Multiparametric MRI finds more diabetic neuropathy than single parameter

CHICAGO -- Multiparametric MRI that incorporates diffusion tensor imaging (DTI) and T2 mapping is more accurate for identifying diabetic peripheral neuropathy than single parameter MRI, according to research shared December 3 at the RSNA meeting.

The study results could "enhance noninvasive evaluation of peripheral nerve integrity in diabetic populations," said presenter Eunsun Oh, MD, PhD, of Soonchunhyang University Seoul Hospital in South Korea.

Diabetic peripheral neuropathy affects up to 50% of people with type 2 diabetes. It tends to be diagnosed with via a nerve conduction study or by electromyography, both of which can be painful for patients and carry low sensitivity, Oh noted.

"We need a noninvasive, sensitive biomarker," she said.

Oh's group conducted research that included 17 diabetic patients with clinically confirmed peripheral neuropathy, 16 diabetic patients without neuropathy, and 12 healthy controls. All underwent knee MRI that included axial T2-weighted Dixon imaging, diffusion tensor imaging (DTI), and T2 mapping sequences. The team measured fractional anisotropy (FA), axial diffusivity (AD), radial diffusivity (RD), mean diffusivity (MD), and T2 relaxation time at the tibial nerve at the calf level, and assessed MRI's performance using area under the receiver operating curve (AUC) analysis.

The investigators reported the following:

• Fractional anisotropy values of the tibial nerve were significantly lower in both diabetic groups (those with and without diabetic peripheral neuropathy) compared to healthy controls (p
• T2 relaxation times were significantly longer in both diabetic groups than in controls (p
• Among the single-parameter models, fractional anisotropy showed the highest diagnostic performance for detecting diabetic peripheral neuropathy (AUC = 0.85).
• The multiparametric model combining fractional anisotropy and T2 relaxation time demonstrated improved diagnostic accuracy (AUC = 0.91), outperforming all single-parameter models.
• Axial diffusivity, radial diffusivity, and mean diffusivity showed lower AUC values and were less effective as standalone markers.

Oh noted that FA reduction occurs early in the disease process, and "indicates microstructural axonal loss even in asymptomatic (subclinical) stages."

The takeaway? "Combining structural (FA) and pathological (T2) markers yields the highest diagnostic accuracy," Oh concluded.

Visit our RADCast for full coverage of RSNA 2025.