Bookmark and Share

Standardization News Search
Feature
Digital Reference Radiographs Offer Possible Savings

by Michael Hutchinson

Digital radiography encompasses a wide range of technologies, including flat panel technologies, computed radiography, and a variety of scintillator and digital camera-based technologies. The marked improvement in technology of digital radiography in the past decade is opening up many opportunities for cost savings. In switching from traditional film radiography to digital radiography, the costs associated with purchasing film, processing film, and chemical waste disposal are eliminated. In addition significant savings can be realized through cycle time reduction and the ease of automation that digital radiography offers.

Lack of Reference Images Slows Implementation

Currently one of the major hurdles to the implementation of digital radiography for a number of industrial uses is the lack of digital reference images. What seems to be at least an interim solution in the aluminum casting industry is to use the existing film reference radiographs (ASTM E 155, Reference Radiographs for Inspection of Aluminum and Magnesium Castings). But the existing reference radiographs have proved inadequate for two main reasons: (1) the difference in spatial resolution between radiographic film and the digital radiographic systems, and (2) the difference in dynamic range between film and many of the digital detectors.

The grain size of common radiographic film ranges in size from about three microns to 10 microns. The pixel spacing for digital radiographic systems that are proposed for the inspection of aluminum castings, without the use of geometric magnification, range from 50 microns to 139 microns. These differences in resolution do not seem to affect the detection of a discontinuity but do affect the grading of the severity level of the discontinuity. Figure 1 is a film radiograph showing plate three of elongated porosity (ASTM E 155) that was digitized with a pixel spacing of 140 microns. Figure 2 is a film radiograph of Plate 5 of the same series digitized with a pixel spacing of 50 microns. And Figure 3 is the same as Figure 1, only digitized at 50 microns. You can see that by comparing these figures, the difference in resolution of a detector (or in this case digitization pixel size) between 50 microns and 140 microns makes a shift of approximately two plates in the apparent severity level.

The difference in dynamic range between film and many of the digital detectors is the second reason that the use of ASTM E 155 reference radiographs has proved inadequate for the grading of the severity level of aluminum castings. Digital detectors with their wide dynamic range, coupled with the limitation on the number of grey level intensities that humans can differentiate, make it necessary to step through the data of a given image with a series of windows. This is done by adjusting the contrast (window width) and then changing the brightness (window level) in a series of steps to view the data. The trouble comes in adjusting the contrast of a production radiograph taken with a digital detector. When using a high contrast, the discontinuity looks worse (higher plate number). When using a low contrast, the discontinuity may not be visible at all. An example of this effect is in Figures 4, 5, and 6. All of the images are from a single 16-bit dynamic range digital radiograph of the ASTM E 155 hardware for elongated porosity in aluminum. Figures 4 and 6 are the same contrast setting and Figure 5 is at a slightly higher contrast setting. In comparing Figures 4 and 5, there is a noticeable difference even though they are both showing plate two and are from the same radiograph. Instead, Figure 5 looks more like Figure 6. Figure 6 is of Plate 7, a difference of five severity levels from Plate 2 (Figures 4 and 5).

ASTM Team Formed

In an effort to overcome these difficulties, ASTM Committee E07 on Nondestructive Testing has sanctioned an industry team to develop a set of digital reference images and the methodologies to use them. These new images and methods will be used in place of the current ASTM E 155 reference radiographs when viewing digital radiographs of aluminum castings. The goal of this effort is to allow the use of digital radiography for aluminum castings without changing the classification of aluminum castings as compared to the traditional film radiography using ASTM E 155 in a statistically significant way.

The approach of the team has been to start with the ASTM E 155 hardware and take new film radiographs of the hardware using the same techniques as are used to create the E 155 reference radiographs. The only difference is the inclusion of a step wedge and two duplex line pair gauges. The addition of the step wedge is to aid in the setting of the proper contrast. The addition of the line pair gauges serve as a check on the resolution throughout the process from radiograph, to the digitized image, to the displayed image by a digital detector vendor’s software. The next step is to digitize these films at a very high spatial resolution (10 -micron pixel size) and to a 16-bit dynamic range. This level of resolution was arrived at by balancing the desire to have no visible loss in resolution against the size of the resulting file. Specifically, changes on the order of 100 microns are difficult to detect with the unaided eye and, since 10X magnification is allowed by some specifications with film radiography, this drove us to at least having a 10-micron resolution. Since this results in a file size of close to a gigabyte per film upon digitization, the team chose not to try for any higher resolution.

The concept in using these reference images is that by starting with an image with 10 micron pixels it is easy to reduce the resolution by steps of 10 microns to get an approximate resolution match to whatever is needed based on detector and geometric magnification. For example if you were operating a detector with a pixel spacing of 139 microns with a geometric magnification of two, you would want to view the reference images at a pixel spacing of 70 microns.

Making the resulting digital images a useful standard requires that there be documentation of when and how to use them. The team is addressing this with usage instructions covering such things as how the digital reference image files are to be loaded, how to select the correct resolution to load and how to normalize the contrast between a production image and a reference image. The instructions also include minimum hardware requirements to view the images, instructions on how the brightness should be set, and how to interpret what is seen. In addition, minimum software requirements are being developed to cover the required operations to use the digital reference images. Finally, restrictions on when the standard can be used, i.e., material thickness limitations and minimum effective resolution of the system, are being documented.

While this seems a fairly simple concept it does get confusing in the details when we try to explain it. To help with this we are developing some demonstration software to demonstrate our concept of how digital reference images would be used and what a radiographer would see in operation.

The last piece to this puzzle is some proof that it works. To demonstrate this we are planning to perform a round robin with industry volunteers in 2004. The basic idea is to take a group of castings with a range of discontinuities and inspect them using film, the E 155 reference radiographs, a suite of digital detectors, and the digital reference images. If the team has done their job well we should not be able to find a statistical difference in how castings are classified between film radiography and digital radiography using the proposed digital reference images.

Conclusions

Traditional film reference radiographs such as ASTM E 155 do not translate directly into the world of digital radiography. Therefore the development of digital reference images is required to capture the savings offered by digital radiography. While there are many details involved in developing digital reference images that will result in similar classification results as traditional film reference radiographs, there are no technical reasons why they cannot be developed and successfully implemented. //

Copyright 2003, ASTM

The work presented in this article is being performed by a team with significant contributions from: Uwe Ewert, Uwe Zscherpel, James Kennedy, Mike Horky, Tom Riechers, Bill Meade, Michael Hutchinson, and The University of Alabama at Birmingham.