== Typical problems in vector reconstruction and their solutions ==

Most commonly available 2D atlases were not designed to be used as
sources for 3D models generation.  Precision of contour definitions
and label placement is sufficient for standard use, but not always for
the automatic reconstruction process. For example, in some published atlases
one may find a number of deliberate simplifications e.g. labels are
omitted because names of some areas are considered obvious or labels
are literally too large to put them inside the represented area. Such
omissions have to be fixed in order to get a correct 3D
reconstruction, indeed for any systematic computer processing.


[[Image(006__final_comparison.gif)]]


=== Non-interactive error correction ===

The vector parser tries to automatically deal with troublesome
situations.  However, if it is unable to correct a problem
automatically, the program will notify the user and save all the
important information about the case as a thumbnail image including
name of the problematic label and its location. These data help the
user to solve the problem manually by removing ambiguities from
contour slides. Below we discuss common inconveniences and our
strategies for fixing them.

=== Leaking structures ===

The most common defect is gaps in structure outlines. It frequently
happens when two contour lines, which should touch are drawn so that
they leave a little space in between. Such an arrangement may not be
visible in a printed atlas but it greatly disturbs the tracing process
where every pixel may influence the results. In this case, the
structure being traced overtakes the space of its neighbour through
the broken contour which we call ''leaking of the structure'' '''(see
the spot indicated by a red arrow on Fig.~3A)'''.

=== Gap filling algorithm ===

It is handled by a heuristic gap filling algorithm ([wiki:barSoftwareGapFillDetails detailed description]). The main idea
behind this algorithm is to expand the contours by applying a [http://en.wikipedia.org/wiki/Mathematical_morphology dilation] filter ([http://www.imageprocessingplace.com/DIP-3E/dip3e_main_page.htm Gonzalez2008]) until the boundary closes. If a very
accurate reconstruction is required, the best policy is to prepare a
precise contour slide so there is no need to apply gap filling. For
the atlases we tested, the gap filling algorithm performed well, the
reconstructed structures were adequate and the gains in time were
tremendous as compared with manual cleaning.

[[Image(002__gapfill.gif)]]


=== Duplicated labels within a single contour ===

Another common difficulty is an absence of borders drawn within an outline containing multiple labels.
This can be a consequence of improper drawing but often
simply follows from lack of sharp transitions between structures
('''see cortical areas in Figure...''').
When the parser detects two or more overlapping regions that were
created based on different labels such labels are changed to
spotlabels in the CAF slide and a notification is logged. Neither
path is removed as there is no way to determine which one should be
discarded and which should be kept. After tracing, the user is notified
about ambiguous delineation and should review the slide in question
and correct it.

=== Labels placed over contours ===

Error correction procedure also looks for labels anchored directly
above contours which may happen when a structure is defined as a thin
strip between other structures or placed outside the brain outline. Since
tracing contours themselves is not allowed, such incidents are
recorded and a thumbnail image is generated allowing user to examine
and correct the error.  Tracing of a region corresponding to the
particular label is skipped but in general tracing is not interrupted
as the whole procedure works non-interactively.

[[Image(003__invalid_labels.gif)]]

=== Unlabelled area ===

Another problem we encountered during tracing was the lack of labels
attributed to some areas.  3dBAR vector parser automatically detects
such unlabeled regions by comparing the sum of traced paths with the
total area of the ''Brain'' structure. The parser locates all
contiguous unlabelled regions consisting of more than a given number
of pixels and traces them.  The resulting structures are called
''Unlabelled'' and may be indexed and reconstructed similarly to
the regular structures.


[[Image(001__unlabelled.gif)]]


[[Image(004__new_edges.gif)]]
[[Image(007__final_electrode.gif)]]