Project information
- Involvement:
Setting up Cloud infrastructure, MRI image preprocessing and quality control pipeline implementation and deployment, Deep Learning pipeline debugging and implementation, AI backend system deployment - Technologies:
Python, Pandas, Numpy, Tensorflow, Keras, FreeSurfer, MRIQC, Docker, AWS - WebApp link: Neurofind
Helped a world-renowned research institute implement and deploy Deep Learning models as a cloud-based user-friendly webapp
The challenge
A world-renowned research institute approached us to help them implement and deploy a Deep Learning-based brain age estimator as a cloud-based webapp easily usable by doctors and researchers. The complex AI backend system contains additional MRI image preprocessing and quality control pipelines on top of the Deep Learning models, and the entire system needs to be deployed on the cloud.
The solution
We took over this project from the client and started from setting up the cloud computing infrastructure, to the final deployment of the AI backend system, fulfilling all the goals and requirements from the client.
The end result
The finished web-app is a web-based user-friendly tool for detecting and quantifying neuroanatomical abnormalities in an individual person from structural Magnetic Resonance Imaging (MRI) scans. It can be useful for researchers investigating the neuroanatomical correlates of a wide range of traits or disorders.With a user account, you can upload a structural MRI scan of a person along with basic demographic information such as age, gender and ethnicity. Within a few hours, Neurofind generates an individualised report including an estimate of the person's brain age compared to their chronological age as a potential marker of general brain health; the extent to which the person's brain structure differs from the brains of a reference cohort of disease-free people and the location and extent of the differences in brain structure from the reference cohort.