Artificial intelligence in x-ray technology​

ZADD Segmentation​

AI-based defect inspection for computed tomography​

The app ZADD Segmentation detects small and fuzzy defects in components reliably and quickly even on poor image quality. For this purpose, the machine learning-based software relies on Artificial Intelligence. Defects and abnormalities are detected, segmented and evaluated using AI for CT data inspection. ZADD thus supports your X-Ray applications in component development, process optimization and fault analysis. ZADD being the acronym for ZEISS Automated Defect Detection, is an optional app for our standard CT inspection software ZEISS INSPECT X-Ray.​

Your advantages with ZADD segmentation at a glance​

  • Time savings with AI

    Time savings with AI

    • Minimize inspection effort​
    • Reliable and fast error detection
  • Robust results & clear reporting​

    Robust results & clear reporting​

    • Reliable results, even if image quality is not perfect​
    • Suitable for mixed and dense materials
  • Easy defect assessment​

    Easy defect assessment​

    • Custom optimization of defect analysis​
    • Simple evaluation and recognition of scrap parts

ZEISS automated defect detection

AI software for your application areas

The image shows a component that can be inspected for defects using artificial intelligence in CT.

Detect defects in components reliably​

Various defects can occur during the complex manufacturing process of components. Especially inside, they are not visible to the naked eye and can have a major impact on the stability and functionality of the component. Artificial intelligence combined with industrial computed tomography makes these hidden problem areas visible early. The ZEISS Automated Defect Detection software specializes in the detection of different defects, so that even on poor image quality with many artifacts, defects can be detected quickly and reliably.​

The image shows the performance of an inline inspection completed in just 60 seconds using AI in CT.

Identify and sort scrap at an early stage​

To be able to sort out defective components in a value chain at an early stage, the 3D data must be evaluated reliably and quickly. Thanks to ZADD, components with critical defects are easily recognized and can be accurately sorted out or, if possible, reworked. Good parts, on the other hand, pass through the further machining process unhindered. The result: a lower rejection rate and high quality of the components. In this way, you can achieve a steady increase in efficiency and maximum process reliability with AI in CT.​

How ZADD Segmentation works​

Good part or bad part? ZEISS Automated Defect Detection (ZADD) supports this decision with Artificial Intelligence. Complete your evaluation with the ZADD Segmentation app for ZEISS INSPECT X-Ray. Watch this video to see how it works.

  • Please note that our software is now called ZEISS INSPECT X-Ray


  • Image of ZEISS VoluMax

    Data aquisition

    • Use our CT portfolio for data acquisition, e.g. the high-power ZEISS VoluMax 9 titan or the high-precision ZEISS METROTOM series for state-of-the-art computed tomography measurements​ 
    • Or import your acquired data from a different CT system into ZEISS INSPECT X-Ray for evaluation.
  • Illustration of segmentation​


    • Find defective areas in the acquired data using Deep Machine Learning (ML) with ZADD Segmentation​
    • Particularly well suited for defect analysis with suboptimal volume data (e.g. due to noise, coarse resolution or artifacts)​
    • Learn from examples, no complicated parameter-tuning is necessary​
    • Use pre-trained ML models to get started quickly ​
    • Create regions of interest (ROIs) for time-optimized evaluations
  • Evaluation​


    • Visualize your data with powerful tools in 3D and 2D​
    • Determine metrics for the detected defects, like diameter, volume or sphericity​
    • Create filters to select defects with a property above or below a given threshold​
    • Make use of complex evaluations like P202, P203, porosity or distance-to-surface
  • Reporting and statistics​

    Reporting and statistics​

    • Evaluation in illustrative measurement reports ​
    • Easy transfer of measurement reports in PDF format ​
    • Archiving the data enables long-term tracking of defects ​
    • Advanced evaluation and statistical functions with ZEISS PiWeb Reporting Plus ​
    • The statistical process control enables the recognition of process correlations and optimizations

Pre-trained models for specific applications​

When using the app ZADD Segmentation in ZEISS INSPECT X-Ray you can benefit from our pre-trained Machine Learning models. Use one out of three available options for alloy castings, hairpin inspection or electronics.​



AI-inspection of hidden defects in alloy castings​



Automated hairpin analysis for e-drive applications​



Solder joint analysis made easy in electronics​

Examples of typical casting defects that ZADD can find​

  • Pores


    A pore is a spherical or ellipsoidal cavity with mostly smooth walls inside the component. Depending on their origin, they can contain air, vapor, hydrogen, or other gases (e.g. from lubricants). They often occur in upper casting layers, but in poorly evacuated areas or undercuts they can be distributed within the whole casting.​

  • Cold run / cold shut

    Cold run / cold shut

    Cold run occurs preferably on flat surfaces with relatively low thickness. This can result in a separation of the cohesion, leaving holes, areas that have not run out, but also rounded edges and overlaps. In die casting, cold run can be seen on very fine and thin surface slates.​

  • Micro-porosity


    Micro-porosity can be understood as an accumulation of small shrinkage holes (micro-shrinkage / interdendritic shrinkage), which can create chains and lead to leakages. This porosity appears in a CT scan with lower resolution as spongious areas.​

  • Wall displacement

    Wall displacement

    If, for example, defects occur in the positioning of the core in the mold before casting, or if the cores shift during the casting process, the geometries of the casting no longer match the CAD model.​

  • Chips


    During rough machining of the component (e.g. saw cut on the feeder), aluminum chips are produced which can fall into the component. Likewise, small protrusions (feathers) can break off during coring and remain in the component. These aluminum residues can lead to defects in the cooling system, for example, during subsequent operation.​

  • Inclusions


    Inclusions are partially or completely embedded impurities in the cast component which are usually denser than the base material. They are caused, for example, by foreign bodies in the casting mold or by contaminated casting material.​

Artificial intelligence (AI) in computed tomography (CT)​

  • Artificial intelligence is ubiquitous. Autonomous driving is just one of many examples of the application of AI. Artificial intelligence is also a topic in industry and thus in computed tomography and is becoming increasingly significant. This is because it enables defect analyses to be carried out even more reliably, accurately and quickly. In industry, a defect is often located inside a component. An optical inspection process for quality control is then no longer sufficient because it does not provide any indication of internal defects. X-ray inspection allows a close look inside a component and can thus detect defects at an early stage. By using AI in CT inspection, a partially automated defect analysis is realized.​

    Explanation of terms:​

    In connection with AI and CT, the terms AI Defect Detection or AI Anomaly Detection are often used. AI stands for Artificial Intelligence and Defect Detection or Anomaly Detection means defect detection or anomaly detection. The addition of "NDT" makes it clear that AI works non-destructively, because NDT stands for non-destructive testing.​

  • AI and computed tomography

    Artificial intelligence is a trend in automation. Process requirements are becoming tighter and tighter, and even in harsh measurement environments, image evaluation and defect analysis must work quickly and reliably. This is especially true for safety-relevant components, e.g. in the automotive industry or aerospace. To increase quality by performing defect analyses faster, while at the same time offering high process reliability, AI is used for reading CT scans. Defect detection with AI eliminates the need for manual tuning of parameters, thus avoiding subjective decisions in defect detection.​

    ZEISS Automated Defect Detection is particularly useful when volume data is affected by overly dense materials or short scan times. While artifacts and noise in the images usually cause faulty detections, the software remains unaffected by these effects.

Contact us for a personal demo

Our service will accompany you from the beginning whether selecting the right ML model or developing a specially trained solution. We support you in the operation, optimization and evaluation performance of the system and solve your individual inspection tasks in many cases.

Form is loading...

If you want to have more information on data processing at ZEISS please refer to our data privacy notice.