Transcript Interpretation of magnetic resonance imaging in the

Interpretation of magnetic
resonance imaging in the chronic
phase of traumatic brain injury
Jussi Laalo 1, Timo Kurki2, Olli Tenovuo*3
1Department
of Radiology, University of Turku, Finland,
2Pulssi Imaging Center, Finland, 3Department of Neurology,
University of Turku, Finland
*presenting author
Introduction
• MRI is the imaging method of choice in postacute TBI or suspected TBI sequels
• Its findings may be of great importance for
differential diagnosis, treatment decisions,
outcome prediction, and medicolegal
purposes
Introduction, continued
• We found no studies over the accuracy of
consultant neuroradiologists’ head MRI reports
- in many studies the report of one
neuroradiologist has been the reference
diagnosis.
• Regarding CT of acute TBI, we have recently
shown that there is a marked variation even
between the most experienced readers in the
detection of brain injury findings
(Laalo J, Kurki T, Tenovuo O, Sonninen P. Reliability of diagnosis of
traumatic brain injury by computed tomography in acute phase.
Journal of Neurotrauma 2009;26:2169-78.)
Objectives
To study eventual differences (in quantity and
quality) in detecting the late stage TBI findings
in MRI between two neuroradiologists and to
compare these interpretations with the original
report
Material and methods
• Randomly selected 89 cranial MRIexaminations from patients with clinically
evident TBI and 11 non-TBI controls (scattered
among TBI images, with blinded evaluation)
• Reviewed independently by two
neuroradiologists: one with 14 years of
experience within the subspecialty (R1), and
another with recent completion of subspecialty
training (R2)
Material and methods
• All examinations were performed on 1,5 T MRscanners and included T2, gradient echo (T2*) and
FLAIR sequences. Eventual other available
sequences were also reviewed.
• The nature, location, and side of the finding was
recorded. The classes used were: brain contusion,
subdural effusion, and diffuse axonal injury (DAI).
DAI included subclasses of spot-like haemorrhages,
spot-like hyperintensities and localized atrophy.
The readers also stated their view of DAI being
evident, possible or absent.
Results
Number of findings
WM hyperintensities
Contusions
Spot-like haemorrhages
Local atrophy
DAI evident/possible
R1
R2
Original
380
182
43
57
98
41*
259
134
35
57
33
40*
173
70
23
52
28
---
*Only 30 of these concerned same subjects
= R2 vs R1
= original vs R1
Agreement
Strong
Good
Moderate
Poor
Results
In the original report, the differences in
interpretations were all directed towards false
negative. These findings and locations were:
• Atrophy in both frontal lobes and the right
temporal lobe
• Juxtacortical T2 hyperintensities in both frontal
lobes
• T2 hyperintesities of deep WM in both frontal
lobes and both parietal lobes and in the corpus
callosum
• Contusion findings in both frontal lobes
Results
There was consensus between the two
neuroradiologists over whether DAI is present or
not in 68/89 (76 %).
In the control group 4/11 patients had findings, all of
which were WM hyperintensities. A patient with 16
WM hyperintensities was reported by R2 as DAI being
possible, all others were reported as DAI absent by
both neuroradiologists.
Conclusions
• The interpretation of TBI findings in late-stage MRI
is difficult, yielding significant variability also
between specialists in neuroradiology.
• This may endanger correct diagnostics and lead to
false treatment decisions and medicolegal
problems.
• Detecting atrophy seems to be the most
demanding task while interpreting images of
chronic TBI.
Conclusions
• Standardized quantitative image analysis
programs should be developed to be used in
clinical practice.
• A visual interpretation of MRI images is too
pendulous, puts the correct diagnostics in risk, and
leaves the vast majority of the available imaging
information unused.