Abstract #0678

A Rapid, Robust, Anatomy and Atlas Guided Lesion Quantification Framework from Diffusion Weighted MR Images

Sumit K. Nath¹, Dattesh Dayanand Shanbhag¹, Rakesh Mullick¹, Uday Patil¹, Marie Luby², Katherine D. Ku², Lawrence L. Latour², Steven Warach², - NINDS Natural History of Stroke Investigators²

¹Imaging Technologies, GE Global Research, Bangalore, Karnataka, India; ²NINDS, NIH, Bethesda, MD, United States

A novel anatomical and atlas guided split-and-merge algorithm is presented for quantifying potential lesions in diffusion weighted MR images. Compared with a conventional non split-and-merge method, our approach leads to highly improved results when analyzed with ground truth.