Radiomics models offered a pooled AUC of 86 percent for differentiating between ruptured and unruptured intracranial aneurysms, according to a recently published meta-analysis.
Adjunctive AI showed no difference in accuracy than unassisted radiologists for intracranial hemorrhage (ICH) detection and had a slightly longer mean report turnaround time for ICH-positive cases, according to newly published prospective research.
Providing an 85.7 percent sensitivity rate for detecting cerebral aneurysms on computed tomography angiography (CTA), the deep learning model also offered a similar AUC (93 percent) in comparison to radiology reports (91 percent).
While cone-beam computed tomography (CBCT) demonstrated high pooled sensitivity and specificity rates for intraparenchymal hemorrhage in the systematic review, researchers noted higher false-negative rates for subarachnoid and intraventricular hemorrhages.
The first radiology triage modality to garner a Breakthrough Device Designation from the FDA, Annalise-Obstructive Hydrocephalus has reported sensitivity and specificity rates of 97.5 percent and 95.3 percent respectively.