Recent advancements in artificial intelligence have led to significant progress in the diagnosis of dementia types, specifically Alzheimer’s Disease (AD) and Frontotemporal Dementia (FTD). Researchers from the College of Engineering and Computer Science at Florida Atlantic University have developed a deep learning model capable of distinguishing between these two forms of dementia using electroencephalography (EEG) brainwave analysis. This innovative approach promises to enhance diagnostic accuracy and improve patient care.
Dementia encompasses various disorders that progressively impair memory, cognitive function, and daily activities. According to projections, approximately 7.2 million Americans aged 65 and older will be affected by Alzheimer’s Disease by 2025. While FTD is less common, it is the second leading cause of early-onset dementia, typically impacting individuals aged 40 to 60. The challenge lies in the fact that both diseases inflict damage on the brain but do so in distinct ways. Alzheimer’s primarily affects memory and spatial awareness, while FTD impacts behavior, personality, and language. This overlap can lead to misdiagnoses, making accurate differentiation essential for effective treatment and care.
Traditional imaging techniques such as MRI and PET scans are effective for diagnosing Alzheimer’s but can be costly, time-consuming, and require specialized facilities. In contrast, EEG offers a more accessible and noninvasive alternative by measuring brain activity through sensors detecting different frequency bands. Despite its advantages, EEG signals can be noisy and vary significantly among individuals, complicating the analysis process. Previous machine learning applications have struggled to consistently differentiate between AD and FTD using EEG data.
To address this issue, the research team at Florida Atlantic University has created a deep learning model that enhances the accuracy and interpretability of EEG readings. By analyzing both frequency and time-based brain activity patterns associated with each condition, the model improves diagnostic capabilities. The findings, published in the journal Biomedical Signal Processing and Control, reveal that slow delta brain waves serve as crucial biomarkers for both Alzheimer’s and FTD, primarily in the frontal and central brain regions.
The study indicates that brain activity in Alzheimer’s patients is more broadly disrupted, affecting additional regions and frequency bands, such as beta. This broader disruption accounts for the relative ease of detecting Alzheimer’s compared to FTD. The model achieved over 90% accuracy in distinguishing individuals with dementia from cognitively normal participants and successfully predicted disease severity with relative errors of less than 35% for AD and 15.5% for FTD.
The researchers employed feature selection techniques to enhance the model’s specificity, raising its ability to identify healthy individuals from 26% to 65%. Their two-stage design, which first detects healthy individuals and subsequently differentiates between AD and FTD, achieved an accuracy rate of 84%, placing it among the leading EEG-based diagnostic methods to date. By merging convolutional neural networks with attention-based long short-term memory (LSTM) networks, the model effectively detects both the type and severity of dementia from EEG data.
Using Grad-CAM visualizations, the model can identify which brain signals influenced its classifications, providing clinicians with valuable insights into the decision-making process. This approach offers a novel perspective on brain activity evolution and highlights specific regions and frequencies that contribute to diagnoses, a capability that traditional diagnostic tools often overlook.
“What makes our study novel is how we used deep learning to extract both spatial and temporal information from EEG signals,” stated Tuan Vo, the first author and a doctoral student in the FAU Department of Electrical Engineering and Computer Science. “By doing this, we can detect subtle brainwave patterns linked to Alzheimer’s and frontotemporal dementia that would otherwise go unnoticed. Our model doesn’t just identify the disease—it also estimates how severe it is, offering a more complete picture of each patient’s condition.”
The research findings also revealed that Alzheimer’s Disease tends to be more severe, impacting a wider range of brain areas and leading to lower cognitive scores. In contrast, FTD’s effects are more localized to the frontal and temporal lobes. These insights align with previous neuroimaging studies but add depth by illustrating how these patterns manifest in EEG data.
“Our findings show that Alzheimer’s disease disrupts brain activity more broadly, especially in the frontal, parietal, and temporal regions, while frontotemporal dementia mainly affects the frontal and central areas,” explained Hanqi Zhuang, a co-author and associate dean at FAU. “This difference explains why Alzheimer’s is often easier to detect. However, our work also shows that careful feature selection can significantly improve how well we distinguish FTD from Alzheimer’s.”
Overall, this study demonstrates that deep learning can streamline dementia diagnosis by integrating detection and severity assessment into a single system. This advancement reduces the time required for evaluations and provides clinicians with real-time tools to monitor disease progression. “This work demonstrates how merging engineering, AI, and neuroscience can transform how we confront major health challenges,” noted Stella Batalama, dean of the College of Engineering and Computer Science. “With millions affected by Alzheimer’s and frontotemporal dementia, breakthroughs like this open the door to earlier detection, more personalized care, and interventions that can truly enhance lives.”
The research team also included co-authors Ali K. Ibrahim, Ph.D., an assistant professor of teaching, and Chiron Bang, a doctoral student, both from the same department. For further information, please refer to the study titled, “Extraction and interpretation of EEG features for diagnosis and severity prediction of Alzheimer’s Disease and Frontotemporal dementia using deep learning,” published in Biomedical Signal Processing and Control in 2026.
