Intercomparison of quality control software responses in measuring the modulation transfer function in mammography

Abstract

The National Cancer Institute (INCA) estimated approximately 74,000 new cases of breast cancer in Brazil in 2023, making it the second most common cancer among women. Mammography is the most effective method for early diagnosis, capable of detecting lesions as small as 2 mm before symptoms or dissemination. To ensure its effectiveness, mammography systems must comply with Brazilian standards, such as RDC nº 611 and IN nº 92 from Anvisa, through quality control testing. One of the primary quality tests is the Modulation Transfer Function (MTF) analysis, which evaluates system performance in the spatial frequency domain to ensure diagnostic accuracy. This study assessed the performance of three software tools — ATIA, COQ, and MAMMOQC — in analyzing the MTF of digital mammography systems, focusing on Retrofit DR and CR technologies. Images were acquired using a Siemens Mammomat 3000 NOVA analog mammograph equipped with a Retrofit DR detector and a CR plate. Digital images were analyzed using ATIA, which automatically positions Regions of Interest (ROIs), and the COQ and MAMMOQC plugins in ImageJ, where ROIs were manually selected. MTF values at 10%, 20%, and 50% were compared between the software using paired t-tests with a 5% significance level. Results showed that ATIA and MAMMOQC exhibited higher agreement for 50% MTF, while COQ often produced higher resolution values, particularly at 10% and 20% MTF, highlighting methodological differences among the tools. Statistical analysis revealed significant discrepancies between software results, particularly for the CR system. For the Retrofit DR system, significant variability was observed in comparisons between COQ and MAMMOQC (p = 0.030). For the CR system, all software comparisons showed significant differences, suggesting greater sensitivity to noise and methodological variations at lower MTF percentages. These findings underscore the importance of understanding the strengths and limitations of each software for mammography quality control and emphasize the need for further validation of ATIA as a reliable tool in this field.