Skip to main content
U.S. flag

An official website of the United States government

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

Validating A.I. pipelines for analysis of live cell image data

Summary

Quantitative time lapse microscopy applied to the analysis of individual cells requires custom designed image analysis algorithms to segment and track single cells. This is a painstaking and labor-intensive process. Deep learning A.I. promises to dramatically change this paradigm by learning the appropriate models for segmentation, mitosis detection and tracking. Each quantitative live cell imaging pipeline is unique, depending the imaging system, cell type, culture condition, and other factors. Methods are needed to rapidly obtain training data and test and validate models. Once obtained, dynamic measurements from single cells can be used to inform next generation models for predicting responses from complex cellular systems. We anticipate that dynamic characterization of cells will help guide manufacturing processes for cellular products and therapies.

To fully realize the potential of deep learning A.I. for large-scale bioimage analysis, we are collaborating with computational scientists (link(s) to ITL) to develop strategies to build and deploy trusted A.l. analysis pipelines. Part of this work involves building and testing high speed imaging systems  for generating data at an appropriate acquisition rate that can test the limits of pre-trained models . When can we have confidence in the quantitative output of a deep learning A.I. model? Under what conditions does the model fail? We are also evaluating the effect of training data on A.I. model performance characteristics such as accuracy, reliability, robustness and bias. These systematic studies involve the acquisition of image datasets under varying conditions for the training and testing of A.I. models.

With trusted A.I. systems, fit-for-purpose cellular measurements of greater complexity than ever before are possible. 

Created June 9, 2021, Updated May 6, 2022