Mon Sep 09 14:00:00 UTC 2024: ## Researchers Discover Optimal Settings for High-Quality Whole Slide Imaging in Digital Pathology

**Hasselt, Belgium – September 9, 2024** – A team of researchers at Hasselt University has identified the optimal settings for generating high-quality whole slide images (WSIs) in digital pathology, balancing image quality with efficiency in scan time and data volume. The findings, published in *PLOS ONE*, offer a valuable contribution to the field, which is increasingly reliant on AI-powered analysis of WSIs for diagnosis and prognosis.

Traditionally, tissue biopsies are examined under microscopes by pathologists. However, digital pathology is rapidly gaining traction, allowing for the generation of WSIs using digital slide scanners. These WSIs provide a wealth of data for analysis through AI and spatial statistics, potentially revealing subclinical features invisible to the naked eye.

The researchers investigated how different scan settings on a ZEISS Axio Scan.Z1 digital microscope impacted the quality and consistency of WSIs. They used human lung cancer tissue samples and evaluated the performance of a deep learning model called StarDist for nucleus segmentation.

Their analysis revealed that while achieving high matching percentages between replicates was possible, the similarity in information from those replicates was relatively low. Additionally, increasing scan time and data volume did not significantly improve the consistency of the results.

Ultimately, the researchers identified an optimal scan setting that balanced high image quality with acceptable throughput time. This setting was selected based on its consistent performance in nucleus detection, resulting in fewer artifacts and minimized inconsistencies between scans.

“This research contributes to establishing a robust methodology for optimizing tissue scans in digital pathology,” stated Dr. Dirk Valkenborg, lead author of the study. “By optimizing the scanning process, we can ensure consistent, high-quality data that is ideal for AI-based analysis, ultimately leading to more accurate diagnoses and better patient outcomes.”

The study highlights the importance of carefully selecting scan settings for digital pathology applications. The researchers’ findings provide valuable insights for optimizing workflows and maximizing the potential of AI in this rapidly evolving field.

**Further research:** The authors acknowledge that their study is limited by the use of only three technical replicates per scan setting. They encourage future research to utilize a more complete factorial design, considering additional factors like artifact detection, image compression, and other available scanner settings. Expanding the analysis to include different tissue types and staining methods would also broaden the applicability of the results.

**Availability:** Data used in the study is available in a public repository on GitHub.

Read More