Designing and analyzing virtual cuts in 3D models of pig bodies by mapping cuts from a statistical atlas
Special Symposium - Innovations in Food Technology (LMC Congress)
Flexible Production, PAT & Modelling (Food-3b)
Keywords: Atlas, image registration, computed Tomography, meat quality
The development and introduction of new products within the
slaughter industry is a slow and expensive process as it is
difficult to analyze the products wrt. earning potential and
quality. The difficulty primarily stems from the unfortunate fact
that it until recently was impossible access the quality of a
product or combination of products until the cutting had been
carried out. With Computed tomography (CT) it is possible to
construct 3D models of the interior of pig bodies without
performing an invasive procedure. A CT scanner generates a 3D
volume of the interior of an object by moving a x-ray source in a
spiral motion around the object while sensors
continuously record the attenuation of the intensity of the x-ray beam.
Given a population representative database of 3D pig models, it
becomes possible to compare the earning potential and the overall
quality of product combinations, as well as to determine which
types of pigs are suitable for a given product range. In order for
such a analysis to be carried out successfully it is essential
that identical cuts can be located in the 3D models and that the
crucial quality parameters can be extracted consistently across
the models. These demands necessitate the establishment of an
almost complete physical correspondence between the models in the
database. This can be accomplished by registration of the 3D pig
models to an atlas. An atlas could simply be a selected 3D pig
model where anatomical points and regions have been identified by
an expect. Cuts or parts of
cuts can be mapped from an atlas to the 3D pig models and vice versa.
Substantial challenges remain in converting the vaguely defined
cuts from product catalogues, which requires human interpretation
and prior knowledge, to concise
mathematical models which can be applied to an atlas.
In addition to mapping cuts, an atlas can also provide important
and valuable statistical information about the population of pigs,
which makes an atlas a widely applicable tool for a variety of
research and development projects, e.g. the development of
slaughter and cutting robots.
The presentation will give a detailed description into the
construction of statistical atlases including the process of
registration of 3D models and will demonstrate a wide range of
cutting examples. Possible pitfalls and problems will likewise be
discussed.
Presented Wednesday 19, 17:25 to 17:30, in session Flexible Production, PAT & Modelling (Food-3b).
