ASL-MRI Shows Promise for Early Dementia Detection Without Radiation

A novel, non-invasive imaging technique may allow physicians to detect signs of dementia well before irreversible brain damage occurs, according to new research presented at RSNA 2025. The approach, which uses arterial spin labeling (ASL) MRI at 3 Tesla, has the potential to identify early changes in cerebral blood flow—offering a critical window for intervention.

“Dementia is frequently diagnosed after irreversible neuronal loss has already occurred,” said Shaheen Haidrus, MD, a third-year radiology resident at Subharti Medical College & Chhatrapati Shivaji Subharti Hospital in Meerut, India. “The need for accessible, radiation-free biomarkers in centers that lack PET or SPECT facilities served as motivation for our research.”

The study included 92 participants: 53 individuals clinically diagnosed with dementia based on DSM-5 criteria, and 39 age- and sex-matched healthy controls. Using ASL-MRI, a technique that magnetically labels water in the blood as a natural tracer, Dr. Haidrus and her team measured regional cerebral perfusion without the need for contrast agents or ionizing radiation.

What sets this approach apart is the automated pipeline developed by the team—one that extracts perfusion metrics and integrates machine learning tools to support early dementia classification. Designed to operate with minimal manual input, the system aims to make dementia screening both efficient and scalable.

Results from the study revealed a marked difference in cerebral blood flow (CBF) between the dementia group and the healthy controls. Patients with dementia showed an average CBF of 43 ml/100g/min, significantly lower than the 54 ml/100g/min seen in the control group. In addition to raw CBF values, patients also scored lower on normalized z-CBF values—another indicator of reduced perfusion.

Importantly, cerebral perfusion values were strongly correlated with cognitive scores on the Mini-Mental State Examination (MMSE), a widely used tool to assess memory and mental status. “We observed unexpectedly prominent frontal hypoperfusion in vascular and mixed dementias, often preceding visible white matter changes,” Dr. Haidrus noted. For Alzheimer’s disease, hypoperfusion was most pronounced in the frontal opercular cortex, lingual gyrus and temporal lobes.

The team’s AI classifier performed best in differentiating Alzheimer’s patients from healthy controls, a reflection, Dr. Haidrus said, of the balanced dataset for that particular subtype. “Our AI classifier also performed best for Alzheimer’s patients and controls, highlighting the importance of balanced data for rarer subtypes,” she said.

She emphasized the broader utility of ASL-MRI as a tool for both screening and follow-up. “ASL-MRI is a repeatable, contrast-free and cost-effective biomarker for screening and follow-up,” she said. “It facilitates early diagnosis, subtype differentiation and longitudinal monitoring without radiation exposure, making it especially valuable for resource-limited settings and follow-ups.”

Dr. Haidrus credited her team and mentors for guiding the research, including Dr. Mahesh Kumar Mittal, Dr. Vivek Kumar, Dr. Akanksha Singh and Dr. Vishal Vishnoi. Looking ahead, she hopes to see standardized ASL quantification integrated into vendor MRI workstations.

“Our goal is to standardize ASL quantification on vendor MRI workstations so that quantitative perfusion can be incorporated into routine dementia protocols,” she said.

The findings point to a future in which dementia could be identified earlier and monitored more effectively—without invasive procedures or radiation—opening doors to earlier and more personalized care.