Calibration Controls

This window contains controls which affect the line-fitting algorithms used in spatial and temporal calibration. It also allows the probe shape to be estimated: i.e. the region within the acquired rectangular image which actually contains ultrasound data. The probe shape can be used to update the current calibration if a new depth setting has been selected.

This window also displays the current values of all the spatial calibration transformations and scales:

Calibration Control Window

Controls affecting the line-fitting algorithm

Variance of Gaussian kernel
This controls the smoothing applied to the image in advance of the edge detection process.
Vertical analysis bands
This sets the number of vertical bands in the image that are used for the edge detection process. They are indicated in the calibration windows by vertical red lines drawn down the image.
Gradient threshold
This threshold governs the way the line detection algorithm detects edge elements. The algorithm scans down the image and picks the first local maximum which exceeds the threshold, or the biggest local maximum if none exceed the threshold.
Pixel linearity threshold
The random sample consensus algorithm requires a minimum proportion of the edge elements to be collinear, otherwise the line is not considered reliable. This parameter determines the vertical tolerance (in pixels) that is used when determining collinearity.
Ransac acceptance proportion
This is the proportion of edge elements that must be collinear for the line to be accepted.
Upside down image?
It is conventional with some probes (eg. endovaginal) to view the image upside down, so that the probe face is at the bottom of the image and objects distant from the probe at the top of the image. By clicking on the `Upside down image?' button, the edge detector will scan up from the bottom of the image for the first strong edge, instead of down from the top. Click this button when the image is inverted.

Estimating the probe shape

There are two reasons for wanting to estimate the probe shape, i.e. the region within the acquired rectangular image which actually contains ultrasound data:
  • The probe shape allows the current spatial calibration to be adjusted after the depth setting on the ultrasound machine has been changed.
  • The probe shape allows more accurate image registration for convex array ultrasound probes.
  • To calculate the probe shape, first ensure that the gain on the ultrasound machine is set high, so that you can clearly see the area containing ultrasound data. In order to calculate the shape correctly, you must be able to see at least a part of the top of the ultrasound data, and both of the sides. This means that, although the procedure can theoretically handle depth changes, panning and zooming of the image, it is not possible to pan or zoom such that the edges of the ultrasound data can no longer be seen.

    The following is an example of an acceptable image, seen in the preview window, for a convex probe:

    Setting probe shape - 1

    Now use the segmentation tresholds to select a grey-scale threshold which includes most of the ultrasound data, and very little outside. The segmentation does not have to be particularly good; as long as most of the ultrasound data is included, the probe shape should be calculated correctly. For instance, the following segmentation for a convex probe is perfectly adequate:

    Setting probe shape - 2

    Having done this, you can now click on the `Update probe shape' button in the calibration controls window, provided that you either have the `Review' or `Preview' window open. If there is already a prerecorded data set loaded into Stradx, the probe shape parameters will be updated with no change to the spatial calibration. You can see whether the estimated parameters are correct by clicking on the `Display probe shape' toggle button: this will cause the probe shape to be displayed in red in both the `Review' and `Preview' windows:

    Setting probe shape - 3

    If you have no data loaded, the probe shape can be calculated `live' from the `Preview' window. If, in addition, there is already a previous estimate of probe shape, then the spatial calibration parameters will be adjusted to reflect the new location of the data. You can tell when this is going to happen, since the `Update probe shape' button will change to `Update probe shape (and calibration)' in this instance:

    Probe shape controls

    The probe shape parameters are all stored in the calibration (.sxc) file, hence you will need to save this file again if the calibration has been updated by changing the probe shape.

    The facility will also work with linear array probes:

    Setting probe shape - 4

    B-scan decompression

    To perform sensorless freehand 3D ultrasound it is necessary to have access to the ultrasound envelope intensity signal direct from the scanner. If, as is normally the case, you are using B-scan data then this will have undergone logarithmic compression which changes the statistical properties of the signal. To undo this process we need to estimate the key parameter of the compression that has been performed. If you are using RF data then this step is not required. In this case the system is performing the image intensity compression directly, and thus knows the correct decompression parameter to use.

    To estimate the decompression parameter you should first, segment out about 40 patches of "fully developed speckle" in some of the ultrasound images. An example of two such patches is shown below.

    Two patches of speckle segmented out

    A good strategy is to segment out a couple of patches in roughly twenty different images. Then press the "Find decompression parameter" button. The system will compare the statistics of the compressed speckle patches with the known statistics of perfect speckle and come up with an estimate for the decompression parameter.

    Beam width profile

    This window displays information about the average width of the ultrasound beam in the elevational direction for the top, middle and bottom thirds of the beam. (The top third of the beam is the third of the beam closest to the probe.) It is assumed that the beam has a Gaussian profile in the elevational direction, and the dimension given is one standard deviation of this Gaussian profile.

    This beam-width information may be saved in and loaded from the calibration (.sxc) files and registration (.sxr) files. It is set and used by the Image registration window, particularly when performing sensorless freehand 3D ultrasound.