This Week in Radiology — May 28, 2026

Welcome to This Week in Radiology. This week we're covering 10 notable papers spanning quality and practice management, advances in neuroradiology, and innovations in CT and hybrid imaging. Let's dive in.

Our first theme focuses on quality, value, and the future of radiology practice. As our field incorporates increasingly complex technologies, defining and measuring our impact becomes paramount.

In the American Journal of Roentgenology, the ACR's Relevance and Impact Committee proposes a new framework for tracking patient-oriented outcomes [5]. They introduce two core concepts: 'diagnostic imaging provenance,' which is the lineage of how a diagnosis is made, and 'relevance,' which connects that diagnosis to the reasons for the exam and downstream decisions. The goal is to create a 'medical imaging life cycle' feedback loop, using data science to link imaging findings directly to patient outcomes like morbidity and mortality, thereby demonstrating radiology's role in high-value care.

Complementing this high-level framework, a paper in the American Journal of Neuroradiology provides a practical guide to quality management practices [10]. It breaks down the key concepts of Quality Control, or QC, which is the reactive detection of defects; Quality Assurance, or QA, which is the proactive prevention of defects through standardized processes; and Quality Improvement, or QI, which uses data to adapt and improve both QC and QA over time. These principles, borrowed from manufacturing, provide a structured approach to achieving consistent excellence in patient care.

Putting these quality principles into practice, a multidisciplinary panel offers consensus recommendations for medical 3-D printing within healthcare facilities, published in the Journal of the American College of Radiology [1]. With the use of patient-specific anatomic models and surgical guides expanding, these 25 consensus statements provide a contemporary reference for establishing safe, consistent, and sustainable 3-D printing workflows, covering everything from quality and safety to economics.

However, a paper in European Radiology sounds a crucial cautionary note regarding the quality control of emerging AI tools [7]. Investigators tested several large language models, including GPT-4o, Mistral Large, and others, on neuroradiology scenarios where United States and international guidelines conflict. They found a profound and systematic bias toward United States guidelines. Even when models were explicitly prompted to follow a non-United States guideline, or when the model itself was developed in France or China, they defaulted to the American recommendation in 90% of cases. This raises significant concerns for safe global deployment. The authors found the most effective mitigation strategy was providing the full text of the relevant local guideline within the prompt, which restored accuracy to over 90%.

Next, we turn to several important updates in neuroradiology, spanning surgical decision-making, endovascular therapy, and molecular imaging.

The New England Journal of Medicine published a multicenter, cluster-randomized trial addressing a key question in the management of children with Chiari I malformation and syringomyelia [2]. The study compared posterior fossa decompression with duraplasty—that is, incising the dura and placing a graft—to decompression alone.

The Study One hundred sixty-two patients under 21 years old were enrolled across 38 centers, with each center randomized to perform only one type of procedure. The primary outcome was the rate of surgical complications within six months.

Results The complication rate was 14% in the duraplasty group versus 6% in the decompression-alone group. This difference was not statistically significant, with an adjusted odds ratio of about 2.6 that had a wide confidence interval. However, secondary outcomes at 24 months showed some interesting trends. Clinical improvement occurred in 58% of the duraplasty group versus 46% of the decompression-alone group. Syrinx reduction was also greater with duraplasty, at a mean of 3.1 millimeters versus 1.2 millimeters. Conversely, the need for repeat decompression was much lower with duraplasty, at 3% versus 14%.

Conclusions The authors conclude that the rate of surgical complications did not differ significantly between the two groups, and that larger trials are needed to clarify the relative benefits and risks. For radiologists who diagnose and follow these patients, this study provides important context for surgical discussions, suggesting that while duraplasty might not have a statistically significant increase in short-term complications, it may offer better long-term syrinx reduction and a lower chance of reoperation.

In the endovascular space, the INSTANT study, published in the American Journal of Neuroradiology, provides an update on the safety of aneurysm coiling [8]. This prospective European study followed 294 patients treated with the OPTIMA coil system. Critically, there were zero deaths or adverse events causing morbidity that were related to the device or the procedure itself. In the unruptured aneurysm group, mortality at 30 days was zero, and morbidity was just 1.2%. In the ruptured aneurysm group, all mortality was related to the subarachnoid hemorrhage and its complications, not the treatment.

Finally, a paper in Science Translational Medicine reports a major advance in molecular imaging for neurodegenerative disease [3]. Researchers developed and validated a new PET tracer, called [C]MODAG-005, that targets alpha-synuclein aggregates. In vitro and animal models showed high affinity and specificity for these pathological proteins. Most importantly, the first-in-human imaging demonstrated that the tracer binds in brain regions affected by alpha-synuclein pathology in patients with Multiple System Atrophy and Parkinson's disease. This could be a revolutionary tool for early diagnosis, monitoring, and evaluating new therapies for synucleinopathies.

Our final theme covers innovations in CT and hybrid imaging, from new hardware to novel applications of existing technology.

An invited commentary in the American Journal of Roentgenology looks to the future with Photon-Counting Detector CT [4]. While the piece is a perspective and doesn't present new data, it highlights how this technology is viewed as not just an incremental update, but a potential paradigm shift for computed tomography.

In European Radiology, a multicenter study uses artificial intelligence to analyze non-contrast chest CTs, providing new insights into the pulmonary vasculature in patients with and without COPD [9]. Among nearly 700 participants, researchers found that the presence of emphysema was associated with a distinct pattern of vascular remodeling: lower volumes of distal small pulmonary vessels, and higher volumes of more proximal vessels. This suggests pruning of the small vessels. Notably, this pattern was present even in control patients who had emphysema but normal lung function. The findings reinforce that emphysema acts as an 'accomplice' in vascular damage, potentially explaining its link to cardiovascular disease, and suggests that these changes can be identified noninvasively on standard chest CTs.

Lastly, a study in the Journal of the American College of Radiology evaluates the cost-effectiveness of using integrated FDG PET/MR with low-dose chest CT for the initial staging of locally advanced breast cancer [6]. Compared to a conventional pathway of contrast-enhanced CT, bone scan, and breast MRI, the integrated PET/MR strategy was significantly more accurate. It improved correct staging for distant metastases from 70% to 97% and reduced false positives. While the upfront imaging cost was higher, the incremental cost-effectiveness ratio was favorable, at about $3,000 per additional patient correctly staged. The authors conclude that this approach is economically attractive for patients with presumed stage III disease and may improve diagnostic efficiency while reducing downstream costs.

If you only have time for one paper this week, make it the New England Journal of Medicine trial on surgical decompression for Chiari I malformation with syringomyelia [2]. It provides the highest level of evidence to date on the contentious issue of whether to add a duraplasty, informing conversations that radiologists frequently have with neurosurgical colleagues and patients.

Here are the key takeaways from this week in Radiology.

First, for pediatric patients with Chiari I and syringomyelia, adding a duraplasty to posterior fossa decompression did not significantly increase short-term complications but was associated with better syrinx reduction and a much lower rate of reoperation at 24 months. The evidence continues to evolve.

Second, be aware that large language models exhibit a strong United States-centric bias. When using them for clinical decision support outside the United States, safe implementation requires providing the full text of local guidelines to ensure recommendations are appropriate.

Third, for initial staging of locally advanced breast cancer, an integrated FDG PET/MR and low-dose chest CT pathway is more accurate and cost-effective than conventional multi-modality staging, significantly reducing misclassification.

Fourth, emphysema seen on non-contrast chest CT is associated with pruning of small pulmonary vessels, a marker of vascular remodeling linked to cardiovascular risk. This finding is present even in individuals with preserved lung function.

Fifth, formal quality management frameworks—encompassing Quality Control, Quality Assurance, and Quality Improvement—are essential for integrating new technologies like 3D printing and AI, and for systematically demonstrating radiology's value in patient care.

That's your roundup for This Week in Radiology. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.