US 12,493,972 B2
Systems and methods for segmenting 3D images comprising a downsampler, low-resolution module trained to infer a complete low-resolution segmentation and generate corresponding low-resolution feature maps from an input downsampled high-resolution 3D image and high-resolution module trained to infer a complete high-resolution segmentation from an input from the low-resolution module and 3D high-resolution image
Dante De Nigris, Montreal (CA); Gabriel Chartrand, Magog (CA); and Simon Ducharme, Montréal (CA)
Assigned to AFX MEDICAL INC., Montreal (CA)
Appl. No. 18/030,698
Filed by AFX MEDICAL INC., Montreal (CA)
PCT Filed Sep. 28, 2021, PCT No. PCT/CA2021/051349
§ 371(c)(1), (2) Date Apr. 6, 2023,
PCT Pub. No. WO2022/073100, PCT Pub. Date Apr. 14, 2022.
Claims priority of provisional application 63/088,717, filed on Oct. 7, 2020.
Prior Publication US 2023/0386067 A1, Nov. 30, 2023
Int. Cl. G06T 7/11 (2017.01); G01R 33/56 (2006.01); G06N 3/0464 (2023.01); G06T 3/40 (2006.01); G06T 7/00 (2017.01); G06T 7/62 (2017.01)
CPC G06T 7/62 (2017.01) [G01R 33/5608 (2013.01); G06T 3/40 (2013.01); G06T 7/0012 (2013.01); G06T 7/11 (2017.01); G06T 2207/10088 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30016 (2013.01)] 11 Claims
OG exemplary drawing
 
1. A system for segmenting 3D images, the system comprising:
a downsampler configured to downsample an input high-resolution 3D image to produce a low-resolution 3D image; and
a computer-implemented neural network module comprising:
a low-resolution module trained to infer a complete low-resolution segmentation from the low-resolution 3D image and to generate corresponding low-resolution feature maps; and
a high-resolution module trained to infer a complete high-resolution segmentation from the input high-resolution 3D image and the feature maps from the low-resolution module.