Spatial calibration

[eegkno]

See the section Phantom registration in: http://perk-software.cs.queensu.ca/plus/doc/nightly/user/ApplicationfCal.html

Simulation steps:

1. Select the Device set named "Saved dataset for both video and tracking (fCal)" from the drop-down box and click Connect
2. Switch to Freehand calibration toolbox by clicking on its tab on the left side of the application window The video source shows up in the main canvas (By clicking the Edit segmentation parameters button - the "edit-like" button in the line of Segmentation parameters - a dialog appears in which the segmentation parameters can be changed)
3. Click Start button to start calibration The calibration begins, the segmented points appear as green dots on the image. The green bar on the bottom shows the progress. When it finishes, the result matrix and the errors are displayed in the toolbox.
4. Show devices checkbox becomes enabled. Clicking on it changes display mode so the tools are displayed in the main canvas instead of the image
5. The results can be saved by clicking on Save results button as a new XML file

[eegkno]

Things to check when performing spatial calibration: See: http://perk-software.cs.queensu.ca/plus/doc/nightly/user/ApplicationfCal.html

Notes:

• Try to use a water tank large enough so that the bottom (or reflection from the bottom or sides) is not visible in the image. It is better to make the wire intersections appear as small, barely visible gray points than large, very bright spots for two reasons.
• Wire intersection positions cannot be determined accurately if it appears as a large bright blob. 2. If the image is too bright then the calibration algorithm may detect too many point candidates and the application may slow down or may appear to hang for minutes.

Optimize image acquisition to avoid incorrect fiducial segmentations: Fiducial line intersection points should appear as small circular-shaped blobs and the rest of the image should be almost black.

• Decrease the dynamic range to improve contrast.
• Decrease gain and increase the threshold to make the background as dark as possible and to make the fiducial line intersections as small as possible, while being still detectable by the segmentation algorithm.
• Reduce reflections by slightly changing the transducer angle or putting sound absorbing material in the water container (such as rubber or sponge).
• Keep the transducer's axial (direction of sound travel) axis orthogonal to the wires for optimal reflection from the wires.
• If the intersection blobs are not circular then set the focal point distance(s) of the transducer to match the depth of the fiducial wires. If two or more candidate fiducials (red points) appear at a wire position in the image then try to increase the MorphologicalOpeningBarSizeMm value (to something between 2.0mm - 10.0mm).
• Probe orientation: make sure that the probe orientation is correct (for fCal_2.x_Wiring_1.x the marked side of the probe shall be close to the A1 point). The wire numbers are shown in fCal during spatial calibration and in the segmentation, parameter setting dialog box, which can help in verification of the correct orientation. To verify the probe orientation: touch a w1 wire with your finger => your finger should appear near w1 wire on the US image. If your finger appears at a different wire (and you are sure that the image orientation is set correctly) then you do not keep the marked side of the probe on the w1 wire's side - in this case simply rotate the probe by 180 deg (and keep it in this orientation while collecting the calibration data). Try moving the sensor closer to the transducer tip. This reduces the lever arm effect and may reduce the calibration error.
• Make sure that the wires appear in the image during calibration where you want to achieve good accuracy (e.g., if you need good accuracy at 3-5cm distance from the transducer then have your calibration wires at 2, 4, and 6 cm from the transducer).

[eegkno]

The wiring should be made with elastic bands to avoid loose wires.

[eegkno]

In the segmentation parameters:

NOTE: This was the main step to perform the calibration correctly

• If green dots do not appear, then cancel the calibration by clicking on Cancel button, and bring up the Segmentation parameter dialog by pressing the edit icon (next to the Segmentation parameters import button):
• Press Freeze button if the image shows all the wires (it seems suitable for segmentation) Check if initial spacing is correct. For giving a hint on image spacing to fCal, select the checkbox next to spacing. It will give you two rulers, a green (G) and a blue (B). You can manually enter two known distances that appear on the image, one horizontal and one vertical. Then set the visual rulers on the image to represent these distances. You can do this by dragging the end points of the ruler.
• It's usually a good idea to unfreeze the image at this point and play with the "Gain" setting on your ultrasound machine to find the optimal value when dots (wires) on the image are all visible but not too large. If wires are still not recognized, you can freeze the image again, and change the other segmentation parameters. The first goal is to find all the candidates (red dots). If the wires are detected as candidates (they are red), the next step is the pattern recognition (make them green).
• When green dots appear on the wires in the parameter editor, then unfreeze, return to the Freehand Calibration tab and see if the segmentation is good enough on the other images.
• If the error is too big, maybe the orientation of the probe is not correct, rotate it 180 degrees.