Abstract #0552
Combined unsupervised-supervised classification of multiparametric PET/MRI imaging data of the prostate
Sergios Gatidis 1 , Petros Martirosian 1 , Thomas Kstner 1 , Ilja Bezrukov 2 , Marcus Scharpf 3 , Christina Schraml 1 , Nina F Schwenzer 1 , and Holger Schimdt 1,2
1
Department of Radiology, University of
Tbingen, Tbingen, BW, Germany,
2
Department
of Preclinical Imaging and Radiopharmacy, University of
Tbingen, Tbingen, BW, Germany,
3
Department
of Pathology, University of Tbingen, Tbingen, BW,
Germany
In this study we implemented and evaluated a combined
unsupervised-supervised classification algorithm for the
analysis of multiparametrc PET/MRI data that allows for
robust classification in cases where prior knowledge
about the data is limited. We applied the proposed
method to [11C]-Choline-PET/MRI data of the prostate and
observed high classification accuracy compared to manual
tumor delineation and to histological slices. Numerous
applications of this approach are conceivable,
especially in areas where histopathological correlation
is difficult (e.g. brain imaging) and thus knowledge
about ground truth is limited.
This abstract and the presentation materials are available to members only;
a login is required.
Join Here
