Quality Control for X-Ray Systems A Tool Chain for NDT Applications

Transcription

1 4th International Symposium on NDT in Aerospace Tu.3.A.2 Quality Control for X-Ray Systems A Tool Chain for NDT Applications Stefan REISINGER *, Alexander ENNEN *, Thorsten WÖRLEIN *, Michael SCHMITT *, Virginia VOLAND * * Fraunhofer IIS/EZRT, Development Centre X-ray Technology (Dr.-Mack-Straße 81, Fürth, Germany / virginia.voland@iis.fraunhofer.de) Abstract. Within the last years, X-ray systems have been established in the field of NDT. Mainly caused by the following aspects the confidence in this technology is still inferior to the confidence in competitive NDT-techniques: Long-term degradation of the system status caused by aging process and abrasion of the system components Fluctuations of image quality during the measurement caused by instability of the X-ray source and possible breakdown of detector elements Poor image quality affected by defect detector elements or other detector characteristics, by a changing imaging geometry during the measurement and by geometric misalignments Insufficient repeatability and measurement accuracy In order to cope with these challenges, we implemented procedures to monitor the X-ray system. The monitoring of the system status is divided into several phases: Prior to use, a computed tomography (CT) system and all its components have to be adjusted to each other in order to prevent any kind of misalignments such as detector or axis tilt. The consequences of unadjusted CT Systems manifest in artifacts or blurred images and therefore result in lower image quality and worse analysis results. The stability of the system components should be controlled regularly during the system lifetime, e.g. weekly. Influencing factors, which are used to forecast the expected image quality and end of life of the X-ray components, are factors such as the spatial resolution, number and distribution of defect detector elements, the contrast sensitivity or size and stability of the focal spot. In preparation the measurement process should be simulated to find suitable parameters for the following measurement. Moreover, for each measurement the image acquisition must be monitored with regard to the image quality to be expected. We therefore defined a quantitative set of characteristics such as e.g. a constant illumination of the detector and present a way to monitor the data acquisition with respect to these characteristics. After the image acquisition is done, a wide range of algorithms can be applied if necessary in order to correct image artifacts and to further improve image quality before the reconstruction task and the final analysis steps occur. Algorithms for the automated adjustment, characterization and monitoring of an X- ray system and correction methods for the further improvement of image quality are presented. License: 1

2 1. Introduction Figure 1.1: Tool chain for NDT applications To ensure the quality of NDT-techniques different procedures have been established. For example the standard for tactile systems is to use calibration phantoms to ensure a stabile system status and resulting measurement inaccuracy. Though within the application area of industrial X-ray systems guidelines for the determination of measurement inaccuracy [7] exist, still no standard is committed in terms of a consistency check as in the medical application. In order to cope with these challenge, we found procedures to monitor the X-ray system. To ensure the stability of an X-ray system different steps are necessary (Figure 1.1). Once while bringing a system into service or after replacement of system components, the single components (X-ray detector, X-ray tube and manipulation system; Figure 1.2) must be aligned with respect to each other. Continuously (daily or weekly) the stability of the system components should be controlled. Besides parasitic drags on system quality has to be logged and corrected (during the measurement or in a post step). Therefore the aim is to develop a tool to monitor the system to allow conclusions over the system status. As a result an expert system can be established to determine suitable parameters and necessary corrections dependent on system status and monitoring of each measurement. Figure 1.2: Drawing of an X-ray system. 2. Tool chain for NDT applications This chapter describes the single steps within the tool chain for NDT applications. 2.1 System Adjustment Uniquely the system components must be aligned with respect to each other. Otherwise influences like tilts of detector and rotatory axis, imprecisely distances between X-ray source and X-ray detector or that the focal spot and the detector centre axis are not arranged 2

3 in the same plain lead to unsatisfactory image quality. Later corrections on the data cannot adjust all these missing alignments even if the maladjustment is completely known. 2.2 Monitoring of system status The stability of the system components should be controlled regularly during the system lifetime, e.g. weekly. Influencing factors, which are used to forecast the expected image quality and end of life of the X-ray components, are factors such as the spatial resolution, number and distribution of defect detector elements, the contrast sensitivity or size and stability of the focal spot. Especially with regard to performance of certification a logging of system status is essential. 2.3 Monitoring of image acquisition During each measurement task some characteristics are monitored to control the image acquisition. If a parameter (homogenous illumination of detector, mean attenuation, movement of optical focal spot) gets out of specified ranges steps can be defined to act on this behaviour. For example repeating single exposures, waiting for a specified time or even restart or abort of the whole measurement can be reasonable. 2.4 Just in time correction Some influences to image quality can be corrected synchronous during the measurement. These corrections are for example determining the position of the center of rotation and determining the position of the optical focal spot [4] or reduction of beam hardening effects [8]. Thus a online reconstruction with a syncronous correction can take place. 2.5 Post processing/ Correction of raw images After the image acquisition is done, a wide range of algorithms can be applied if necessary in order to reduce image artefacts [1] and to further improve image quality before the reconstruction task and final analysis steps occur [2]. These analysis steps are for example examination of dual energy measurements [5]. 3. Quality control for X-ray systems To simplify the use the tool chain of image processing all single steps have to be linked with each other. The challenge of the X-ray tool chain is monitoring of the time variable status of the X-ray system depending on its life time. The most important issue for a typical user of an X-ray system is the long-term management of best parameters (voltage, power, number of projections, integration time, etc.) for each measuring task. To use the full capacity of a system by means of determination and using of resolution limit and best quality of generated images is dependent on the selection of suitable parameters. Degradation of system status (such as basic spatial resolution (Figure 3.1) and detector efficiency (Figure 3.2)) or the state of single components brings up the demand to adapt the necessary parameters. Moreover these parameters are dependent on the measurement task and given requirements. 3

4 Figure 3.1: Determination of basic spatial resolution according to EN and ASTM Complete failures of single components come along with stealthy degradations. Thus the status of the X-ray system and the single components is logged over the time. With a certain empirical values it is possible to predict coming degradations and complete failures. This allows early replacement of single components especially because sourcing of new components is very time expensive due to long time of delivery. Hence down times of X- ray systems caused by breakdown of single components can be reduced. Figure 3.2: Determination of detector efficiency according to ASTM E [6]. Another part of quality control for X-ray systems is a just in time correction of influences like misalignments (Figure 3.3). Especially in the case of extensive measuring tasks or if the time for the reconstruction of the dataset which take some hours or days like described in [3] it is important to be able to react on parasitic drags instantly. In a extrem case these influences are not until noticed if the whole measurement and reconstrution task is finished and are irreparable. This would effect high costs and loss of time. Figure 3.4 shows a plot of logged values. It is observable that for a single projection no signal was triggered. Therefore a simple reiteration of this measurement fragment would lead to more usable results. 4

5 Figure 3.3: Observation of drift of focal spot. It is obvious, that the movement of the focal spot changes its direction during the long-term observation. Figure 3.4: Observation of influencing values during the measurement. At the marked instant of time all monitored values show anomalous values. This is an indication to insufficient image quality (In this case the tube was dropped out). If a synchronous correction is not possible at least a logging of system behaviour during the measurement to predict necessary corrections on the raw images or the reconstructed volume (Figure 3.5) is recommended. Figure 3.5: left: Ring artefacts with / without correction. Right: Detector shift with / without correction. 5

6 As a follow up, conclusions can be drawn from the final image quality to necessary parameters of coming up measurement tasks. Conclusion The main benefit of these tools is that the whole tool chain of image processing is available in one software solution. Therefore the possibility of data exchange between the single processing steps is given. The coming steps for this project are the generation of empirical values of the degradation of system components and involved impact to image quality. Having this data it is possible to specify necessary adaption of measurement parameters und give a forecast of system live time. References [1] S. Reisinger, V. Voland, S. Burtzlaff, M. Schmitt und N. Uhlmann, Rohbildkorrekturen für die Artefaktreduktion in der Computertomographie, Jahrestagung der DGZfP BB 127-CD, [2] V. Voland, J.-I. Tiemegni, T. Wörlein, N. Uhlmann und F. Kargl, Steuerung von Röntgenexperimenten, Jahrestagung der DGZfP BB 127-CD, [3] V. Voland, J. Freitag und N. Uhlmann, A CT System for the Analysis of Prehistoric Ice Cores, Elst (Hrsg): Microelectronic Systems: Circuits, Systems and Applications, Berlin, Springer, [4] V. Voland, M. Schmitt und S. Reisinger, Geometric Adjustment Methods to Improve Reconstruction Quality on Rotational Cone-beam Systems, in ICT Conference, Wels, [5] M. Firsching, F. Nachtrab, J. Mühlbauer und N. Uhlmann, Detection of Enclosed Diamonds using Dual Energy X-ray imaging, 18th World Conference on Nondestructive Testing, [6] ASTM E e1 Standard Practice for Manufacturing Characterization of Digital Detector Arrays, ASTM International, West Conshohocken, PA, 2003, DOI: /C A, [7] VDI/VDE Bestimmung der Messunsicherheit und der Prüfprozesseignung von Koordinatenmessgeräten mit CT-Sensoren. [8] M. Franz, S. Kasperl und M. Stamminger, Synchronous Arftefact Reduction in Industrial Computed Tomography, Technisches Messen TM 77, pp ,