Temporal Calibration

Temporal calibration allows Stradx to measure the precise delay between the grabbed images (B-scans) and the incoming positions from the position sensor. It does this by comparing the B-scans with the position sensor information in one of two ways: the first is more accurate, but the second is quicker to perform.

In both cases, the position sensor must be attached to the probe, and you must have set up Stradx so that it can acquire B-scans and positions. The temporal calibration window receives and displays these live, much like the preview window. Ensure that the acquisition rate is set to default, so that Stradx is receiving video frames as fast as possible, otherwise the temporal calibration will not be as accurate as it could be. Comments below the image tell you what to do next, and what the current estimated temporal offset is between the images and the positions.

If you want the most accurate acquisition system, be sure to follow the suggested calibration order.

Accurate Temporal Calibration

This is the default calibration technique. You will need a water bath with a flat bottom which will give a line image in the B-scans. Stradx automatically fits a line to each image (using the same algorithms as used for spatial calibration). The temporal offset between the positions and images can then be calculated by comparing the location of the position sensor (as the probe is moved) with the change in position of the detected line in each image.

By maximising the correlation between these two data sources across several seconds of probe movement, temporal calibration to within +/- 5ms is possible, even if the image acquisition rate is only 25Hz. This is much more accurate than the quick method described below.

Initial state

Calibration is performed in several stages, each guided by comments at the bottom of the temporal calibration window. First, hold the probe vertically such that the bottom of the water bath is orthogonal to the plane of the B-scan. This will result in a nearly horizontal line in the image, as above. Once the probe is sufficiently stationary, you will be asked to move it up and down:

Start moving

As you start moving the probe, Stradx will register that the probe is being moved, and start to record this movement over several seconds:

Keep moving

The probe should then be moved straight up and down, at least once. There is no need to move the probe particularly quickly, but it must be moved by a reasonable amount. Make sure that during the motion:

  • The probe is not translated sideways
  • Stradx always fits a line to the image of the bottom of the bath
  • Otherwise, the calibration may be less accurate, or may fail completely.

    If the calibration is successful, the latest calculated temporal offset will be displayed:

    One calibration completed

    If the calibration was not successful, a red bar will be displayed, indicating what the problem was. Generally, either the probe was not moved up and down far enough, or Stradx could not successfully fit the line in all the images:

    Distance error Calculation error

    It is also possible that the correlation between the position sensor and the location of the line was ambiguous. The correlation algorithm needs some variation (for instance a change of direction) in the movement in order to fit the two data streams together reliably.

    If you only want to perform one calibration, then clicking on `Accept' will set the current temporal offset to the one you just calculated. Alternatively, if you perform five or more calibrations, the results of each calibration will be averaged (allowing for outliers) to give you an even more accurate final result. Clicking on `Accept' now will use the averaged result rather than the last one.

    Stradx also keeps a record of which position sensor and video source the calibration relates to (eg. Composite PAL/Polaris in the examples on this page). This label is displayed in the temporal calibration window and also stored in the setup file, along with the calibration result itself.

    Several calibrations completed

    If you are having trouble getting the automatic line detection to work, then you can change the number of vertical scan lines, and a variety of other parameters, using the calibration controls panel. Generally, the best results can be achieved when the acoustic power and gain of the ultrasound machine are set fairly low.

    For convex probes, you will probably also want to define a region to limit the search for a horizontal line in the image. This can be done exactly as for spatial calibration. Clicking on the 'Define scan area' button allows you to define a (convex) polygon bordering the region. Comments below the image show what each mouse button will do:

    Defining a region

    Once the region has been defined, subsequent line fitting will only happen within this region:

    Calibration using a region

    Quick Temporal Calibration

    This is an alternative calibration technique which does not require a water bath, but is only accurate to approximately +/- 40ms. It can be selected by clicking on the 'Quick calibration' toggle button. The technique looks for a sudden change in both the images and in the positions, after the probe was initially held stationary.

    Once again, comments below the image will show you what to do at each stage. Start by holding the probe against your skin, or anything else which will give an image.

    Quick initial state

    When the instruction in the window changes and tells you to move the probe quickly, you should lift the probe rapidly away from your skin. Be careful not to drop the probe as you do this! The system expects the position sensor to move at least 5 cm in one tenth of a second, otherwise it will not attempt to calculate a time offset. You may find that rotating the probe as you pull it away from your hand improves the suddenness of the motion.

    Quick movement

    The system will then calculate the correct time offset to match up the movement recorded by the position sensor with the change in the image. As with the more accurate calibration, you can either accept the latest result, or wait until you have calculated five or more temporal offsets, and use the average.

    If the calibration was not successful, a warning will be displayed in red indicating why this was the case:

    Quick movement error

    Note that, particularly for the Polaris tracking system, it is very difficult to move the probe sufficiently fast to enable this sort of calibration, without also causing the Polaris to lose track of the position sensor. In this case it would be better to use the alternative, more accurate, method.