Artificial Intelligence-Based Diffraction Analysis (AIDA) for Point-of-Care Breast Cancer Classification
Abstract
The overall goal of this project is to advance the next generation imaging cytometer, AIDA (Artificial Intelligence-based Diffraction Analysis), for automated molecular screening on individual cancer cells. AIDA will integrate cutting-edge developments in computational optics and machine learning: digital diffraction imaging and deep neural network. We will implement an AIDA imaging system equipped with multiple light sources with different wavelengths. This setup will allow us to detect different molecular markers through color-based multiplexing. We will implement a deep-learning framework for cellular analyses. Specifically, we will train deep neural networks to i) recognize individual cells directly from diffraction images, ii) extract levels of molecular information, and iii) unravel hidden phenotypes for cell stratification. Then, we will perform a clinical study to test AIDAs clinical utility. Cellular samples will be obtained from breast cancer patients and will be color-stained for triple markers: HER2, ER/PR. We will then apply AIDA to image a large number of individual cells and automatically extract their features; these data will be used to construct the molecular profile of a given sample.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 2021
- Accession Number
- AD1153338
Entities
People
- Cesar M Castro
- Hakho Lee
- Kwonmoo Lee
- Michelle Specht