Effect of a Computer-aided Diagnosis System on Clinicians’ Performance in Detection of Small Acute Intracranial Hemorrhage on Computed Tomography

Rationale and Objectives

To analyze the effect of a computer-aided diagnosis (CAD) system on clinicians’ performance in detection of small acute intracranial hemorrhage (AIH) on computed tomography (CT).

Materials and Methods

The authors have developed a CAD scheme that used both image processing techniques and anatomic knowledge based classification system to improve diagnosis of small AIH on CT. A multiple-reader, multiple-case receiver operating characteristic (ROC) study was performed. Twenty clinicians, including seven emergency physicians, seven radiology residents, and six radiology specialists were recruited as readers of 60 sets of brain CT, including 30 cases that show AIH smaller than 1 cm, and 30 controls. Each reader read the same 60 cases twice, first without, then with the prompts produced by the CAD system. The clinicians ranked their confidence in diagnosing a case of showing AIH, which produced the ROC curves.

Results

Significantly improved performance is observed in emergency physicians, average area under the ROC curve (Az) increased from 0.8422 to 0.9294 (P = .0107) when they make the diagnosis without and with the support of CAD. Az for radiology residents increased from 0.9371 to 0.9762 (P = .0088). Az for radiology specialists increased from 0.9742 to 0.9868, but was statistically insignificant (P = .1755).

Conclusions

CAD can improve the clinicians’ performance in detecting AIH on CT. In particular, emergency physicians can benefit most from the CAD and improve their performance to a level approaching that of the average radiology residents.

Key Words: Observer performance study, receiver operating characteristic, computer-aided diagnosis, computed tomography, acute intracranial hemorrhage