Evident Technology Center Europe GmbH

Artificial intelligence (AI) solutions that automatically extract information from digital histology images have shown great promise for improving pathological diagnosis. Prior to routine use, it is important to evaluate their predictive performance and obtain regulatory approval. This assessment requires appropriate test datasets. However, compiling such datasets is challenging and specific recommendations are missing. A committee of various stakeholders, including commercial AI developers, pathologists, and researchers, discussed key aspects and conducted extensive literature reviews on test datasets in pathology. Here, we summarize the results and derive general recommendations on compiling test datasets. We address several questions: Which and how many images are needed? How to deal with low-prevalence subsets? How can potential bias be detected? How should datasets be reported? What are the regulatory requirements in different countries? The recommendations are intended to help AI developers demonstrate the utility of their products and to help pathologists and regulatory agencies verify reported performance measures. Further research is needed to formulate criteria for sufficiently representative test datasets so that AI solutions can operate with less user intervention and better support diagnostic workflows in the future.

Microfluidics is used increasingly for engineering and biomedical applications due to recent advances in microfabrication technologies. Visualization of bubbles, tracer particles, and cells in a microfluidic device is important for designing a device and analyzing results. However, with conventional methods, it is difficult to observe the channel geometry and such particles simultaneously. To overcome this limitation, we developed a Darkfield Internal Reflection Illumination (DIRI) system that improved the drawbacks of a conventional darkfield illuminator. This study was performed to investigate its utility in the field of microfluidics. The results showed that the developed system could clearly visualize both microbubbles and the channel wall by utilizing brightfield and DIRI illumination simultaneously. The methodology is useful not only for static phenomena, such as clogging, but also for dynamic phenomena, such as the detection of bubbles flowing in a channel. The system was also applied to simultaneous fluorescence and DIRI imaging. Fluorescent tracer beads and channel walls were observed clearly, which may be an advantage for future microparticle image velocimetry (μPIV) analysis, especially near a wall. Two types of cell stained with different colors, and the channel wall, can be recognized using the combined confocal and DIRI system. Whole-slide imaging was also conducted successfully using this system. The tiling function significantly expands the observing area of microfluidics. The developed system will be useful for a wide variety of engineering and biomedical applications for the growing field of microfluidics.