back to abstract
[ Compact chromatin
| Gallery B ][ Mémoire
de DEA | Gallery A ][ TEM
methods ][ Methodol
Three-Dimensional Reconstruction and Analysis
of Compact Chromatin
in the Nucleus of G0 Rat Lymphocytes (plates).
text in black correspond to paper legends
text in green is not on the paper
Chromatin domains have been identified as the weakly-connected components of the bulk. This segmentation is a simplified version of 3D-watershed segmentation: bulk initial components (usually a single large one) are first eroded 'to break' their weak interconnections. Then, connected -components labelling, by surface tracking, is done. A reconstruction process is applied by conditioned morphological dilation on each labelled body. As these are part of the initial bulk, a label-propagation is accomplished during conditional dilation. The following plates illustrate the full reconstruction process.
PLATE 4. Compact chromatin (white) and peripheral chromatin (gray) in contact with the nuclear enveloppe of a rat G0 lynphocite. These slices are a sample of a series of 63.
Fig. 1. Consecutive sections of the nucleus of a lymphocyte. The compact chromatin is white and all other structures black. The chromatin is shown without morphological erosion, segmentation or labelling. A sheet of peripheral chromatin of irregular width is interrupted only by the nuclear pores. The central mass is the perinucleolar body of chromatin. The nucleolus is not contained in this section. This body is continuous with the peripheral sheet by means of thin filaments.
PLATE 5. Segmentation and reconstruction of weakly-connected clumps. Colors identified each connected component. Interface surfaces bewteen chormatin bodies is also measured.
Fig. 2. Section showing the individualization of the compact chromatin bodies after erosion and color labelling. The peripheral gray region is the user marked-out chromatine; the white pixels are those changed during internal erosion.
Fig. 3. The same section depicted in figure 2 after conditional dilatation. The information is reincorporated to the chromatin bodies already individualized.
Surface Rendering Visualization
PLATE 6. 3D-rendering of the labelled nucleus. The different bulk components are weakly-connected along color boundaries. Each body can be indiviudually visualized and quantified.
Fig. 4. External view of a nucleus. The areas of different colors are the regions of contact with the nuclear envelope of the chromatin bodies labeled with the same color. The yellow areas show the regions of contact of the perinucleolar body. The orange voxels represent the nuclear pores.
PLATES 7 (a) Smallest 38 (from 41) bodies. (b) BIggest 5 bodies. Small isolated grains are remarkedly few in number (tens) and they may be rather an artifact from reconstruction errors, due to image misalignement and deformations not fully corrected.
PLATE 8. A single large chromatine
body. Orange voxels represent the pore
distribution at the nuclear enveloppe.
Fig. 5. Incomplete reconstruction of a large body of perinucleolar chromatin. The central space free of compact chromatin corresponds to the nucleolus. Four regions of contact with the nuclear envelope can be seen as convex smooth surfaces, the one on the right is not completely reconstructed and appears irregular.
Fig. 6. Two perinucleolar chromatin bodies completely reconstructed. The holes occupied by the nucleoli are covered by chromatin. The areas of contact with the nuclear membrane are diametrically opposed.
PLATE 9. Hand-made analysis of chromatine domains, from slice-by-slice drawings from the TEM micrographs (original design and photograph by Dra. Clara Esquivel and Dr. Gerardo Vázquez-Nim, Facultad de Ciencias of the UNAM ).
PLATE 10. Computer equivalent of domain identification, with automatic 2D-segmentation, slice alignement and 3D-weak-connected component extraction, labelling and 3D-reconstruction. At this time (1993) resolution was very low (~80x100 pixels), but this analysis can now be effectuated at full acquisition resolution (512x512xNslices).
PLATE 11. A polar-projection of chromatin domain into the plane allows to fully visualize domain distribution in a sphere centered at the center of mass of the nucleus. Initial information about pore locarion is superposed as black speckles. Must detected bodies contact the nuclear enveloppe, thus very few remain fully inside the nucleus. Studies on different nuclei reveal very similar patterns.
Fig. 7. Peripheral surface map showing the domains of contact
of the color labeled
chromatin bodies. The pores are shown in white.
PLATE 12. A polar thickness projection of chromatine bulk. Depth thickness from periphery to the center of mass is coded in gray levels, to render visible the location of thicker clumps. Light-gray regions correspond to bulky chromatin, and dark-gray levels correspond to thin zones . This is a "volume-casting" technique in which voxels are ray-traced to measure depthness. Unfortunately, pore channels are very tortuous, which blurs an expected observation: pores should be located along dark region borders.
PLATE 13. The polar thickness
map, as superposed on the nucleus. Color represents thin (dark-red)
peripheric chromatin, mostly associated with the nuclear enveloppe. Thick,
bulky zones penetrate into the core of the nucleolus (light-yellow) (see
Plate10). The apparent higher resolution
(compare with the surface renderings)
comes from the 3D-Bresenham traversal of the volume, which is done at a
sub-voxel resolution, and averaging measurements for depth values as well
as surface intersections.
back to abstract
[ Compact chromatin | Gallery B ][ Mémoire de DEA | Gallery A ][ TEM methods ][ Methodol .2D-3D ][ Home ]
page by Jorge
Márquez January 25, 1998. email@example.com