The adoption of synthetic intelligence (AI) in medical imaging has skilled substantial progress lately, with greater than 450 AI-based medical units gaining approval in each Europe and the USA. As hospitals grapple with rising prices and workers shortages, AI has emerged as a pivotal device in making healthcare smarter and extra environment friendly. Nevertheless, in a panorama flooded with AI choices and affected person outcomes at stake, the importance of choosing the fitting AI answer is vital.
Consequently, the medical imaging group was rightfully intrigued when the Radiology Division from Stockholm South Emergency Hospital (Södersjukhuset, SÖS) unveiled findings at Rӧntgenveckan from a comparative evaluation of three well-known, commercially out there AI algorithms for the notification and triage of intracranial hemorrhage (ICH). Whereas one of many algorithms is Aidoc’s, the identities of the opposite two suppliers stay respectfully confidential. The soon-to-be-published examine revealed vital disparities within the efficiency of those algorithms, regardless of evaluation of the identical information. Aidoc’s algorithm demonstrated a 30% greater constructive predictive worth (PPV) and a 1.25% larger damaging predictive worth (NPV) when in comparison with the common outcomes of the opposite two algorithms.
The examine concerned the retrospective evaluation of 3902 non-contrast CT mind scans by Aidoc’s ICH algorithm run on its award-winning AI working system (aiOS™), and two different algorithms from distributors “A” and “B” deployed through a AI market PACS vendor. To make sure a good comparability, every vendor’s algorithm complied with the identical strategies for the evaluation of retrospective information, which meant that solely DICOM-related information from the PACS was analysed, with out further help from some other information sources. Comparative sensitivity, specificity, PPV and NPV outcomes for all three algorithms are proven under:
| Sensitivity | Specificity | PPV | NPV | |
|---|---|---|---|---|
| Vendor A | 60.2% | 97.1% | 48.4% | 98.2% |
| Vendor B | 62.8% | 97.4% | 52.4% | 98.3% |
| Aidoc | 90.3% | 99.0% | 80.3% | 99.5% |
For full particulars of this examine, please obtain the pre-publication report: