Radiology in public focus

Brain Iron on MRI Predicts Cognitive Impairment, Decline

A special MRI technique that measures brain iron levels may help predict the onset of mild cognitive impairment and cognitive decline in cognitively unimpaired older adults, according to a study published in Radiology.

Excess iron accumulation in the brain can contribute to neurodegeneration by inducing oxidative stress, exacerbating amyloid toxicity, disrupting tau protein function and ultimately leading to neuronal death. Recent research links elevated brain iron levels and Alzheimer’s disease.

A team led by Xu Li, PhD, associate professor of radiology at Johns Hopkins University and research associate at the F.M. Kirby Research Center for Functional Brain Imaging at the Kennedy Krieger Institute in Baltimore, used quantitative susceptibility mapping (QSM) MRI to evaluate 158 cognitively unimpaired participants in the Johns Hopkins BIOCARD Study, following them for up to 7 ½ years.

They found that higher baseline magnetic susceptibility on MRI in the entorhinal cortex and putamen—two brain regions important to memory and other cognitive functions—was associated with a higher risk of mild cognitive impairment.

“QSM can detect small differences in iron levels across different brain regions, providing a reliable and non-invasive way to map and quantify iron in patients, which is not possible with conventional MR approaches,” Dr. Li concluded.

Read the related RSNA News story.

Molecular Breast Imaging May Benefit Women with Dense Breasts

Research published in Radiology showed that adding molecular breast imaging (MBI) to digital breast tomosynthesis (DBT) significantly improves invasive cancer detection in women with dense breasts, with only a modest increase in recall rates compared to DBT alone.

MBI is considered a safe, relatively inexpensive option for supplemental screening. The Density MATTERS Trial, led by Carrie B. Hruska, PhD, professor of medical physics at Mayo Clinic in Rochester, MN, enrolled 2,978 women aged 40–75 across five sites from 2017 to 2022. Participants underwent two annual screening rounds with DBT and MBI. Most were postmenopausal and 82% had category C density, indicating heterogenous density.

MBI detected 30 cancers missed by DBT, 71% of which were invasive (median size 0.9 cm), and 90% were node negative. MBI added 6.7 cancers per 1,000 screenings in the first year and roughly 3.5 per 1,000 in the second year. Node-positive detection also improved, with DBT alone finding 57% of such cases in year one, while DBT plus MBI found 100%.

“Someone who’s having their routine annual screen every year should not be diagnosed with advanced breast cancer,” Dr. Hruska said. “That’s just unacceptable. With a supplemental screening every few years, we hope to find cancers earlier and see the diagnosis of advanced cancer go way down.”

Read the related RSNA News story.

AI Hybrid Strategy Improves Mammogram Interpretation

A hybrid reading strategy for screening mammography, deployed retrospectively to more than 40,000 exams, reduced radiologist workload by 38% without changing recall or cancer detection rates, according to a study published in Radiology.

“AI models perform well but sometimes make mistakes,” said Sarah D. Verboom, MSc, a doctoral candidate in the Department of Medical Imaging at Radboud University Medical Center in the Netherlands. “Identifying exams where AI is uncertain is crucial.”

Combining radiologist review with stand-alone AI interpretation for cases where the model is confident, Dr. Verboom and colleagues analyzed 41,469 mammograms from 15,522 women in the Dutch National Breast Cancer Screening Program. AI assessed each exam for probability of malignancy and uncertainty. Confident low-risk cases were classified as normal; confident high-risk cases triggered recall. Uncertain cases underwent review by two radiologists.

Performance of the model matched standard double-reading, and when AI was certain, AUC reached 0.96. Sensitivity nearly matched radiologists (85.4% vs 88.9%).

“The key component of our study is the value of uncertainty quantification,” Dr. Verboom said. “It could help address workforce shortages and build trust in AI implementation.”

Read the related RSNA News story.