World Congress on Medical and Biomedical Engineering, San Antonio, Texas, 6-13 August. Physics in Medicine and Biology, 33(Supplement I):8, 1988.
A method is presented which insures that the sum of the weighting arguments of any temporal linear filter is zero, while preserving the significant aspects of the filter. By insuring that this sum is zero, constant features as well as noise which has an expected value of zero is suppressed in the final image. A new sequence is formed by subtracting the average image from each image in the original sequence. This new sequence is used to define the signature vectors of features that are to be enhanced and/or suppressed. These signatures are used in the filter definition to create a new linear filter which is then applied to the original sequence. A mathematical derivation of the convolution of these two operations shows that the sum of the weighting arguments of the new filter is zero. The method is applied to the classical matched filter and the eigenimage filter. A comparison is made between the technical characteristics of the modified and the original filters. This method is illustrated on a cardiac study and a simulated sequence.