AI Microscopy Solutions

Enhancing Image Acquisition, Analysis and Service

Discover the seamless integration of AI into microscopy, and unleash the potential of your research like never before.

Classification of cellular phenotypes with AI
Classification of cellular phenotypes with AI
Classification of cellular phenotypes in U2OS + LLC2 cell culture. A few cells were manually assigned (left). ZEISS Intellesis Classification then predicted the complete dataset (right).

The Future of Microscopy with AI-Powered Solutions

Artificial Intelligence (AI) has been a game-changer in many sectors, and microscopy is no exception. ZEISS AI Microscopy Solutions leverage the power of artificial intelligence to revolutionize the way you acquire images, analyze and interpret microscopic data and manage the performance of your microscope.

Lithium ion battery cathode - analyzed with AI
Lithium ion battery cathode - analyzed with AI
Left: Cathode from a lithium-ion battery imaged with SEM with BSE detector, 10 kV Right: Image segmentation with red (aluminum current collector), yellow (nickel cobalt manganese - NCM - storage particles), and green (free space between the particles), using ZEN Intellesis AI module. Image courtesy: University of Aalen, Germany.

Integrating AI into Microscopy

Our AI microscopy solutions are designed to enhance precision, speed, and scalability in your research, offering unprecedented insights into the microscopic world. Whether you are in the field of life sciences, materials science, geosciences, electronics or education and routine, our AI microscopy solutions are here to transform your microscopic exploration into an experience of the future.

The Significant Impact of AI in the Entire Workflow

AI has significantly impacted the following areas of microscopy, encompassing sample detection, image acquisition, preprocessing, and image analysis:

  • Automatic Sample Detection: Traditionally manual, microscopy sees a transformation with AI. ZEISS AI Sample Finder streamlines sample placement, focus adjustment, and identification, reducing manual efforts for swift experiment initiation.
  • AI-Guided Image Acquisition: In life sciences, it's common to identify small regions of interest from large field of view images. Manual methods are time-consuming and prone to bias. AI excels at swiftly sifting through extensive data. ZEN's Guided Acquisition Module seamlessly integrates microscopy automation with AI-powered image analysis, enabling efficient guided acquisition workflows.
  • Denoising: In applications like live cell imaging, working with high photon doses is impractical due to the risk of photobleaching and phototoxicity. This often results in noisy images, requiring denoising for meaningful scientific insights. ZEN addresses this through the Intellesis Denoising module, utilizing AI tools like the Noise2Void algorithm to effectively preserve information for quantitative image analysis.
  • 3D Reconstruction: AI-driven 3D reconstruction of X-ray tomography images accelerates data acquisition by over 10X while maintaining image quality, surpassing traditional methods. ZEISS Xradia 3D X-ray platform, featuring the Advanced Reconstruction Toolkit (ART), employ deep learning-based reconstruction to enhance acquisition throughput across diverse fields, including geosciences, pharma, biosciences, electronics, batteries, and engineering materials.
  • Image Segmentation: A crucial step in microscopy image analysis workflows, particularly in bioimage analysis. Deep learning has demonstrated remarkable effectiveness in automating segmentation at both the pixel level (semantic) and object level (instance). ZEISS's software ecosystem, encompassing ZEN and arivis, offers diverse AI-powered segmentation tools that harness both conventional machine learning and deep learning approaches.
  • Object Classification: Once images are segmented to identify various objects, the next step often involves classifying these objects into different groups. ZEN incorporates user-friendly AI-powered classification algorithms that can be easily trained by anyone to categorize objects into custom classes.
  • Predictive Service: Realizing the full potential of the microscope relies on its optimal functionality. AI plays a crucial role in predicting potential issues, minimizing unplanned downtime. ZEISS Predictive Service proactively identifies and addresses potential problems before they occur, safeguarding against unplanned downtime.

AI-Based ZEISS Solutions

ZEISS Advanced Reconstruction Toolbox

ZEISS Advanced Reconstruction Toolbox

State-of-the-Art Reconstruction Technologies for Your ZEISS X-ray Microscope or microCT

arivis Cloud – Enhanced Image Analysis Solution

ZEISS arivis Cloud

Run pre-defined workflows and train deep learning models on our cloud-based platform for AI-driven image analysis.

arivis Pro

ZEISS arivis Pro

arivis Pro gives you the power to handle your datasets without constraints and get your results in next to no time.

AI Toolkit

AI Toolkit

Boost Your Image Analysis Capabilities with Machine Learning

ZEISS Labscope

The Easy-to-use Imaging App



Software Suite for Connected Microscopy from the Materials Lab to Production

ZEISS AI Sample Finder​

ZEISS AI Sample Finder​

Reduce your time to experiment from minutes to just seconds.

ZEISS Axiovert 5 digital

ZEISS Axiovert 5 digital

Artificial Intelligence will automate your routine work and make your processes more efficient and your results more reproducible.


  • AI-based image analysis in microscopy utilizes artificial intelligence, including machine learning and deep learning, to enhance the analysis of microscopic images. It involves training algorithms to recognize patterns and structures, improving the accuracy and efficiency of tasks such as sample detection, image segmentation, denoising, object classification, and 3D reconstruction. This technology accelerates and automates image analysis in microscopy for more effective scientific research and diagnostics.

  • In microscopy, deep learning involves the training of artificial neural networks to independently analyze microscopy images by learning patterns and structures from datasets annotated by experts. This training enables the networks to perform tasks such as image segmentation, object detection, and classification with increased precision and efficiency. For a deeper understanding of the distinctions between machine learning and deep learning, you can refer to our AI eBook.

  • The integration of AI in microscopy guarantees the acquisition and analysis of multidimensional image data at high throughputs and in a reproducible manner—fundamental for scientific discovery. These technologies automate routine manual tasks, liberating valuable time for researchers to design more experiments and innovate at an accelerated pace.

  • a. Whether seeking new cancer therapies through subcellular observation in a fluorescence microscope, designing battery materials via microstructure analysis with SEM, or optimizing hydrocarbon yield through mineralogy study with an X-ray microscope, AI provides insights from both small and large datasets at higher throughputs. This versatility underscores the transformative impact of AI in accelerating scientific discoveries across numerous domains in microscopy.

  • Traditional image analysis relies on predefined rules, such as simple thresholding, to extract information from images. For instance, using a fixed threshold to segment images based on pixel intensity. However, these rule-based approaches can be unreliable when images deviate from ideal conditions, leading to irreproducible results. In contrast, AI-based analysis goes beyond specific rules, as AI systems learn hidden patterns from diverse datasets. This enables them to generalize and maintain robustness, even when imaging conditions change, ensuring more reliable and adaptable results in microscopy.

  • AI is progressively addressing everyday challenges, and this trend extends to microscopy, where every aspect of the workflow stands to benefit. As generative AI continues to evolve, serving as a valuable assistant, we anticipate more natural language-driven interactions with microscopy systems. Imagine a user loading a multi-well plate with numerous experiments and simply asking, "Please image these samples at 10X and summarize the insights from all the wells." This vision foresees a seamless transition from sample to insights, eliminating the need for manual setup, data transfer, preprocessing, segmentation, and classification. The future of microscopy with AI promises heightened productivity, even in handling massive experiments involving tens of thousands of iterations.

  • For those proficient in Python coding, implementing AI-based solutions in microscopy involves utilizing open-source libraries. However, this requires expertise in handling large datasets and a deep understanding of managing image patches and reconstructing them to their original size. Fortunately, ZEISS offers a range of AI tools through a user-friendly, no-code interface accessible to everyone. In both ZEN and arivis software ecosystems, users can leverage tools for denoising, segmentation, classification, and more using AI. For example, the arivis Cloud platform provides infrastructure for training custom AI models in image segmentation, usable either on the cloud or locally in arivis Pro or ZEN software packages, ensuring consistent results across platforms.

  • Whether it is ZEN, ZEN Blue, arivis Pro, arivis Hub, or arivis Cloud, every ZEISS software incorporates AI tools for image analysis. This ensures that users benefit from advanced capabilities, making each software package a powerful tool for enhancing productivity.

  • Deep learning (DL) is a subset of machine learning (ML), but when people mention machine learning, they typically refer to conventional ML, where predefined digital filters are applied to training data. This conventional approach involves feature engineering, where someone assembles these filters, often working with a limited number of features, typically up to a few tens. This limitation can make it challenging for these algorithms to capture the complexity present in some images. In contrast, deep learning is a feature learning technique that optimizes millions of parameters by learning from vast training data, making it more robust against changing image conditions. In essence, the key distinction lies in the few engineered features of ML versus the multitude of learned features in DL.