US 12,249,067 B2
Dynamic multimodal segmentation selection and fusion
Tao Tan, Nuenen (NL); Hongxiang Yi, Excelsior, MN (US); Rakesh Mullick, Bangalore (IN); Lehel Mihály Ferenczi, Dunakeszi (HU); Gopal Biligeri Avinash, Concord, CA (US); Borbála Deák-Karancsi, Budapest (HU); Balázs Péter Cziria, Budapest (HU); and Laszlo Rusko, Budapest (HU)
Assigned to GE Precision Healthcare LLC, Waukesha, WI (US)
Filed by GE Precision Healthcare LLC, Wauwatosa, WI (US)
Filed on May 24, 2022, as Appl. No. 17/664,702.
Prior Publication US 2023/0386022 A1, Nov. 30, 2023
Int. Cl. G06T 7/00 (2017.01); A61B 8/08 (2006.01); G06T 7/149 (2017.01); G06T 7/174 (2017.01)
CPC G06T 7/0012 (2013.01) [A61B 8/0883 (2013.01); G06T 7/149 (2017.01); G06T 7/174 (2017.01); G06T 2207/10136 (2013.01); G06T 2207/20061 (2013.01); G06T 2207/20124 (2013.01); G06T 2207/30048 (2013.01)] 25 Claims
OG exemplary drawing
 
1. A system, comprising:
a memory that stores computer executable components; and
a processor that executes the computer executable components stored in the memory, wherein the computer executable components comprise:
a reception component that receives a segmentation dataset comprising a combination of different image segmentations of an anatomical object of interest respectively segmented via different segmentation models from different medical images captured of the anatomical object, wherein the different medical images and the different image segmentations vary with respect to capture modality;
a ranking component that employs a dynamic ranking protocol as opposed to a static ranking protocol to determine ranking scores for the different image segmentations that control relative contributions of the different image segmentations in association with combining the different image segmentations into a fused segmentation for the anatomical object, wherein the dynamic ranking protocol comprises determining the ranking scores based a plurality of parameters, comprising:
confidence measures associated with the different image segmentations that reflect measures of confidence of the different segmentation models with respect to the different image segmentations, and
a type of the anatomical object; and
a fusion component that combines the different image segmentations based on the ranking scores to generate the fused image segmentation.