Authors: Elena Doering, Merle C. Hönig, Tobias Deußer, Gerard N. Bischof, Thilo van Eimeren, Alexander Drzezga, Lotta M. Ellingsen

Published on: February 06, 2024

Impact Score: 8.05

Arxiv code: Arxiv:2402.04299

Summary

• What is new: A predictive image-to-image translation model that forecasts future brain scans, using early data to predict Alzheimer’s disease progression.
• Why this is important: Limitations in using repeated brain scans for Alzheimer’s diagnosis due to radiation, cost, and patient burden.
• What the research proposes: Development of a model that uses convolutional neural network architecture with long-short term memory to predict future brain metabolism from initial scans.
• Results: High accuracy in predicting future brain scans up to seven years post-baseline, particularly for individuals with mild cognitive impairment.

Technical Details

Technological frameworks used: Convolutional Neural Network (CNN) with Long-Short Term Memory (LSTM)

Models used: Predictive Image-to-Image Translation (I2I)

Data used: [$^{18}$F]-FDG PET data from 161 individuals

Potential Impact

Medical imaging, Alzheimer’s disease diagnostics, healthcare providers, and companies developing treatments for cognitive impairments.

Want to implement this idea in a business?

We have generated a startup concept here: NeuroForecast AI.