Back to results
US 11,599,191 B2Granted
Automatic localization of a structure
AssigneeSiemens Healthcare GmbH
InventorsMartin Berger, Anton Nekovar
Life of this patent
Estimating the term…
Abstract
According to a computer-implemented method for the automatic localization of a structure, an individual image which maps an object with the structure is displayed on a display surface. A gaze movement of a user is captured by a gaze capture system and, on the basis thereof, a gaze capture signal is generated. By a computer unit, dependent upon the gaze capture signal, at least one image point of the individual image is identified, which at least partially maps the structure.
Claims14
A computer-implemented method for an automatic localization of a structure, wherein an individual image which maps an object with the structure is displayed on a display surface, the method comprising:
- displaying an individual image on the display surface;
- capturing a gaze movement of a user by a gaze capture system while the individual image is displayed on the display surface;
- generating a gaze capture signal based on the captured gaze movement;
- identifying, by a computer, dependent upon the gaze capture signal, at least one image point of the individual image, wherein the at least one image point at least partially maps the structure;
- displaying at least one additional individual image on the display surface, therein defining an image sequence comprising the individual image and the at least one additional individual image;
- repeating the capturing, the generating, and the identifying for each additional individual image to provide at least one further image point for each additional individual image displayed in the image sequence, wherein each further image point at least partially maps the structure;
- and
- segmenting the structure at least partially automatically by the computer based on the identified at least one image point of the individual image and the at least one further image point for each additional individual image.
5 more elements
A computer-implemented method for training a software module for machine learning, the method comprising:
- displaying, on a display surface, a training individual image which maps a training object with a training structure;
- capturing a gaze movement of a user by a gaze capture system;
- generating a training signal based on the captured gaze movement;
- identifying, by a training computer, at least one image point dependent on the generated training signal, wherein the at least one image point at least partially maps the training structure;
- displaying at least one additional training image on the display surface, therein defining an image sequence comprising the training individual image and the at least one additional training image;
- repeating the capturing, the generating, and the identifying for each additional training image to provide at least one further image point for each additional training image displayed in the image sequence, wherein each further image point at least partially maps the training structure;
- and
- segmenting, by the training computer, the training structure at least partially automatically based on the at least one image point and the at least one further image point for each additional training image;
- and
- training the software module based on a result of the segmentation by the training computer.
7 more elements
An apparatus for an automatic localization of a structure, the apparatus comprising:
- a computer;
- a display device having a display surface configured to display an image sequence comprising a plurality of individual images, wherein individual image of the plurality of individual images maps an object with the structure on the display surface;
- a gaze capture system configured to capture a gaze movement of a user of the apparatus and generate a gaze capture signal based on the captured gaze movement for each individual image of the plurality of individual images displayed in the image sequence;
- and
- a software module for machine learning,
- wherein the computer is configured, dependent upon a respective gaze capture signal, to identify at least one image point of the respective individual image of the plurality of individual images in the image sequence, wherein each at least one image point of the respective individual image at least partially maps the structure,
- wherein the display device is further configured to display at least one additional image on the display surface, therein defining an image sequence comprising the individual image and the at least one additional image,
- wherein the gaze capture system and computer are further configured to repeat, for each additional at least one image, the capture of the gaze movement, the generation of the gaze capture signal, and the identification of at least one further image point of the respective at least one additional image, wherein the at least one further image point at least partially maps the structure,
- wherein the computer is further configured, based on the identified at least one image point of the individual image and the at least one further image point for each additional individual image, to segment the structure at least partially automatically, and
- wherein the software module is configured to be trained based on a result of the segmentation by the computer.
7 more elements
Dependencies are inferred from the claim text.