MINDS Lab

Projects

Goal

The project aims to develop machine learning algorithms to analyze multi-modal neuroimaging data with the goals of better understanding healthy aging and dissecting Alzheimer’s disease heterogeneity.

The project is supported by NIH R01 AG067103.

These patterns of coordinated amyloid deposition were estimated using unsupervised machine learning.
These patterns of coordinated amyloid deposition were estimated using unsupervised machine learning.

Goal

The aim of this project is to develop deep learning techniques to identify individuals at early stages of Alzheimer’s disease and predict their future cognitive performance.

This figure depicts the positive predictive value of machine learning models trained with different feature combinations and evaluated across different datasets.
This figure depicts the positive predictive value of machine learning models trained with different feature combinations and evaluated across different datasets.

This project is supported by BrightFocus award A2021042S.

Goal

The goal of this project is to use machine learning models to investigate structural and functional brain correlates of late life depression.

The project has received support by the McDonnell Center for Systems Neuroscience.

Late-life depression was significantly associated with reduced volume within this structural network, which was identified by unsupervised machine learning on gray-matter tissue density maps from subjects of the UK Biobank study.
Late-life depression was significantly associated with reduced volume within this structural network, which was identified by unsupervised machine learning on gray-matter tissue density maps from subjects of the UK Biobank study.

Goal

By augmenting the process with artificial intelligence, this project aims to build deep learning techniques that will identify whether a recommendation for a follow-up exam has been made based on incidental imaging findings.

We use XNAT, anĀ imaging informatics software platform, to get image data and radiology reads from the clinic. Then the reports are processed using an artificial intelligence model that we have developed, which aims to automatically identify whether a recommendation for a follow-up has been made. These cases are flagged and delivered to a nurse coordinator, who is tasked with contacting the patient and primary care to ensure that follow-up is taking place.
We use XNAT, an imaging informatics software platform, to get image data and radiology reads from the clinic. Then the reports are processed using an artificial intelligence model that we have developed, which aims to automatically identify whether a recommendation for a follow-up has been made. These cases are flagged and delivered to a nurse coordinator, who is tasked with contacting the patient and primary care to ensure that follow-up is taking place.

The project has received support by the Big Ideas Program.

Our People

The lab, led by Aristeidis Sotiras, PhD, is powered by a team of researchers developing unique computer algorithms and machine learning techniques to better understand brain development, brain aging, and brain tumor segmentation.

Latest in the CIRC