This page is used to visualise the results of the estimation of cortical thickness from CT data. This visualisation is automatically selected when the 'thickness' task page is selected.
In the middle of the visualisation controls are some text messages that show whether the current estimate was successful, what the various results are and what the average (least mean square) error was between the modelled and measured CT values along the displayed line, in HU.
At the top, there is a dropdown list for determining what is mapped to the surface, i.e. thickness, mass, density (CT value), FWHM blur (image resolution), endocortical thickness, inner density, outer gap (joint space), outer density or surface offset. These various properties may or may not be available, depending on what technique was used to measure the cortex. If the particular value is not available, then the surface will be shown entirely coloured gray.
After this, there is another dropdown list which selects the colour map to be used to display this data. Various colour maps are available: the best choice will depend on the specific application. The following four sliders control the range of values used for the colour scales when mapping these values.
The smoothing slider can be used to smooth any of the individual estimates over the surface. Setting smoothing to '0' turns it off, '1' just results in any outlier or missing data being filled, whilst higher values also smooth the remaining data. The 'fill missing data' slider controls to what extent missing values are filled in by smoothing.
The 'reject angles above' and 'reject offset above' sliders control outlier rejection for measurement data, i.e. whether they are shown in grey on the surface and subsequently filled in by surrounding values when smoothing. The former is the maximum acceptable angle between triangles in the measured outer cortical surface and the original surface. The latter is the maximum acceptable distance between the original surface and the new outer cortical surface. For angle rejection, this may also lead to the marking of a whole region as an outlier, if it is surrounded by triangles which are all themselves marked as outliers.
The 'reject offset above' slider also controls whether the data-based thickness measurements are compared to other surface locations to determine if they are consistent. If the slider is set to anything less than 40, the thickness is compared to the distance to the next surface intersection. This allows rejection of cortical measurements which really represent the whole object thickness (across both cortices) or where there is only one cortex and no inner compartment. This comparison will only be performed if the surface intersection data is available, for instance by comparing the surface to itself.
The bottom three checkboxes control what is displayed in the visualisation window below the image and 3D windows. This contains a graph of CT values in HU against distance along the currently defined line in mm. The graph can display the 'original data' (in blue), which are just the CT values along the line through the data. It can also display the 'optimised blurred model' (in red), which is what the optimisation process attempts to match as closely as possible with the original data. The 'deconvolved model' (also in red) is the unblurred CT value distribution from which the optimised blurred model is created, and it is this which gives the cortical properties. The default is to display all three graphs.
The 'deconvolved model' checkbox also controls display of the location of the thickness measurement, drawn as a vertical dashed red line. For manual measurements, this indicates where the user clicked in the data. For measurements based on surfaces, this indicates the location of the original surface, and may also include the location of the nearest surface intersection along the negative surface normal direction. This will normally indicate the full extent of the bone region, for example.
The range of CT values (in HU) on the vertical axis is the same as the range of CT values that can actually be seen in the CT image data. Hence to change this range, use the rotate tool to change the windowing parameters in the top left image window, and the graph will also be updated dynamically.
When cortical thickness is mapped over an entire surface, the results are updated dynamically as the calculation progresses, but ensuring that this update does not slow down the actual calculation. It is likely that the actual calculation is much faster than the update. Once the mapping is complete, these results show the averages over the whole surface, taking into account the known error in each of the individual results at each location.