Campus Event Calendar
Event Entry
What and Who
A Data-Driven Approach for the Generation of Quantitative Digital Atlases of Human Anatomy
Patrick Ernst
DFKI Kaiserslautern
D5 application talk
AG 5
AG Audience
English
Note: We use this to send email in the morning.
Date, Time and Location
Friday, 2 December 2011
13:00
60 Minutes
E1 4
433
Saarbrücken
Abstract
In medicine spatial relations are important information to describe
anatomical facts. Based on them, digital quantitative atlases can be
build. The aim of my masters thesis was to create such atlases using a
data-driven approach and to realize a semantic integration of this
generated knowledge using Semantic Web technologies.
In my talk I firstly want to present a system architecture which
allows the data-driven extraction of spatial relations. Relations are
thereby formulated using fuzzy sets and fuzzy logic. Based on them,
the creation of two different atlas types is introduced. Single model
atlases are constructed by incorporating all information of one
particular relation into one overall model representation. For
extended atlases a clustering algorithm is previously applied in order
to group anatomies containing similar relations. Leveraging such
atlases I explain two applications. Spatial consistency checks can be
used to check the plausibility of the output of automatic object
recognition techniques against quantitative atlases. Furthermore, an
interface is presented capable to transfer quantitative atlases into
qualitative models.
The last part of my talk concerns the retrieval of spatial relations
and atlases. Therefore, a SPARQL engine is proposed which directly
relies on a spatial database and 3D images. The engine generates query
results dynamically, when they are requested. Furthermore, I describe
some use cases exemplifying new possibilities which emerge from such
an semantic integration.
anatomical facts. Based on them, digital quantitative atlases can be
build. The aim of my masters thesis was to create such atlases using a
data-driven approach and to realize a semantic integration of this
generated knowledge using Semantic Web technologies.
In my talk I firstly want to present a system architecture which
allows the data-driven extraction of spatial relations. Relations are
thereby formulated using fuzzy sets and fuzzy logic. Based on them,
the creation of two different atlas types is introduced. Single model
atlases are constructed by incorporating all information of one
particular relation into one overall model representation. For
extended atlases a clustering algorithm is previously applied in order
to group anatomies containing similar relations. Leveraging such
atlases I explain two applications. Spatial consistency checks can be
used to check the plausibility of the output of automatic object
recognition techniques against quantitative atlases. Furthermore, an
interface is presented capable to transfer quantitative atlases into
qualitative models.
The last part of my talk concerns the retrieval of spatial relations
and atlases. Therefore, a SPARQL engine is proposed which directly
relies on a spatial database and 3D images. The engine generates query
results dynamically, when they are requested. Furthermore, I describe
some use cases exemplifying new possibilities which emerge from such
an semantic integration.
Contact
Gerhard Weikum
5000
--email hidden
passcode not visible
logged in users only
Petra Schaaf, 12/01/2011 10:16 -- Created document.