The experimental process of collecting images from macromolecules in an ele
ctron microscope is such that it does not allow for prior specification of
the angular distribution of the projection images. As a consequence, an une
ven distribution of projection directions may occur. Concerns have been rai
sed recently about the behavior of 3D reconstruction algorithms for the cas
e of unevenly distributed projections. It has been illustrated on experimen
tal data that in the case of a heavily uneven distribution of projection di
rections some algorithms tend to elongate the reconstructed volumes along t
he overloaded direction so much as to make a quantitative biological analys
is impossible. In answer to these concerns we have developed a strategy for
quantitative comparison and optimization of 3D reconstruction algorithms.
We apply this strategy to quantitatively analyze algebraic reconstruction t
echniques (ART) with blobs, simultaneous iterative reconstruction technique
s (SIRT) with voxels, and weighted backprojection (WBP). We show that the e
longation artifacts that had been previously reported can be strongly reduc
ed. With our specific choices for the free parameters of the three algorith
ms, NMP reconstructions tend to be inferior to those obtained with either S
IRT or ART and the results obtained with ART are comparable to those with S
IRT, but at a very small fraction of the computational cost of SIRT. (C) 20
01 Academic Press.