Image display

The image display dialog box is used to control the display and calibration of any 16-bit data, which could come from DICOM files or another imaging format.

Converting image data

DICOM files containing CT data include calibration values for converting the raw data (i.e. as stored in the data files) to Hounsfield Units (HU), or (GY) or any other unit. The unit supplied in the data files is shown at the top of this dialog, where '??' means that it was not defined in the data files. The conversions are represented by a multiplier and an offset, both of which are displayed in the following text fields. If the units are un-defined ('??'), then these values can be edited to perform any linear conversion of the raw data to the display data. In this case you are free to supply any textual value for the units. 'HU' should be used if the conversion is to Housefield Units, since this is recognised by other parts of the software.

In order to display 16-bit data (for instance in HU or GY, or MRI data), it also needs to be 'windowed', i.e. a specific range of the data values are selected for conversion to the 8-bit range which is actually displayed as an image. This is achieved by specifying a window centre (the value that will be displayed mid-grey) and a window width (the range of values which will be displayed, with the lowest in this range displayed as black, and the highest as white). This windowing can either be done using the rotate tool, or by using the next two sliders in this dialog.

If the data is CT and includes a scan of a bone density phantom, Stradview can also calculate the conversion between the stored Hounsfield Units and bone density in mg/cm3. The location of the phantom in the data is determined by using the first four landmarks. The first two landmarks should be placed near the centre of the rods at each edge of the phantom in the first CT image in which the phantom fully appears, except for the Image Analysis Solid HA phantom. For this phantom, the landmarks should be placed on the rod with the highest density, and the rod with the second-to-lowest density: the lowest density rod is not visible since it looks the same as the background material. The next two landmarks should be placed at the same locations, and in the same order, in the last CT image in which the phantom fully appears. Having done this, pressing the 'Calibrate for BMD' button will calculate the calibration parameters.

Stradview currently recognises the Mindways liquid and solid K2HPO4 phantoms, the Siemens Osteo HA, the Image Analysis Solid HA and the QRM Bone Density 3-insert and 6-insert HA phantoms. If you are not using one of these phantoms, or if Stradview does not detect the phantom which you are using, the calibration values will be incorrect. You can verify whether the calibration is reasonable by looking at the various estimated rod values in the dialog box showing calibration results.

Smoothing image data

The image smoothing sliders apply Gaussian filters to the raw data during conversion to display data. The 'within each image (xy)' slider controls the filtering within each original image, whereas the other slider controls filtering across images, so using them both results in 3D filtering of the raw data. They are kept separate since the noise may not be the same in each of these dimensions. Using these filters will smooth the data, which can help for both automatic segmentation and cortical thickness mapping if the data is particularly noisy.

When changing the windowing or filtering, the 8-bit display data has to be re-created from the raw 16-bit data. This is done as needed, but can take a while if all the data needs to be processed, for instance if you are currently displaying a reslice or volume rendering, or if the frame filter is being used.