Dattesh D. Shanbhag1, Rakesh Mullick1,
Sumit K. Nath1, Catherine Oppenheim2, Marie Luby3,
Katherine D. Ku3,
1Imaging Technologies, GE Global
Research, Bangalore, Karnataka, India; 2Department of
Neuroradiology, Universit Paris-Descartes, Paris, France; 3NINDS,
NIH, Bethesda, MD, United States
In
this work, we demonstrate a fully automated, fast and robust analysis
pipeline for segmenting the perfusion lesion on different PWI maps (MTT,
Tmax, TTP) and mismatch in acute ischemic stroke setting. The automatically
segmented perfusion lesion and mismatch volume showed a strong correlation of
0.9 and 0.88 respectively, when compared to manually segmented PWI lesion on
MTT maps. Variability for perfusion lesion volume estimates were lower compared
to manual inter-rater variability, but was higher for mismatch estimates.
Overall, Tmax PWI lesion had a lower volume compared to MTT PWI lesion.
Keywords