Deep learning algorithm for early diagnosis of Alzheimer's disease

This technology is an algorithm that utilizes deep learning tools to identify and diagnose diseases from MRI scans.

Unmet Need: Automated neuroimaging for detection of Alzheimer’s disease and preceding states

Therapeutic interventions at the early stages of Alzheimer’s disease (AD) have been shown to slow disease progression. Therefore, developing methods for early detection of Alzheimer’s disease is critical. Diagnostic modalities are limited and underdeveloped since MRI scans require visual inspection by trained medical professionals. Automated deep learning-based algorithms show great promise in their application to neuroimaging diagnostics, yet current deep learning MRI technologies are inaccurate in their ability to classify diseases.

The Technology: Application of deep learning-based neuroimaging modalities for accurate classification of Alzheimer’s disease and detection at early stages.

This technology is a machine learning tool that utilizes 3D deep convolutional neural networks for classification of Alzheimer’s disease and for detection of early-stage disease. The method is based on longitudinal data from structural MRI scans. The algorithm performs in-depth regional analysis to localize the most predictive areas, in addition to analyzing whole brain volumes. This technology can be used to classify states of mild cognitive impairment as well as predict the likelihood of progression to Alzheimer’s disease.

This technology was able to accurately classify patients that already had mild cognitive impairment, as well as predict the likelihood of progression to AD.

Applications:

• Automated deep learning-based MRI tool for classification of Alzheimer’s disease and detection of early-stage disease
• Automated tool for monitoring of patients for early-stage disease progression
• Research tool for development of machine learning-based imaging modalities for detection of
  other diseases

Advantages:

• Utilization of 3D deep convolutional neural networks
• In-depth regional analyses
• Increased predictive accuracy in detecting mild cognitive impairment

Lead Inventor:

Frank Provenzano, Ph.D.

Patent Information:

Patent Pending

Related Publications:

• Feng X, Lipton ZC, Yang J, Small SA, Provenzano FA. “Estimating brain age based on a uniform healthy population with deep learning and structural magnetic resonance imaging” Neurobiol Aging. 2020 Jul; 91: 15-25.

• Feng X, Guo J, Sigmon HC, Sloan RP, Brickman AM, Provenzano FA, Small SA. “Brain regions vulnerable and resistant to aging without Alzheimer’s disease” PLoS One. 2020 Jul 29; 15(7): e0234255.

• Feng X, Yang J, Lipton ZC, Small SA, Provenzano FA. “Deep Learning on MRI Affirms Prominence of the Hippocampal Formation in Alzheimer’s Disease Classification” bioRxiv. 2018; 456277.

Tech Ventures Reference:

• IR CU19273

• Licensing Contact: Jerry Kokoshka