AI-Aided Brain Volumetry Analysis Demonstrates Accuracy in Identifying Frontotemporal Dementia and Its Subtypes

The use of the VUNO-Med DeepBrain (VUNO, Seoul, South Korea) automated artificial intelligence (AI)–aided brain volume analysis software enabled the detection of frontotemporal dementia (FTD), as well as the identification of distinct FTD subtypes based on the extraction of quantitative data from MRI scans. The results, which were presented at the 2024 meeting of the Alzheimer’s Association International Conference (AAIC), have implications for the clinical diagnosis of FTD, which has posed a challenge historically. In 2023, VUNO Med Deep-Brain was granted 510(k) clearance by the Food and Drug Administration (FDA).

The study included a total of 385 participants from Neuroimaging in Frontotemporal Dementia (NIFD) initiative datasets with available MRI, who were categorized as having the behavioral variant (BV; n=57), semantic variant (SV; n=40), and progressive nonfluent aphasia (PNFA; n=34) FTD subtypes, with the remaining participants comprising a Normal Control (NC) group (n=127). Researchers conducted comparative analysis using VUNO-Med DeepBrain to identify regional volume distribution differences in 104 areas of the participants’ brains.

VUNO-Med DeepBrain demonstrated accuracy in distinguishing between:

• Participants with FTD vs those in the NC group (95% CI, .88 to .89)
• Participants with BV vs SV vs PNFA (95% CI, .79 to .81)
• Participants with BV vs SV vs PNFA vs NC (95% CI, .82 to .83)

Additionally, significant brain volume differences were identified across the FTD subtypes:

• Compared with the SV, PNFA, and NC groups, participants with BV demonstrated anterior-posterior gradient in the frontal lobe (P<.01).
• Participants with SV showed left-side dominant asymmetric atrophy, especially in the inferior temporal lobe, as compared with the other 3 groups (P<.01)

"Our breakthrough FDA-Cleared AI technology drives real world clinical value across radiology and neurology workflows and contributes to improving the quality of patient care," said Rob Damiani, Director of Business Development at VUNO. "Patients and neurologists appreciate our easy to understand, customizable reports. We think that  is critical to offer a user friendly, affordable, value based solution with flexible deployment options, (On-Premise or Cloud), that will provide patients with the progression of their neurodegenerative status or disease."

The authors of this study are affiliated with VUNO and the Gangnam Severance Hospital of Yonsei University College of Medicine.