for example, such as a code with more than 1 and less than 8 bits) that serves to represent that image and/or ROI in an atlas. Two-dimensional QBinary codes may be used to both construct QBinary atlases and QBinary-controlled atlases with or without QB-Codes registration. Additionally, the present invention may offer a set of algorithms to sequentially auto-contour (or auto-segment) individual slices of a volume set of digital images without or without QB-Codes registration, provided one slice is semi-automatically contoured by a human operator.
1. A method of QB-Code generation to efficiently use atlas-based segmentation using any of binary or quasi-binary representations of image regions of interest, or a combination of binary and quasi-binary representation of image regions of interest, comprising:
a. identifying a region of interest (ROI) in an image;
b. generating a QB-code for the ROI by searching to identify an applicable QB-Code in a binary atlas or a quasi-binary atlas; and
c. determining a gold standard contour, wherein said gold standard contour is attached to the generated QB-Code.
2. The method of claim 1 , wherein generating the QB-Code comprises one or more of:
a. a pure binary atlas segmentation;
b. a quasi-binary atlas segmentation;
c. a binary controlled atlas segmentation; and
d. a quasi-binary controlled atlas segmentation.
3. The method of claim 2 , further comprising atlas segmentation with registration of the QB-Code.
4. The method of claim 1 , further comprising utilizing the QB-Code and applying sequential edit propagation to achieve fast slice-by-slice auto-contouring of volume image data.
5. The method of claim 4 , wherein achieving fast slice-by-slice auto-contouring of volume image data comprises:
a. generating or receiving a slice of an ROI as a contoured ROI; and
b. repeating the following steps until all images are processed:
i. editing the contoured ROI and inserting the edited contoured ROI and the associated QB-code into the binary and/or quasi-binary atlas; and
ii. automatically segmenting a next slice by propagating the contoured ROI and an associated gold standard contour.
6. The method of claim 5 , further comprising one or more of:
a. inserting the edited contoured ROI into a binary atlas;
b. inserting the edited contoured ROI into a quasi-binary atlas;
c. inserting the edited contoured ROI into a binary controlled atlas; and
d. inserting the edited contoured ROI into a quasi-binary controlled atlas.
7. The method of claim 1 , wherein generating the QB-Code comprises one or more of:
a. thresholding the ROI;
b. thresholding to transform the ROI;
c. thresholding to segment the ROI;
d. generating a local binary pattern involving the ROI; and
e. generating a customized binary code involving the ROI.
8. The method of claim 1 , wherein generating the QB-Code comprises one or more of:
a. (multi-) thresholding the ROI;
b. (multi-) thresholding to transform the ROI;
c. (multi-) thresholding to segment the ROI;
d. generating a local binary pattern involving the ROI; and
e. generating a customized binary code involving the ROI.
9. The method of claim 1 , further comprising:
a. creating the binary atlas or quasi-binary atlas;
b. grouping atlas entries;
c. generating one or more representative samples for each group of atlas entries; and
d. repeating the following steps until a QB-Code is generated for each of the one or more representative samples for each group:
i. generating the QB-Code for each of the one or more representative samples for each group; and
ii. querying whether the QB-Code is generated for each group.
10. The method of claim 1 , further comprising creating the binary atlas or quasi-binary atlas to generate a pure QBinary atlas, or a QBinary controlled atlas.
11. The method of claim 1 , further comprising utilizing an image registration algorithm for QB-Code generation.
12. The method of claim 1 , further comprising registering the QB-Code.
13. A system of QB-Code generation to efficiently implement atlas-based segmentation using any of binary or quasi-binary representations of image regions of interest, or a combination of binary and quasi-binary representation of image regions of interest, comprising:
a QB-Code Generator configured to (a) identify a region of interest (ROI) in an image and (b) generate a QB-code for the ROI by searching to identify an applicable QB-Code in a binary atlas or a quasi-binary atlas;
wherein a gold standard contour attached to the generated QB-Code is determined.
14. The system of claim 13 , wherein generating the QB-Code comprises one or more of:
a. a pure binary atlas segmentation;
b. a quasi-binary atlas segmentation;
c. a binary controlled atlas segmentation; and
d. a quasi-binary controlled atlas segmentation.
15. The system of claim 13 , wherein generating the QB-Code comprises one or more of:
a. thresholding the ROI;
b. thresholding to transform the ROI;
c. thresholding to segment the ROI;
d. generating a local binary pattern involving the ROI; and
e. generating a customized binary code involving the ROI.
16. The system of claim 13 , wherein generating the QB-Code comprises one or more of:
a. (multi-) thresholding the ROI;
b. (multi-) thresholding to transform the ROI;
c. (multi-) thresholding to segment the ROI;
d. generating a local binary pattern involving the ROI; and
e. generating a customized binary code involving the ROI.