Factor analysis of medical image sequences (FAMIS)

Summary FAMIS is a method used to summarize the information underlying time or energy sequences of medical images using only a small number (3 to 5) of images and associated curves. FAMIS is based on a linear additive model. and describe the information included in the whole series as the sum of several components that are physiologically or physically meaningful. For instance, FAMIS allows one to separate a complex dynamic process into a sum or more easily understandable processes (like a region presenting an increasing uptake over time, and another in which uptake does not vary over time). It also allows one to separate primary from scattered photons in SPECT and PET for scatter correction.
we have been involved in developping new methods for FAMIS for about 10 years, to improve the accuracy of the method and find new applications to it.

Our main results are about:
  • The choice of the optimal metrics to be used in FAMIS to best identify the parameters of FAMIS model as the function of the statistical properties of the images.
  • The introduction of priors in FAMIS, to help identify a unique solution to the model.
  • The determination of the statistical properties of factors and factor images resulting from FAMIS.
  • The generalization of FAMIS so that the solution can be estimated in a subspace of dimensions greater than the number of factor to be identified.
Recent applications of our work regarding FAMIS are:
  • Scatter correction in planar scintigraphie or SPECT.
  • Cross-talk correction for dual-isotope SPECT.
  • Creation of fused perfusion-function images from gated PET or gated SPECT.
  • Characterization of C11-choline kinetic in brain tumor and necritic tissues from dynamic PET scans.
We also participated to the development of Pixies, a software implenting FAMIS..
Fundings This work has been partly supported by l'Association pour la Recherche sur le Cancer, la Fondation pour la Recherche Médicale, and the French "Ministère de la Recherche".
Publications - N. Freedman, R. Chisin, R. Rubinstein, Y. Shoshan, T. Siegal, I. Buvat. Factor Analysis (FA) in the evaluation of dynamic PET brain tumor imaging with C-11 choline. J. Nucl. Med. 43, 249P, 2002. Poster (PDF)
- S. Hapdey, M. Soret, C. Riddell, H. Benali, I. Buvat. Generalized spectral factor analysis using Klein-Nishina priors for cross-talk correction in simultaneous 99mTc/123I brain SPECT. J. Nucl. Med. 42, 139P, 2001.
- I. Buvat, S.L. Bacharach, A.N. Kitsiou, V. Dilsizian, H. Benali, R. Di Paola. Fused images for combined assessment of myocardial perfusion and function. J. Nucl. Med, 41, 157P, 2000. Poster (PDF)
- I. Buvat, S. Hapdey, H. Guillemet, H. Benali and R. Di Paola. Generalized factor analysis of medical image series (FAMIS) for accurate quantitation. Eur. J. Nucl. Med., 27, 966. 2000. Oral presentation (PDF)
- I. Buvat, S. Hapdey, H. Benali, A. Todd-Pokropek, R. Di Paola. Spectral factor analysis for multi-isotope imaging in nuclear  medicine. Information Processing in Medical Imaging. A. Kuba, M. Samal and A. Todd-Pokropek eds, Springer, Berlin, 442-447, 1999. Article (PDF)
- I. Buvat, H. Benali, R. Di Paola. Statistical distribution of factors and factor images in factor analysis of medical image sequences. Phys. Med. Biol. 43, 1695-1711, 1998. Article (PDF)
- H. Benali, I. Buvat, F. Frouin, J.P. Bazin, R. Di Paola. Foundations of factor analysis of medical image sequences: a unified approach and some practical implications. Image and Vision Computing. 12, 375-385, 1994.
- I. Buvat, H. Benali, F. Frouin, J.P. Bazin, R. Di Paola. Target apex-seeking in factor analysis of medical image sequences. Phys. Med. Biol. 38, 123-138, 1993. Article (PDF)
- H. Benali, I. Buvat, F. Frouin, J.P. Bazin, R. Di Paola. A statistical model for the determination of the optimal metric in factor analysis of medical image sequences. Phys. Med. Biol. 38, 1065-1080, 1993.Article (PDF)
- F. Frouin, L. Cinotti, H. Benali, I. Buvat, J.P. Bazin, P. Millet, R. Di Paola. Extraction of functional volumes from medical dynamic volumetric data sets. Comput. Med. Imaging Graph. 17, 397-404, 1993.
More... A FAMIS tutorial Slides in French and associated comments (PDF)
Contacts Irène Buvat : buvat@imed.jussieu.fr