X-ray computed tomography measurement

X-ray computed tomography (XCT) measurement is a non-destructive technique that enables inspection and dimensional analysis of industrial products^[1]. Its principle is similar to the traditional XCT system, which capturing a number of projection images from different angles and reconstrucing the 3D volume^[2]. Since XCT can reveal both external and internal features in the product without any destructive operation, it has been used for quality control of products and drawn significant attentions in manufacturing such as aerospace^[3], automotive^[4], food packaging^[5], and additive manufacturing^[6]. The detailed principle, structure, and applications of the general XCT system can be found in the article entitled CT scan^[7].

Measurement using X-ray computed tomography

Industrial XCT system

Typical X-ray computed tomography system in hospitals

X-ray computed tomography (XCT) has been widely used in both medical and industrial applications^[8]. Basically, it relies on the penetration ability of X-ray to detect the internal structure of the object^[9]. The mostly used XCT machines in hospitals are characterized by a large table for holding the patient and a gantry with hardwares inside^[10]. However, the industrial XCT system is different. A general industrial XCT system mainly includes the following modules^[11]:

• X-ray tube: generates the X-ray beam by accelerating electrons towards a target.
• Detecter: receives X-rays passing through the object and converts the X-ray intensities into electrical signals.
• Manipulator: rotates the object during the scan, providing stable rotation and high positioning accuracy.
• Linear motion axes: control the movement of the manipulator, detector, and other components.
• Reconstruction Module: reconstructs 3D volumes from projection images using reconstruction algorithms.
• Analysis Software: specialized software tools for measurement, defect inspection, dimensional analysis, and reporting.

Projections during a industrial XCT scan

In industrial XCT systems, the X-ray tube and detector are usually static during the scan, and the object is rotated by the manipulator^[11]. One of the differences between medical XCT and industrial XCT lies in that, medical XCT requires detecting the tissues inside the body and low radiation dose for human being, while industrial XCT requires high stablity and super resolution for details in the product^[12]. Currently, industrial XCT has been used in production lines, food factory, etc., to inspect the possible defects in the product without destructions and control the manufacturing quality^[2].

Expanded for dimensional measurements

In recent years, industrial XCT was expanded beyond structural inspection to include dimensional measurement applications^[13]. The goal is to use industrial XCT system, like a rular or caliper, to measure both the internal and external dimensions of the product^[14]. This is especially valuable for inspecting and controlling the quality of additive manufactured products, which often contain complex internal and external structures^[15].

Procedures and main operations

A brief work flow for the measurement using X-ray computed tomography

One complete XCT measurement basically includes the following steps: projection acquisition, pre-processing, reconstruction, post-processing, surface determination, and final evaluation^[16].

Projection acquisition

Projection acquisition aims to physically collect a number of projection images by rotating the object while irradiating it with the X-ray beam^[17]. Prior to irradiation, experienced technicians usually calibrate the XCT system, determine acquisition parameters, decide the orientation of the product, fix the product on the manupulator, and supervise the scanning process^[11]. This process generally take minutes or hours to obtain high-quality projection images suitable for the XCT measurement task. Finally, the projection images are saved in the computer for subsequent processing^[18].

Pre-processing

In practice, projection images contain both structrual information and noise^[19]. A high level of noise in the images can cause significant errors in the final measurement results. This noise may originate from unstable X-ray energy, electrical noise in the detector, physical interactions between the X-ray and the product (such as beam hardening and scattering), vibrations of the mechanical system and the product, and other factors^[20]. Although the noise can be controlled to some extent by improving hardware manufacturing quality and calibration accuracy, it cannot be completely eliminated. Therefore, filtering and correction of the projection images are necessary to ensure measurement accuracy^[21].

Reconstruction

After collecting and necessarily pre-processing on the projection images, a reconstruction should be conducted to reveal the real 3D volume^[9]. This process relies on reconstruction algorithms which are based on the Beer-Lambert's law^[22][23]. More details can be referred to Tomographic Reconstruction. After reconstruction, a voxelized volume is obtained, with the gray value of each voxel representing its X-ray attenuation coefficient (or, simply put, the material’s density)^[24]. As one of the core steps in XCT measurement, the quality of reconstruction significantly influences the final result. Reconstruction is typically performed using the software provided with the XCT system^[25].

Post-processing

The step following reconstruction is post-processing of the 3D volume. The goal of post-processing is also removing noise and improving the volume quality (such as improving contrast and clarifying boundaries, etc.)^[26][27].

The detected surface by a surfce determination technique^[28]

Surface determination

After obtaining a high quality 3D volume, although the details of the product structure can be clearly observed, the dimension can still not be extracted^[29]. To deal with that, surfaces (or boundaries between solid and air) have to be determined. Surface determination techniques detect the voxels belonging to surfaces and generate points from the voxels. Therefore, the output of surface determination is a point cloud that represents the surfaces of the product^[30].

Final evaluation

Following surface determination, the points on the feature should be fitted using standard mathematical models to obtain the characteristics of the target feature, such as diameters, distances, lengths, etc.^[11]. Besides, the uncertainty of the XCT measurement results should also be estimated^[31].

Current challenges

Measurement uncertainty

For dimensional measurements, the measurement uncertainty is as important as the measurement accuracy^[32]. However, there are many error sources in XCT measurement processes, and the influence of these errors on the final measurement errors is not yet fully understand^[33]. Currently, uncertainty estimation relies on scanning an artifact using the same setup as in measurement tasks. More efficient and accurate methods for estimating uncertainties are still needed and are the focus of ongoing research^[16].

Strategy optimization

There are many factors in the XCT measurement procedure that can significantly influence the results, such as acquisition parameters, filters used in pre-processing, and the reconstruction algorithm^[34]. Furthermore, these factors may have strong interactions with each other. Therefore, developing an appropriate strategy to achieve optimal measurement accuracy remains a challenge^[35].

Higher resolution

The resolution of the XCT system depends on the quality of the hardware, such as the resolution of the detector and the spot size of the X-ray source^[36]. Furthermore, for large industrial products, inspecting a large volume requires a smaller magnification during scanning, which results in larger voxel sizes and consequently lower resolution^[37]. Achieving a large field of view while maintaining high resolution remains a significant challenge^[38].

Measurement efficiency

To ensure measurement accuracy, the quality of the projection images must be sufficiently high^[39]. This typically requires longer exposure times and a larger number of projections, resulting in extended acquisition times that can range from tens of minutes to several hours^[18]. The long measurement time hinders its application in production lines. Several approaches have been tried, such as reducing the exposure time or the number of projections, but the measurement accuracy is usually compromised^[40][41]. Improving measurement efficiency while maintaining the accuracy is a goal for the future^[18].

Main suppliers

Currently, there are several main suppliers of XCT systems used for measurements. More specifically, these system are usually called Metrological X-ray Computed Tomography Systems. Main suppliers, newest models, and characteristics are listed in the table.

Supplier	New model	Characteristic
Zeiss	METROTOM 1500[42]	High resolution and traceable accuracy. Suitable for a large variety of parts and applications. DAkkS calibration for reliable measurements.
Nikon	MCT225[43]	High accuracy. Wide range of sample sizes and material densities.
Comet Yxlon	FF35 CT Metrology[44]	Dual tube configuration for highest versatility. Non-sequential fast data acquisition. Premium tubes and detectors for a broad inspection range.
Waygate Technologies	Phoenix V|tome|x M Neo[45]	Highly productive Dual|tube scanner. Dynamic 41 detectors for 2-3x faster CT scans or doubled resolution. higher power on a smaller focal spot.
North Star Imaging	X5000[46]	Large scanning envelope for larger parts while maintaining the ability to inspect small components.
Tescan	UniTOM HR[47]	Spatial resolution down to 600 nm. Perform true 4D imaging with Dynamic CT. Accommodate a broad sample range.

See also

• X-ray

• X-ray tube
• CT scan
• Tomography
• Tomographic reconstruction