Accuracy of non-medical and medical individuals in identifying cerebral cortical abnormality from three-dimensional printed models of magnetic resonance images in children with hypoxic ischemic injury

Abstract

Effective communication of imaging findings in term hypoxic ischemic injury to family members, non-radiologist colleagues and members of the legal profession can be extremely challenging through text-based radiology reports. Utilization of three-dimensional (D) printed models, where the actual findings of the brain can be communicated via tactile perception, is a potential solution which has not yet been tested in practice. We aimed to determine the sensitivity and specificity of different groups, comprising trained radiologists, non-radiologist physicians and non-physicians, in the detection of gross disease of the cerebral cortex from 3-D printed brain models derived from magnetic resonance imaging (MRI) scans of children. Ten MRI scans in children of varying ages with either watershed pattern hypoxic ischemic injury (cortical injury) or basal-ganglia-thalamus hypoxic ischemic injury pattern with limited perirolandic cortical abnormalities and 2 normal MRI scans were post processed and 3-D printed. In total, 71 participants reviewed the 12 models and were required to indicate only the brain models that they felt were abnormal (with a moderate to high degree of degree of confidence). The 71 participants included in the study were 38 laypeople (54%), 17 radiographic technologists (24%), 6 nurses (8%), 5 general radiologists (7%), 4 non-radiologist physicians— 3 pediatricians and 1 neurologist (6%) and 1 emergency medical services staff (1%). The sensitivity and specificity for detecting the abnormal brains of the 71 participants were calculated. Radiologists showed the highest sensitivity (72%) and specificity (70%). Non-radiologist physicians had a sensitivity of 67.5% and a specificity of 75%. Nurses had a sensitivity of 70% and a specificity of 41.7%. Laypeople (non-medical trained) had a sensitivity of 56.1% and a specificity of 55.3%. Radiologists’ high sensitivity and specificity of 72% and 70%, respectively, validates the accuracy of the 3-D-printed models in reproducing abnormalities from MRI scans. The non-radiologist physicians also had a high sensitivity and specificity. Laypeople, without any prior training or guidance in looking at the models, had a sensitivity of 56.1% and a specificity of 55.3%. These results show the potential for use of the 3-D printed brains as an alternate form of communication for conveying the pathological findings of hypoxic ischemic injury of the brain to laypeople.