US 12,412,280 B2
Medical image segmentation and atlas image selection
Matthias Lenga, Mainz (DE); Tobias Wissel, Lubeck (DE); and Rolf Juergen Weese, Norderstedt (DE)
Assigned to KONINKLIJKE PHILIPS N.V., Eindhoven (NL)
Appl. No. 17/780,974
Filed by KONINKLIJKE PHILIPS N.V., Eindhoven (NL)
PCT Filed Dec. 7, 2020, PCT No. PCT/EP2020/084860
§ 371(c)(1), (2) Date May 28, 2022,
PCT Pub. No. WO2021/116011, PCT Pub. Date Jun. 17, 2021.
Claims priority of application No. 19214693 (EP), filed on Dec. 10, 2019.
Prior Publication US 2023/0005158 A1, Jan. 5, 2023
Int. Cl. G06T 7/168 (2017.01); G06T 7/37 (2017.01)
CPC G06T 7/168 (2017.01) [G06T 7/37 (2017.01); G06T 2207/10081 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30056 (2013.01); G06T 2207/30084 (2013.01)] 15 Claims
OG exemplary drawing
 
1. A method for medical image segmentation, the method comprising:
obtaining a segmentation function, the segmentation function being configured to receive a medical input image and to produce an image segmentation,
obtaining multiple test images and associated test image segmentations,
determining a segmentation quality for the multiple test images by comparing the associated test image segmentations with an image generated by the segmentation function,
selecting a plurality of the multiple test images having the segmentation quality over a threshold as multiple atlas images,
obtaining the multiple atlas images,
receiving a medical image,
registering the received image to the multiple atlas images thus obtaining multiple registered images and multiple corresponding registering transformations configured to register the received image on the multiple atlas images,
applying the segmentation function to the multiple registered images thus obtaining multiple registered image segmentations,
applying inverses of the multiple registering transformations to the multiple image segmentations thus obtaining multiple image segmentations, and
determining an output segmentation from the multiple image segmentations.