Rewire ai


Programmed Intelligence Procedure to Standardize & Quantify Empirical Analysis Kit

Artificial intelligence is paving the way to better and faster cures at Washington State University.


Laboratory data analysis can be fast or accurate, but a team at Washington State University in Vancouver has innovated a way to achieve both. Using artificial intelligence to drive biomedical research analysis, scientists are now able to gather more data from experiments, with fewer confounds and less bias.

Currently, researchers spend large amounts of time on analytic tasks that are difficult for humans to conduct accurately and reliably. Many mistakes and biases are introduced into an experiment in the final analytic steps, and are easily avoidable with use of computational assistance.
In words of one WSU researcher, "computers are very good at the tedious, labor-intensive analytic tasks that are hard for humans. Incorporation of artificial intelligence into these biomedical research procedures can dramatically improve both speed and reliability of the resulting data, and discoveries."

In biomedical research, analysis of cellular histology is used to detect physiological changes that underly normal- and disease-state cellular processes. Often, days or weeks of experimentation culminates in microscopic analysis of digitally imaged cells and tissues. "Unfortunately, when human observation is introduced during these final histological steps, the process is often subject to unintended bias.
Automating and standardizing the analysis of these histological images and cellular quantification greatly improves the accuracy, precision, and reproducibility of data," said one user of WSU's analytics AI.

WSU's AI team has developed software, called PIPSQUEAK (Programmed Intelligence Procedure to Standardize & Quantify Empirical Analysis Kit), for the automated analysis and quantification of cellular physiology. "It all began with a beta version of the PIPSQUEAK software. Because of a 2018 WSU Commercialization Gap Fund grant, we were able develop the use of artificial intelligence (AI)," said primary investigator Dr. John Harkness. "A beta version of PIPSQUEAK was released in 2016 and can complete analysis of an image in approximately 10 seconds, compared to the 10-15 minutes required by hand, reducing the time required for analysis by roughly 99%. The beta version can analyze approximately 200 images in 1 hour; a process that would easily take a skilled researcher 40 hours to complete by hand."

According to PIPSQUEAK early adopters, "The development of this AI has great potential to push biomedical research forward. Current techniques used to identify and measure cell physiology require lengthy image preparation and manual identification of cells."

PIPSQUEAK beta is now available here, where you can learn more about WSU's push to bring artificial intelligence to the forefront of biomedical research.

For description of protocol, see:

Please cite Slaker, Harkness, and Sorg (2016) when using the PIPSQUEAK macro. (PDF here)
Slaker M, Harkness JH, and Sorg BA. A standardized and automated method of perineuronal net analysis using Wisteria floribunda agglutinin staining intensity, IBRO Reports, 2016

You can download PIPSQUEAK from ImageJ by adding our repository to your FIJI Update Sites. Instructions here.

Thank you for support from:

Washington State University
2016 WSU Postdoctoral ADARP Grant
2018 WSU Office of Commercialization Gap Fund
Rewire Neuroscience, LLC