Fiducial Extraction¶
Extracts bright circular fiducials in the 3D volume
Input¶
16-bit unsigned 3D volume
Output¶
The input data is not modified. A list of the detected fiducials can be exported as annotations of the input data set.
Description¶
The algorithm detects and extracts bright fiducials from a 3D image set. Several modes of operation are supported:
In 2D mode the detection is run on the image currently shown in the axial view and 2D circles are detected, whereas in 3D mode the detection is run either only on the current slice or on the entire volume with the output being 3D spheres. Depending on the mode the center and/or the radius of the fiducial are fitted to the image data.
The following parameters are available:
- ROI: Size of the region of interest to analyze around a potential candidate. It must be larger than the expected diameter of the fiducial.
- Minimum contrast: Minimum intensity difference between the fiducial and its background after normalization of the slice intensities.
- Fixed diameter: When checked only fiducials of the specified diameter will be detected. Fiducials with different diameters will not be detected. Only the center of the fiducial will be optimized to the image data, the diameter will always stay at the specified value.
- Minimum coverage: Minimum “circularity” meaure in range [0;1].
- Score threshold: Minimum detection confidence in range [0;1].
- Global detection: Perform detection on entire volume (only in 3D mode).
- Only on slice: Run detection only on current axial slice (only in 3D mode).
When the fixed diameter checkbox is unchecked, the user must click the fiducial in the axial view. The radius and center will then be computed based on the image data. If the image data does not allow automatic extraction, the clicked point will be taken as the fiducial center and a warning printed in the log window.
The list view shows all detected fiducials. Clicking on an entry will center the MPR views on the selected fiducial.
The entries of the list can be manually reorded or automatically sorted by z-coordinate (Sort button). They can be exported to a text file or as annotations of the input data set.