The American National Standard "Calibration and Use of Germanium Spectrometers for the Measurement of Gamma-Ray Emission Rates of Radionuclides" has been reissued as N42.14-1999. The performance tests in it can be used to make sure that a gamma spectroscopy program is set up correctly. The same tests can also be used to verify the improvements made by program developers. However, sometimes the tests in this ANSI standard are not enough. To satisfy certain quality assurance requirements, it is necessary to demonstrate that the results are correct either by hand calculations or by comparing the results to known values.

Introduction

The recently re-issued ANSI standard N42.14-1999, “Calibration and Use of Germanium Spectrometers for the Measurement of Gamma-Ray Emission Rates of Radionuclides” ¹ includes a straightforward set of steps to verify the performance of a software package for gamma spectrum analysis. The major editorial change in this revision consisted of reorganizing and renaming Section 8 "Verification of the Analysis Process" to make this section easier for the user to follow ². This reorganization is believed to place the performance tests in an order more in line with what a user would follow in setting up and performance testing a Ge spectrometer. All tests have been changed to involve the use of commercially available radionuclide sources emitting either single or multiple g rays. The tests include guidance for evaluating automatic peak-finding algorithms, for testing the independence of peak-area from gross peak-height-to-baseline-height ratio, for finding doublets and fitting their areas, for evaluating and correcting for pulse pileup (random summing), for an assessment of the magnitude of coincidence summing, for determining the bias and accuracy of the full-energy peak efficiency, and for nuclide identification and activity calculations. Many of these tests are also described in reference ³.

The ANSI standard is primarily intended to provide a basis for the routine calibration and use of Ge semiconductor detectors for the measurement of gamma-ray emission rates and, thereby, the activities of the radionuclides in a sample. For users of gamma spectrometry systems going through this process helps to verify that the system meets the necessary criteria for the expected needs. In this era of computerized systems, a fair amount of the testing, of course, evaluates the software being used to analyze the spectra. It also measures the user’s grasp of the capabilities of the analysis software, and its inherent restrictions.

While ANSI N42.14 provides guidance on testing the performance of gamma spectroscopy software, it does not address certain aspects of software development. These issues are better addressed in various quality assurance standards ^4,5,6. At the same time, the QA standards are not specific to gamma spectrometry. We have found a combination of the N42.14 and the various QA standards serves us well in developing and verifying software for gamma spectroscopy.

In our previous work ^7,8,9, we have used the ideas presented in the ANSI N42.14 to evaluate various software packages that were available commercially through Canberra and Nuclear Data at the time. Recently, due to the use of certain Canberra software in the analysis of material being sent to the Waste Isolation Pilot Plant (WIPP) in Carlsbad, New Mexico, Canberra was required to demonstrate compliance with the U.S. DOE (Carlsbad Area Office) Quality Assurance Program Document – CAO-94-1012 (Revision 2) 10. The software testing requirements of this document include those of NQA-1 (Supplement 11S-2 - Supplementary Requirements for Computer Program Testing) and NQA-2, Part 2.7.

The testing phase of the CAO quality assurance program is discussed in section 6.5.2.4 of the document, which in part reads:

“Test requirements and acceptance criteria shall be specified, documented, and reviewed and shall be based upon applicable design or other pertinent technical bases. Appropriate tests, such as design-driven tests, requirements-driven tests, hardware integration tests, and in-use tests, shall be controlled. Software testing, using documented test plans, test cases, and test results are the primary methods of software validation.

Design-driven tests shall be used to demonstrate the capability of the software to produce valid results for test problems encompassing the range of intended use as defined by the software documentation. Testing of software used for operational control shall demonstrate the required performance over the entire range of the controlled function or process. Acceptable test methods consist of:  

• Hand calculations
• Calculations using comparable proven problems
• Empirical data and information from confirmed published data and correlations or technical literature
• Comparison with other validated software of similar purpose
• Manual inspections or qualitative checks not involving numerical manipulation (examples include visual inspection of database reformatting or data plotting).”

In this paper we will give examples of how these different QA test methods have been applied during the verification and validation testing of various GENIE-2000 algorithms, and of some problems one encounters during such an exercise. We will also show examples of results obtained for some of the performance tests described in the ANSI N42.14.