Technology

Catch up on the top AI-related news and research in radiology over the past month.

In recent interviews, Andrew Del Gaizo, MD, Nicholas Galante, MD and David Larson, MD, MBA, shared their thoughts on the rapidly increasing imaging volume in radiology, key ramifications and possible solutions.

Catch up on the top radiology content of the past week.

In a recent interview, Manisha Bahl, MD, discussed new breast ultrasound research findings that showed 16 percent higher specificity for breast cancer with adjunctive AI in comparison to unassisted interpretation.

AI software identified breast cancer missed by radiologists on one prior screening DBT exam in 26.8 percent of patients and on three prior screening DBT exams in 11 percent of patients, according to a study involving over 300 women that was presented at the Society of Breast Imaging conference.

In a recent interview with Diagnostic Imaging, Sujith Kalathiveetil, MD, FACC, discussed the evolution of coronary computed tomography angiography (CCTA), AI-enabled tools and the potential for future innovations with CCTA.

Are we overestimating the capacity of AI to address the ambiguity inherent to real-world medicine?

PlaqueSegNet, an emerging deep learning model for coronary plaque quantification based off CCTA exams, offered a greater than 90 percent intraclass correlation coefficient agreement with intravascular ultrasound (IVUS) and expert readers across four separate datasets.

In a recent interview, Andrew Del Gaizo, MD, and Nicholas Galante, MD, shared their thoughts and perspectives on increasing imaging volume, the sunsetting of PowerScribe 360, AI hype and promise, and potential opportunities in radiology.

Facilitating real-time visualization of deep vessel structures and high-resolution surface detail within a single workflow, the hybrid intravascular ultrasound (IVUS)/optical coherence tomography (OCT) platform will be launched in the second quarter of 2026.

Catch up on a variety of new FDA clearances in radiology from the past week.

Catch up on the top radiology content of the past week.

Key features with version 1.2 of the 3D mammography algorithm include selectable thresholds for navigating sensitivity and specificity tradeoffs with AI.

The Bunkerhill Contrast CAC and Bunkerhill Contrast AVC are reportedly the only AI-powered software modalities to garner 510(k) clearance for detection and quantification of coronary artery calcium and aortic valve calcium on contrast-enhanced CT scans.

In a recent interview, Connie Lehman, MD, PhD, discussed key changes to the National Comprehensive Cancer Network (NCCN) guidelines on breast cancer screening, including the use of AI-powered five-year risk assessment beginning at the age of 35.

The Philips Spectral CT Verida system reportedly provides reconstruction of 145 images per second, allowing entire CT exams to appear in less than 30 seconds.

Employing the SwiftMR software with deep learning reconstruction solutions from original equipment manufacturers (OEMs) may provide additional sequence coverage as well as scan time reduction.

In recent interviews with Diagnostic Imaging, Andrew Del Gaizo, MD, Nick Galante, MD and Marc Succi, MD, shared their perspective on the recent comment from NYC Health and Hospitals CEO Mitchell Katz, MD, about potentially replacing radiologists with AI.

In a recent interview, Marc Succi, MD, discussed findings from a new study examining the clinical reasoning capabilities of 21 large language models (LLMs), including GPT-5, Grok 4 and Claude 4.5 Opus.

In the latest episode of her “Breast Imaging in Focus” series, Manisha Bahl, MD, discusses key findings from a new study looking at partially autonomous AI-supported screening with mammography and digital breast tomosynthesis (DBT).

Catch up on the top radiology content of the past week.

In a recent interview, Lior Fisher, MD, discussed new research findings for AI software that may enable single view ultrasound detection of valvular and ventricular dysfunction by non-cardiologists.

The neuropacs™ MRI-based AI software provided a 96 percent or better AUROC for differentiating between Parkinson’s disease, atypical parkinsonism, multiple system atrophy Parkinsonian variant and progressive supranuclear palsy, according to a prospective multicenter study published in JAMA Neurology.

Catch up on the top radiology content of the past week.