Data Management

As imaging instrumentation develops, the resulting datasets increase in volume and in the number of dimensions. Handling these datasets for visualization in 3D, 4D, or 5D becomes an ever-increasing computational challenge. Ensuring you have software tools available in your facility to cope with data sizes and volumes is critical to enable your users to visualize and subsequently quantify the structure or events of interest in their multi-dimensional datasets. A combination of high-power computing as well as software optimized to quickly manage these datasets is critical to ensure that your users can “see” in full detail the data they acquired with your instrumentation.

Depending on the size and scope of requirements, ZEISS has different recommendations for data visualization:

• arivis Vision4D: Your modular software for working with multi-channel 2D, 3D and 4D images of almost unlimited size. If your users regularly work with datasets of more than 100GB, arivis Vision4D is ZEISS’ recommended solution. The unique ImageCore technology is RAM independent, and this ensures that arivis Vision4D gives you and your users the power to handle datasets without constraints.
• ZEISS ZEN Software: The user interface used for all ZEISS light microscopes comes complete with the built-in 3D visualization engine designed by arivis. This ensures smooth 3D handling of datasets up to ~100 GB in a familiar software environment. The performance of ZEN is hardware dependent, e.g., more RAM and a faster SSD drive can significantly improve the handling of big data up to 500 GB.
• ZEISS ZEN lite: The free 3D visualization tools available in ZEN lite are also extremely capable. This provides your users with powerful image visualization from their own computers.
  

In most cases, image acquisition is just one step in a complex experimental workflow that ultimately results in quantifiable data. How experiments are planned and configured as well as the subsequent choices for processing and analysis are all key in ensuring accurate and reproducible experimental outcomes for your facility’s users.

ZEISS offers a selection of image analysis choices including a wide range of easy-to-use solutions contained within the acquisition software, such as the ZEN Bio Apps. For applications demanding more complex analysis or involving large datasets more than 100GB in size the arivis Vision4D platform is a powerful solution. When a customized analysis approach is required the APEER platform provides a cloud-based solution that delivers full flexibility in addition to full transparency. Any new analysis function that is programmed by researchers can be quickly and easily integrated into your user’s workflows using APEER, without the need for any programming knowledge and with the assurance of full transparency in terms of processing approach.

With image processing increasingly becoming the bottleneck for gaining insight from imaging data, being equipped with the latest processing and analysis trends is key to staying abreast of the ever-changing experiments that facility users create.

As computing power increases so too does the capability and availability of approaches to train neural networks to recognize or differentiate features of interest. Choosing the best machine learning approach for segmentation, restoration, or classification is vital for the success and reproducibility of your results. Whether the goal is segmenting particular objects of interest or identifying regions of the sample for subsequent acquisition, keeping abreast of the numerous developments and approaches to machine learning ensures you can deliver optimal results for users of your facility.

ZEISS provides a variety of machine learning tools to suit all applications. ZEISS ZEN Intellesis is a quick and easy means of training and deploying segmentation. The push-button ease of the module makes it ideal for the core facility environment when ease of use is often a key factor in onboarding users to the approach. In addition, deep learning neural networks can be trained using our APEER platform or via publicly available networks such as U-Net and then deployed using ZEN Intellesis, thus providing the most flexible approach to machine learning segmentation all within the common ZEISS user interface, ZEN.

If time or manpower is limiting, ZEISS also provides a custom service to train your neural network on your behalf so you can focus on addressing your scientific questions. Contact us if you would like to learn more about this service or if you require a customized solution.