LINEAR DIMENSION REDUCTION OF SEQUENCES OF MEDICAL IMAGES .3. FACTOR-ANALYSIS IN SIGNAL SPACE

A method is presented for improving the precision of factor analysis b y utilizing physiological information. The first step is an optimal li near dimension reduction, whereby the data are projected onto a low-di mensional signal space. Then, principal component analysis is performe d in the signal space rather than in the entire data space. This impro ves the precision of the principal components. Unlike ordinary princip al component analysis, the present method is not degraded when the tim e intervals are subdivided, provided that the signal space is correct. Alternatively, but with identical results, the covariance matrix can be calculated from the whole data space. The covariance matrix is then transformed and principal component analysis is performed in either a low-rank matrix or a low-dimensional submatrix instead of in the whol e covariance matrix. Factor analysis using the intersection method wit h a theory space may be improved by employing the present method. In s imulations based on a [C-11]flumazenil study with 27 frames, the propo sed method required only 58 per cent of the radioactivity to produce t he same precision as the intersection method and only 27 per cent when compared to ordinary principal component analysis.