Transcript Certainty of Stroke Diagnosis

Certainty of Stroke Diagnosis:
Incremental Benefit with CT
Perfusion over NC-CT & CTA
Richard I. Aviv, Julia Hopyan,
Anthony Ciarallo, et al
(including Ting-Yim Lee)
Radiology 2010;255:142-153
stroke <1 hr - R hemisphere
ICA, MCA, ACA
termination
occlusion
L
R
stroke <1 hr - CTA source images (SI)
Aviv et al Clin Radiol. 2007;62:447
CT-Perfusion matched defects:
no penumbra! no brain at risk to salvage!
mean transit time MTT
matched CBV = infarction
blood volume CBV
blood flow rCBF
Murphy et al. Stroke 2006. 37:1771
Aviv et al. AJNR 2007; 28:1975
Murphy et al. Radiology. 2008 247:818
CBV for infarction identification
CT perfusion
match or
mismatch
Murphy, Fox, Lee
et al, Stroke 2006
37:1771
CT stroke protocol - 4 (64 +) slice CT




NON CONTRAST CT
~ 20 sec
CTA NECK & HEAD
~ 20 sec
MIP-MRPs (2D) - coronal, sagittal, rotating obliques (secs)
a) MIP MPRs done immediately by techs *
CT-P 2cm (8cm) BASAL GANGLIA & VENTS ~ 40 sec
a) rCBF (cerebral blood flow ml/100 g brain/minute)
b) CBV (blood volume)
c) MTT (transit time)
d) flow maps done immediately by techs
POST CONTRAST CT
~ 20 sec
TOTAL SCANNING ~100 sec NOT MUCH MORE TIME
THAN SIMPLE NCCT ALONE
* no time waste time for rendering; MPRs off operating
console by techs
* main advantage of CT over MR stroke protocol is time
acute stroke – large mismatch CBV
post
contrast CT
shows
grey matter
to
white matter:
shows CBV
CTA SI
shows CBF
CTA SI: Volume or
Flow Weighted?
Sarma et al; AJNR
In Press
matched opercular defect = infarction
CBF
CBV
MTT
sample MPR reformats
no larger cerebral vessel occlusion
right MCA branch occlusion
contrast disappears with scrolling
NC-CT
PC-CT
slab CTA-SI
images show loss of grey matter density on NCCT,
more conspicuous on PCCT, and well seen on SI
Certainty of Stroke Diagnosis:
Incremental Benefit with CT
Perfusion over NC-CT & CTA
Richard I. Aviv, Julia Hopyan,
Anthony Ciarallo, et al
(including Ting-Yim Lee)
Radiology 2010;255:142-153
purpose
systematically evaluate diagnostic
benefits and inter- & intra-observer
reliability of incremental CT protocol
 confirm clinically suspected stroke combined imaging & clinical data as
reference standard

materials and methods


IRB and patient consent
191 consecutive patients ≤3hr stroke with CTP
and MRI <30 days
 4 readers



2 neurology fellows (1st year of stroke fellowship)
2 neuroradiologists (6/12 post fellowship)
incremental review 1 month apart:


NCCT
NCCT and CTA-SI
 NCCT, CTA-SI, and CTP

anonymized data, no clinical data, K-PACS
materials and methods

5-point scale for confidence of ischemia/
infarct presence
 1 stroke definitely absent
 2 stroke probably absent
 3 equivocal
 4 stroke probably present
 5 stroke definitely present
materials and methods

reference standard
 experienced neurologist (5 years)
 review clinical and imaging (CTP/MRI) data
 TIA considered positive
statistical analysis

logistic regression analysis compare real vs
observed diagnosis, adjusting for confidence
 Akaike information criterion (AIC) and ROC
compare performance for each read
 generalized estimating equation to correlate data
from 4 readers, adjusting for confidence
 generalized linear model, binomial distribution
results





191 patients (55% ♂); 67±16yrs
mean NIHSS and ASPECTS 9
time to onset 117±59 mins
median MRI F/U 3 (0-29) days
final diagnosis



stroke 123 (64%)
 large vessel (70%)
 small vessel (30%)
TIA 36 (18%)
stroke mimic 32 (17%)
results

observed diagnosis significantly related
to real diagnosis for all 3 protocols, after
adjusting confidence (LinRegr and GLM)
confidence correct diagnosis:
 3.3 NCCT
 6.4 NCCT+CTA-SI
 13.3 NCCT+CTA-SI+CTP
confidence distribution each sequence
confidence score
NCCT
N (%)
NCCT
CTA N (%)
2 or 4 (probable)
454 (59)
401 (52)
NCCT
CTA
CTP
N (%)
208 (27)
3 (equivocal)
180 (24)
76 (10)
34 (4)
1 or 5 (certain)
130 (17)
287(38)
522 (68)
fewer equivocal/probable responses
Increased certain diagnoses
with CTP usage
results
ROC: confidence score of ≥4 optimal
results

ROC AUC
 0.67
NCCT
 0.72 NCCT+CTA-SI
 0.81 NCCT+CTA-SI+CTP
results
Fair
Moderate
Substantial
NC-CT
CBV
CTA-SI
MTT
CBF
DWI
NC-CT
MTT
CBF
CTA-SI
CBV
DWI
NC-CT
MTT
CBF
CTA-SI
CBV
DWI
NC-CT
CBV
CTA-SI
MTT
CBF
DWI
discussion & conclusions

CTP added to NCCT & CTA improves stroke
detection amongst inexperienced readers
 13x NCCT or 6x CTA
 less experienced readers simulate real
world experience
 real advance: CT infarct diagnosis easy!
 improved confidence of diagnosis
 reliable inter and intra-observer agreement
 neuroradiologists not content with poor
performance with NCCT alone to show infarct