I have some comments. The search of 4D space, x,y,z,t is interesting
because three of the dimensions are coupled, x,y,z, but the fourth is
not. What I mean by this is that if you are interested in a wide range
of x, you are also interested in a wide range of y and z (normally).
But what spatial range you are interested in is usually not coupled to
the temporal range. For example, perhaps you want a large area of the
image, but at a very specific time. Or the converse is you want a very
small part of the image over a wide time range. This decoupling of the
one dimension has a definite effect on the kind of index tree you want
to have here. The multi-dimensional index in the amanzi-index project
below has only coupled dimensions. We have discussed what it would
take to decouple the dimensions, but have not taken any action.
However, perhaps there is a simplification for your domain. I am under
the impression that there is no correlation between images across
time. In other words, images at different times are in fact images of
different patients, and so cannot be compared. In this case, perhaps
you want one temporal index to find the images, and a set of different
spatial indexes to investigate the contents of each image. In that
case the index in neo4j-spatial is nearly OK, except it is not (yet)
fully 3D. We did much of the work to refactor it to n-dimensional last
year, but I do not believe it is completed. Perhaps this can be
completed as a small project here?
On Feb 6, 10:44 pm, Peter Neubauer <
peter.neuba...@neotechnology.com>
wrote:
> Well, there arehttp://
www.cgal.org/Manual/latest/doc_html/cgal_manual/SearchStructur...
> you could implement, I know Carig Taverner has started on something
> like that athttps://
github.com/craigtaverner/amanzi-indexand it could use
> The Neo4j Heroku Challenge -
http://neo4j-challenge.herokuapp.com/
> >> Hack Real Problems. Win Real Prizes -
http://www.t-b-d.org