AI-powered automation is moving upstream, shifting the focus from just lesion detection to the very beginning of the imaging process: acquisition and post-processing. Most imaging costs are tied to these upstream tasks. By automating them, Linda Moy, MD, explained at ARRS 2026, providers can improve image quality while simultaneously lowering costs and increasing operational efficiency…
- On the DL: Deep learning (DL) reconstruction tools allow for noisier or lower-resolution initial imaging that AI then transforms into high-quality images. DL algos can reduce MRI scan times by >50%—while maintaining, or even improving, diagnostic quality.
- Keep the Noise Down: AI denoising algorithms have demonstrated a 60% reduction in radiation dose, alongside a 39% reduction in image noise.
- Table for 3D? Using three-dimensional ceiling cameras and infrared imaging to automate table height and positioning can save 32 seconds per exam, as well as reduce radiation dose by some 20%.
- No Contrast: Synthetic AI images can now simulate contrast-enhanced MRI, without actual contrast agents. This is a breakthrough for pregnant patients, those with severely impaired renal function, or populations concerned about gadolinium deposition.
These advancements significantly improve the patient experience, particularly for peds patients who may no longer require sedation, and claustrophobic patients who benefit from shorter time in the scanner.
Mind the Mirage: Rads must remain vigilant regarding AI “hallucinations.” While rare, these tools can occasionally invent lesions that do not exist or remove true lesions during the reconstruction process.
Bottom Line: AI reconstruction works seamlessly in the background to provide higher signal-to-noise ratios and reduced artifacts, allowing rads to focus on downstream diagnostic tasks with better data in less time.
Leave a Reply