Is image selection a useful strategy to decrease the transmission time in teleradiology? A study using 100 emergency cranial CTs

This study examines the suitability of working with a selection of images in a teleradiology consulting system in neurological or neurosurgical emergency situations. The teleradiology system was based on IBM-compatible personal computers, video digitization for data acquisition, and data transmission by Integrated System Digital Network. Forty normal and 60 abnormal emergency cranial computed tomograms were shown to a radiologist on call who presented all cases he regarded as pathologic to a neuroradiologic expert by teleradiology. To reduce transmission time, only a selection of images from the CT study was presented (up to four images per case). For each case the on-call radiologist's diagnosis (Don-call), the expert's diagnosis on the teleradiology screen (Dmonitor), and the expert's diagnosis on the original film (Doriginal) was documented, together with an estimation of the agreement between those diagnoses. There was clinically relevant disagreement between the on-call radiologist's diagnosis and the neuroradiologist's diagnosis based on the image selection on the teleradiology monitor in 23 % of cases. A clinically important discrepancy between the neuroradiologist's diagnosis based on the image selection and his diagnosis using the original films was found in 30 % of cases. This was due to the presence of clinically relevant information on images not transferred by the on-call radiologist. Image quality of the transferred images was sufficient in all cases. Drastic selection of images from a complete CT study leads to a high rate of incorrect diagnoses and is not appropriate to reduce transmission time in teleradiology.