Abstract #1466

Can a Human Operator Be Replaced by a Computer for Processing of Bolus-Tracking Perfusion Data?

Matus Straka¹, Michael Mlynash², Greg Zaharchuk¹, Gregory W. Albers², Roland Bammer¹

¹Lucas Center, Department of Radiology, Stanford University, Stanford, CA, USA; ²Stroke Center, Department of Neurology and Neurological Sciences, Stanford University, Stanford, CA, USA

Bolus-tracking DSC-MRI PWI is an important technique in acute stroke, but its use in clinical routine is hampered by need of manual processing. We have developed a fully automated system aimed on perfusion processing. A key part of the system is the automated selection of AIF and VOF signals in the data. In this study, we compared manual and automatic selections on 30 cases with focus on noisy data and data with motion. The results indicate high success (>90%) rate of the system on various bolus-tracking datasets and allow usage of such algorithm in clinical routine.