Image quality control for outputs of digital pathology and spatial omics experiments

In November 2021 Zegami completed a project for Ultivue, a spatial biology company that specialises in multiplexed tissue staining and imaging. The company produces reagents that enable researchers and clinicians to obtain whole slide images of tissues at high resolution to help find novel biomarkers using their novel immunohistochemistry platform.

Ultivue have a complex image processing pipeline that requires images to be of high enough quality to ensure their image analysis provides accurate and meaningful results. Feedback on the quality of images (e.g. blurring, poor staining, cell necrosis) needs to be fast and standardised so that adjustments can be made on the subsequent analysis for each specific problem.

Zegami has helped us to find innovative ways to efficiently detect common artifacts in large collections of tissue images – and to locate them on the slides affected.

Florian Leiss, PhD – VP of Digital Health Strategies at Ultivue

Zegami worked closely with the Ultivue team to understand their user requirements and implement a prototype demonstrating how a customer facing solution could work. The result was a system that could scale to a large number of cohorts that combined unsupervised and supervised machine learning techniques with imaging and visualisation tools. This allowed the customer to easily and quickly identify quality control issues within the imaging pipeline.

Part of the next stage is to use the initial implementation and learnings as a basis to design and develop a customer facing product for Ultivue’s end-users.

Whole Slide Images were sliced and then coloured within the Zegami platform according to observed QC issues