Reliability of Acute Traumatic Brain Injury Diagnosis with CT

1. Reliability Of Diagnosis Of Traumatic Brain Injury By Computed Tomography In The Acute Phase Reliability Of Diagnosis Of Traumatic Brain Injury By Computed Tomography In The Acute Phase Olli Tenovuo Olli Tenovuo Department of Neurology Department of Neurology University of Turku University of Turku Finland Finland

2. Introduction Introduction CT is important in the diagnosis and evaluation of acute TBI. CT is important in the diagnosis and evaluation of acute TBI. CT is not very reliable in showing non- haemorrhagic brain injuries, particularly small contusions or traumatic axonal injury. CT is not very reliable in showing non- haemorrhagic brain injuries, particularly small contusions or traumatic axonal injury. The interpretation of these vague findings seems to be very difficult, and might depend on the experience of the reader. The interpretation of these vague findings seems to be very difficult, and might depend on the experience of the reader.

3. Introduction, continued Introduction, continued There is also evidence of difficulty in diagnosing small epidural and subdural haemorrhages. There is also evidence of difficulty in diagnosing small epidural and subdural haemorrhages. As a great number of examinations are performed in off-duty hours, when the readers are radiology residents on call, there seems to be a potential risk of missing the radiological diagnosis of acute TBI. As a great number of examinations are performed in off-duty hours, when the readers are radiology residents on call, there seems to be a potential risk of missing the radiological diagnosis of acute TBI.

4. Purpose of the study Purpose of the study To evaluate the accuracy of the CT diagnosis of acute TBI. To evaluate the accuracy of the CT diagnosis of acute TBI. To compare the CT interpretation of traumatic findings among experienced readers and between experienced and less experienced readers. To compare the CT interpretation of traumatic findings among experienced readers and between experienced and less experienced readers.

5. Material and methods Material and methods 100 acute cranial CT scans from 2003, where a suspected acute TBI was indicated. 100 acute cranial CT scans from 2003, where a suspected acute TBI was indicated. Setting: the emergency ward of a university hospital. Setting: the emergency ward of a university hospital.

6. Material and methods, continued Material and methods, continued We evaluated We evaluated the rate of misinterpretations, the rate of misinterpretations, the nature of the findings most often missed, the nature of the findings most often missed, the differences in interpretation related to the experience of the reader, the differences in interpretation related to the experience of the reader, the variation among experienced readers reports. the variation among experienced readers reports.

7. Material and methods, continued Material and methods, continued In those cases where the reports of the three study readers were exactly the same, this was classified as the final diagnosis. In those cases where the reports of the three study readers were exactly the same, this was classified as the final diagnosis. As this was not conclusive in one third of the scans, the final diagnosis in these was formed after a group consensus, and at this stage all eventual later examinations were also reviewed in order to strengthen the conclusion. As this was not conclusive in one third of the scans, the final diagnosis in these was formed after a group consensus, and at this stage all eventual later examinations were also reviewed in order to strengthen the conclusion.

8. Results Results Subdural haemorrhages did not cause any difficulty, even on-call residents found them accurately. Subdural haemorrhages did not cause any difficulty, even on-call residents found them accurately. Brain contusions were more difficult to detect; on-call residents missed 70 % of these. Brain contusions were more difficult to detect; on-call residents missed 70 % of these. Residents also missed some intraventricular and subarachnoidal haemorrhages and oedema; concerning these findings, their accuracy was moderate. Residents also missed some intraventricular and subarachnoidal haemorrhages and oedema; concerning these findings, their accuracy was moderate.

9. Results, continued Results, continued Practically all of the residents mistakes were false-negatives . Practically all of the residents mistakes were false-negatives . The reports of the two experienced neuroradiologists (NR1, NR2) differed significantly. The reports of the two experienced neuroradiologists (NR1, NR2) differed significantly. A neuroradiologist in training (NR3) was placed between the two. A neuroradiologist in training (NR3) was placed between the two.

10. Results, continued Results, continued NR1 made very few random errors. NR1 made very few random errors. NR2 did not miss any of the contusions, and missed only few subarachnoid haemorrhages, but also had a substantial number of false- positive findings on brain contusions. NR2 did not miss any of the contusions, and missed only few subarachnoid haemorrhages, but also had a substantial number of false- positive findings on brain contusions. NR3s accuracy did not reach NR1s, but it was more consistent than NR2s. Contusions were more difficult for NR3; both false-positive and negative findings occurred. NR3s accuracy did not reach NR1s, but it was more consistent than NR2s. Contusions were more difficult for NR3; both false-positive and negative findings occurred.

11. Results, continued Results, continued In four patients (= 13 % of positive scans), the CT-scan was reported normal, although it showed acute intraparenchymal traumatic lesions In four patients (= 13 % of positive scans), the CT-scan was reported normal, although it showed acute intraparenchymal traumatic lesions In retrospective analysis, this misinterpretation did not seem to influence the recovery In retrospective analysis, this misinterpretation did not seem to influence the recovery

12. Discussion Discussion Many signs of brain trauma are difficult to interpret in CT Many signs of brain trauma are difficult to interpret in CT Experience helps, but even among the most experienced there are marked differences. Experience helps, but even among the most experienced there are marked differences.

13. Conclusion Conclusion The low reliability of CT diagnosis (especially concerning other than neuroradiologists interpretations) should be taken into account in diagnostic and treatment decisions concerning acute head injuries The low reliability of CT diagnosis (especially concerning other than neuroradiologists interpretations) should be taken into account in diagnostic and treatment decisions concerning acute head injuries The diagnosis of acute TBI or the level of care and follow-up must not be based solely on cranial CT imaging The diagnosis of acute TBI or the level of care and follow-up must not be based solely on cranial CT imaging