The text below was written in 2006, and has now been largely superseded by events. Specifically, FTD has been recognised as a valid analysis method in its own right, as a result of comparison with other analytical methods; our software offerings are now split in TWO : TrackWorks (which creates the data files and images necessary for analysis) and FastTracks (which allows analysis to be carried out anywhere in the world, using the data files created by TrackWorks).
The new software actually consists of 3 modules : an image capture module, an automated track counting module, and a results review module.
At this point in time, the results of the automated counting process have not yet been rigorously verified by a sufficient number of users (although Prof. Gleadow has carried out some comparisons involving the same mounts counted in the traditional manner, and again with our new software, and the results have been very encouraging).
We therefore decided to provide user access to the process in three modules, as follows :
a) the first step is the capture and storage of suitable images in a very specialised manner, using a proprietary technique.
b) the next step is the automated counting process which can now be carried on off-line on a different computer, if the user so desires. This is a considerable advantage, as it takes the load off the microscope system, which is the "bottle-neck" in the process.
c) the final step allows the user to review and modify the track count. There will probably be some small number of marginal objects which may be mis-identified as false positive or false negative, and the user will have the opportunity to re-categorise these. In our experience, the number of false hits has been very small indeed. We believe at this time that the level of errors incurred during automatic counting is of the same order as that incurred as a result of manual counting.
We should also add that a major advantage of this technique is that, on current indications, the track density limit for counting is about an order of magnitude greater than that for manual counting. Also, the reviewing and correction process takes only a fraction of the time that would be required for a manual count, and the results, needless to say, are much more consistent, because they are not subject to the inevitable and well-known variabilities of a human operator.
(Because some operator intervention is required, we actually hesitated to call this process "fully automated", but believe that this is a fair label)