Interview with Lauren Prince, Pathology Intern at CRL and Aiforia

Lauren Prince, an intern working for CRL and Aiforia, tells us about her time training neural networks to build AI models for toxicology studies at CRL.
Written by Aiforia

Tell us about your internship at Charles River Laboratories and Aiforia:

 

Lauren Prince

 

I am utilizing Charles River Laboratories’ pathology expertise and slides and incorporating Aiforia’s platform to build two deep-learning artificial intelligence (AI) models for CRL. 

Dr. Dan Rudmann from CRL is guiding and instructing me in pathology to successfully develop these models. I have also been employing support from Aiforia in advancing the models through sharing of extensive knowledge of the program. 

Describe your projects working with CRL and Aiforia:

Lauren Prince1The first is the rat ovarian follicle project. The objective of these studies is to design and train a convolutional neural network (CNN) that has the ability to identify, distinguish, and count different stages of ovarian follicles in rats. Two models will be developed-for H&E and PCNA-stained rat ovaries. This project is important in assessing the reproductive effects of drugs and other chemicals in rodent models.

The second study is looking at delayed-type hypersensitivity in skin. The objective of this study is to design and train a CNN to recognize characteristics of delayed-type hypersensitivity (DTH) lesions and then score the lesion severity in monkey skin. This model is crucial to impartially classify delayed-type hypersensitivity reactions in the skin.

What was it like using the Aiforia Platform?

Lauren Prince2After engaging in basic training, the Aiforia software is self-explanatory and quite easy to use. There are many advanced settings to understand and employ to further advance the model and the Aiforia employees were crucial in assisting me in understanding these parameters. 

The knowledge base and quick start guide were also extremely helpful in utilizing and understanding all the advantages and advanced settings of the platform.

What were your expectations working with an AI-based software?

I was expecting to need to learn about computer programming in order to use the AI-based software but was pleasantly surprised when I learned the opposite. With basic pathology training, I was able to develop and finalize two AI models. The Aiforia software is easy to use!

What do you think is the biggest benefit of using Aiforia for this type of pathology research?

These two projects were chosen to be developed using AI software because they are subject to high subjectivity, time requirements, and are laborious when employing manual counting.  The Aiforia models greatly improve speed and reproducibility of the assays.