A computer-implemented method and image processing device for marking objects on 3d images

An image processing device comprising:

• a processor configured to:
• obtain a 3D image including at least one object to be marked, wherein the 3D image may be represented by a plurality of 2D slices;
• display at least one slice of the 3D image to a user;
• enable the user to select a position of the 3D geometric form on the 2D slice so that the 3D geometric form at least partially overlaps with the object to be marked;
• in response to the user selection automatically define a 3D region of interest within the 3D image which comprises the 3D geometric form at the selected position and clipping the image data from the 3D region of interest from the 3D image;
• apply a trained algorithm to the image data from the 3D region of interest, wherein the output of the trained algorithm is a 3D volume within the 3D region of interest which is predicted to cover the object, wherein the enabling, defining, and applying are repeated at least once such that at each repetition an additional 3D volume is generated that depends on additional user input, and
• wherein each additional 3D volume is joined together with the currently existing 3D volume resulting in a larger existing 3D volume;
• output the 2D slice together with the marked 3D volume.

A non-transitory computer-readable storage medium having stored a computer program comprising instructions, which, when executed by a processor, cause the processor to:

• obtain a 3D image including at least one object to be marked, wherein the 3D image may be represented by a plurality of 2D slices;
• display at least one slice of the 3D image to a user;
• enable the user to select a position of the 3D geometric form on the 2D slice so that the 3D geometric form at least partially overlaps with the object to be marked;
• in response to the user selection automatically define a 3D region of interest within the 3D image which comprises the 3D geometric form at the selected position and clipping the image data from the 3D region of interest from the 3D image;
• apply a trained algorithm to the image data from the 3D region of interest,
• wherein the output of the trained algorithm is a 3D volume within the 3D region of interest which is predicted to cover the object, wherein the enabling, defining, and applying are repeated at least once such that at each repetition an additional 3D volume is generated that depends on additional user input, and
• wherein each additional 3D volume is joined together with the currently existing 3D volume resulting in a larger existing 3D volume;
• output the 2D slice together with the marked 3D volume.