Back to results
US 10,420,626 B2Granted
Fiducial markers, systems, and methods of registration
AssigneesCanon U.S.A., Inc., The Brigham and Women's Hospital, Inc.
InventorsJunichi Tokuda, Nobuhiko Hata, Takahisa Kato, Brian Ninni, Laurent Chauvin
Life of this patent
Estimating the term…
Abstract
Exemplary methods, apparatus, and systems are provided for automated detection and registration of medical images using fiducial markers and processing algorithms. By either clustering fiducial markers having a different number, size, shape, configuration, or material property or by using fiducial markers arranged in a ring shape, wherein their arrangement is asymmetric, and then applying a feature extraction to extract the fiducial marker objects, and registering the fiducial marker objects with a model of the fiducial frame, automatic registration can be achieved.
Claims18
A method of automatic registration, comprising:
- obtaining image data of a medical device, the medical device comprising a plurality of fiducial markers arranged as a fiducial frame on the medical device, said fiducial frame comprising:
- (a) at least two clusters of fiducial markers, each cluster comprising a different number, size, shape, configuration, or material property of fiducial marker(s), or
- (b) at least three fiducial markers arranged in a ring shape, wherein their arrangement is asymmetric;
- detecting fiducial marker objects within the image data by:
- applying a feature enhancement to the image data to enhance the fiducial marker objects, the feature enhancement including shape data of the fiducial markers,
- applying a feature extraction based on the shape data of the fiducial markers to extract the fiducial marker objects, and
- for image data of the medical device comprising the at least two clusters of fiducial markers, applying a clustering algorithm to define clusters of the fiducial marker objects;
- defining a representative point for each extracted fiducial marker object or for each cluster of fiducial marker objects;
- and
- registering the fiducial marker objects with a model of the fiducial frame, the registering step comprising:
- (a) for image data of the medical device comprising the at least two clusters of fiducial markers,
- (i) calculating average positions of fiducial markers for each cluster in the model of the fiducial frame,
- (ii) point to point matching of the average positions of the fiducial markers for each cluster with the model of the fiducial frame, and then
- (iii) matching individual fiducial marker objects using an iterative closest point algorithm, or
- (b) for image data of the at least three fiducial markers arranged in the ring shape, estimating a circle to which the fiducial marker objects are aligned and matching a model circle to the estimated circle, the model circle based on model markers included in the model of the fiducial frame, and rotating the ring shape until the representative points of the fiducial marker objects match the model of the fiducial frame.
12 more elements
A system comprising:
- a medical device comprising a fiducial frame, wherein the fiducial frame comprises:
- (a) at least two clusters of fiducial markers, each cluster comprising a different number, size, shape, configuration, or material property of fiducial marker(s), or
- (b) at least three fiducial markers arranged in a ring shape, wherein their arrangement is asymmetric;
- and
- a non-transitory computer-readable storage medium storing a computer-executable program for performing a method of fiducial registration, the method comprising:
- obtaining image data of a medical device;
- detecting fiducial marker objects within the image data by:
- applying a feature enhancement to the image data to enhance the fiducial marker objects, the feature enhancement including shape data of the fiducial markers,
- applying a feature extraction based on the shape data of the fiducial markers to extract the fiducial marker objects, and
- for image data of the medical device comprising the at least two clusters of fiducial markers, applying a clustering algorithm to define clusters of the fiducial marker objects;
- defining a representative point for each extracted fiducial marker object or for each cluster of fiducial marker objects;
- and
- registering the fiducial marker objects with a model of the fiducial frame, the registering step comprising:
- (a) for image data of the medical device comprising the at least two clusters of fiducial markers, (i) calculating average positions of fiducial markers for each cluster in the model of the fiducial frame, (ii) point to point matching of the average positions of the fiducial markers for each cluster with the model of the fiducial frame, and then (iii) matching individual fiducial marker objects using an iterative closest point algorithm, or
- (b) for image data of the at least three fiducial markers arranged in the ring shape, estimating a circle to which the fiducial marker objects are aligned and matching a model circle to the estimated circle, the model circle based on model markers included in the model of the fiducial frame, and rotating the ring shape until the representative points of the fiducial marker objects match the model of the fiducial frame.
12 more elements
A method of automatic registration, comprising:
- obtaining image data of a medical device, the medical device comprising a plurality of spherical fiducial markers arranged as a fiducial frame on the medical device, said fiducial frame comprising at least three fiducial markers arranged in a ring shape, wherein their arrangement is asymmetric;
- detecting fiducial marker objects within the image data by:
- applying a feature enhancement to the image data to enhance the fiducial marker objects, the feature enhancement including shape data of the spherical fiducial markers, and
- applying a feature extraction based on the shape data of the spherical fiducial markers to extract the fiducial marker objects;
- defining a representative point for each extracted fiducial marker object;
- and
- registering the fiducial marker objects with a model of the fiducial frame, the registering including estimating a circle to which the fiducial marker objects are aligned and matching a model circle to the estimated circle, the model circle based on model markers included in the model of the fiducial frame, the registering step comprising rotating the ring shape until the representative points of the fiducial marker objects match the model of the fiducial frame.
4 more elements
Dependencies are inferred from the claim text.