Computers in Biology and Medicine 36 (2006) 1268–1287
www.intl.elsevierhealth.com/journals/cobm
A supervised method for calculating perfusion/diffusion
mismatch volume in acute ischemic stroke
Judy R. James
a
, Karmen K. Yoder
b
, Olaniyi Osuntokun
c
, Andrew Kalnin
b
,
Askiel Bruno
c
, Evan D. Morris
a,b,∗
a
Department of Biomedical Engineering, Indiana University-Purdue University, (IUPUI) Indianapolis, IN, USA
b
Department of Radiology, Indiana University School of Medicine, Indianapolis, IN, USA
c
Department of Neurology, Indiana University School of Medicine, Indianapolis, IN, USA
Received 9 March 2005; accepted 18 May 2005
Abstract
Diffusion and perfusion (MR) imaging modalities identify overlapping but not identical areas of tissue as lesion
following a stroke. It is thought that the ‘mismatch’ between modalities may represent tissue that could be recovered
with proper (thrombolytic) treatment. We have designed a tool for semi-automated segmentation of the images and
calculation of the mismatch volume. We present results from software phantoms and clinical data. Phantom results
show our mismatch volume calculations are unbiased at realistic noise levels. Clinical data show that raters using
our tool are consistent, fast (15 min per subject) and indistinguishable from an expert using manual segmentation.
᭧
2005 Elsevier Ltd. All rights reserved.
Keywords: Stroke; Diffusion; Perfusion; Mismatch volume; Penumbra; Segmentation
1. Introduction
1.1. Stroke
Acute stroke (cerebral infarction) is caused by sudden oligemia (ischemia), which deprives brain
cells of oxygen and nutrients. In the absence of blood flow, a pathophysiological chain of events be-
gins that leads to cell death. There are reports that suggest that there is a level of cerebral blood flow
∗
Corresponding author. R2 E124, 950 W. Walnut St, Indianapolis, IN 46202, USA. Fax: +1 317 274 8124.
E-mail address: emorris@iupui.edu (E.D. Morris).
0010-4825/$ - see front matter
᭧
2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compbiomed.2005.05.007