CHALLENGE: Customer is a pathology software team and was looking to improve the accuracy of pathology image analysis algorithms from baseline and experimental level to production level, for tracing and delimitating cancel cells from non-cancer cells in tens of thousands of blood stream images. But the task was difficult for researchers and software developers in the team because they’re not specialized in AI.

SOLUTION: The team used Cerbrec Graphbook to fine-tune and improve their baseline and experimental AI models for image segmentation and image classification tasks by just drag-and-drop, without needing previous AI tuning experience or steep AI development learning curve.

RESULT: The new custom AI models built with Cerbrec Graphbook increased the accuracy of customer's pathology image algorithms by 30%, meeting their accuracy requirement for production. The cost of completing the project by the team was also 80% lower than when projected without Cerbrec, due to shorter time to delivery and less expensive talent needed.

Contact us for inquiry or demo: sales@cerbrec.com