Observer reliability of CT Perfusion scans

Background: CT perfusion (CTP) provides potentially
valuable information to guide treatment decisions in
acute stroke. Assessment of inter-observer reliability of
CTP has, however, been limited to small, mostly single
centre studies. We performed a large, internet- based
study to assess observer reliability of CTP interpretation
in acute stroke.
Methods: We selected 24 cases from the IST-3, ATTEST
and POSH studies to illustrate various perfusion abnormalities.
Observers were presented with non-contrast CT,
maps of cerebral blood volume (CBV), cerebral blood flow
(CBF), mean transit time (MTT), delay time (DT) and thresholded
‘‘penumbra’’ maps (dichotomised into penumbra
and core), together with a short clinical vignette. A structured
questionnaire asked observers to comment on
presence of perfusion deficit, its extent compared to
ischemic changes on NCCT, and on whether CTP supported
a decision to give or withhold thrombolytic therapy.
We assessed observer agreement with Krippendorff’salpha
(K-alpha).
Results: Fifty seven observers commented on one to 24
scans (median¼10). Inter-observer agreement was substantial
for presence of an abnormality on perfusion
sequences (k alpha values CBV¼0.73, CBF¼0.64,
MTT¼0.77, DT¼0.78 and PM¼0.8). There was fair
agreement for mismatch between perfusion and NCCT
(k alpha values CBV¼0.21, CBF¼0.26, MTT¼0.39
DT¼0.38 and PM¼0.39). There was only slight agreement
on treatment decisions. (k alpha¼0.2).
Conclusions: Despite substantial agreement between
observers on presence of perfusion abnormalities, there
was less agreement on the degree of ‘‘mismatch’’, and only
slight agreement on treatment decisions.