Technology That Can Save Lives

Thanks to Microsoft, the Department of Computer Science is developing technology that can save lives.

Currently, medical images can be analyzed by a class of deep neural networks called convolutional neural networks (CNN). It’s a type of machine learning that analyzes medical images and can make predictions about diagnosis.

In June, Microsoft released the source code for neural networks that use encrypted data. That was the starting point for research conducted at Cal Poly Pomona over the summer that was funded by a Discovery Through Research gift from Microsoft. The gift was made possible by alumna Maria Alvarez, who is the Engineering General Manager, Shared Engineering Services at Microsoft-Artificial Intelligence and Research Division.

“Being able to help students and give back to my alma mater is meaningful. It’s important for students to be prepared for the future,” she says.

Alvarez (’95, master’s in computer science) also supports student success through scholarships.

The Microsoft gift funded a project that builds on the single-model CNN called “Privacy Preserving Inference with Convolutional Neural Network Ensemble.” Due to privacy concerns, medical imagery is encrypted and the analysis for this project is done with the encrypted data.

Computer science student Michael Nguyen learned from his experience in working with developers at Microsoft.

“I learned to professionally communicate with Microsoft developers to debug and discuss problems that arose. The developers were helpful and quick to respond. My CPP advisor was also always available. I felt surrounded by helpful people who wanted my project to succeed.”

“The innovation that this research brings is the ensemble, or the ability to analyze images from several neural networks,” says faculty advisor Tingting Chen, associate professor of computer science. “It’s unique because different hospitals may be working with different models. Combining them can increase the accuracy, without compromising data security.” Increasing accuracy means better diagnosis of imagery and that can save lives.”

For information on how you can support Discovery Through Research, contact Development Director Melissa J. Martinez at 909-869-4160 or melissam@cpp.edu. 

 You can also click here to give online.