Radiologists read scans in the first-in-first-out (FIFO) order or based on the 'stat' marker based on clinical indications. This might be suboptimal because scans with critical findings might be on bottom of the work list.
Artificial intelligence(AI) / deep learining algorithms have been built which can detect scans with emergency findings like hemorrhages. midline shift or fracture[1, 2] from head CT scans.
If these algorithms are used to preread the head CT scans and prioritize scans with critical findings, it can lead to an improvement in turn around time for critical scans. In this study, we demonstrate that such an reordering of radiologist work-list will improve read times in a simulated environment.
We considered a scan critical if it has any of intracranial hemorrhage, fracture, mass effect and/or midline shift.
We ran deep learning algorithms which can detect intracranial hemorrhages, skull fracture, midline shift and mass effect on a training dataset of 21095 scans. Predictions of these algorithms were used as features and we trained a random forest model to predict criticality of scan. The probability score output from this random forest model is used to assign a new scan a criticality score in 0-1.
CQ200 dataset consisted of 214(41[19%] critical) scans. Simulated trial showed that radiologists would take significantly lower time to read critical scans in algorithmically reordered queue compared to usual FIFO queue (9 mins vs 32 mins, p<0.01).
Deep learning can successfully reorder radiologists' work-list and this will significantly decrease time to read critical scans.
Link to complete publication here: https://scholar.google.com/citations?view_op=view_citation&hl=en&user=dqpMNRUAAAAJ&cstart=20&pagesize=80&citation_for_view=dqpMNRUAAAAJ:YsMSGLbcyi4C
