公开(公告)号：WO2019175870A1

公开(公告)日：2019-09-19

申请号：PCT/IL2019/050274

申请日：2019-03-12

Applicant: PERSIMIO LTD.

Inventor： BOOK, Gilad ,  YOSIBASH, Zohar ,  TRABELSI, Nir

IPC: G06T7/136 ,  G06T7/174 ,  G06T7/00

Abstract: A method for automated segmentation of a bone in an image dataset of a region of a human body, wherein said region comprises at least two adjacent bones, and wherein one of the bones comprises a shaft region and a head region, the method comprising identifying in the image dataset voxels belonging to a contiguous volume comprising both bones based on a density threshold; extracting a subset comprising the head region of the first bone, based on a boundary image slice; and applying a trained machine learning classifier to said subset, to generate a classification of each of voxel as belonging to one of the bones or a none-bone tissue.