CDMAM phantom for Image Quality Assessment and Quality Control in digital mammography: A narrative review.
Authors
Affiliations (2)
Affiliations (2)
- Faculty of Medicine, Department of Medical Imaging Technology, Medical University of Lodz, Lodz, Poland.
- Department of Medical Physics, Copernicus Memorial Hospital in Lodz Comprehensive Cancer Center and Traumatology, Lodz, Poland.
Abstract
The CDMAM (Contrast Detail Mammography) phantom is widely used in European mammography quality control as a subjective observer-based tool for quantitative evaluation of low-contrast detectability. With the evolution of breast imaging technologies and the increasing adoption of automation, a technically focused synthesis addressing phantom design, analysis methodology, metrological performance, and practical limitations is needed. This review provides a concise technical overview of the current state of knowledge on CDMAM-based image quality assessment and quality control in digital mammography. A structured literature review was conducted covering publications from 1998 to 2025. Relevant studies were identified using PubMed, Web of Science, Scopus, SpringerLink, Elsevier ScienceDirect, and IOPscience. More than 60 scientific articles addressing the CDMAM phantom were analyzed, with emphasis on phantom construction and versions (CDMAM 3.4 vs. 4.0), automated analysis software and metrics (including IQF<sub>inv</sub>), dose-image quality relationships, applications beyond 2D full-field digital mammography (FFDM) such as digital breast tomosynthesis (DBT), synthetic 2D imaging (s2D), and contrast-enhanced spectral mammography (CESM), and emerging computational approaches including model observers and artificial intelligence. The literature supports CDMAM as a useful and standardized surrogate of low-contrast detectability under controlled conditions; however, its ability to represent clinical lesion detectability is inherently limited by the use of simple test objects in a uniform background. CDMAM 4.0 demonstrates improved metrological performance relative to CDMAM 3.4, including enhanced sensitivity for the smallest details and reduced measurement uncertainty. Automated analysis substantially improves repeatability and enables harmonized long-term trend monitoring; however, software version dependence can bias IQF<sub>inv</sub> and requires harmonization strategies. The relationship between average glandular dose (AGD) and low-contrast detectability is nonlinear, indicating an optimal dose range beyond which further dose escalation yields limited perceptual benefit. For DBT and s2D, CDMAM-based assessments remain feasible but require careful interpretation due to reconstruction-dependent noise and depth-related effects; in CESM, applicability is primarily limited to the low-energy component. Hybrid frameworks combining CDMAM metrics with model observers (e.g., NPWE, CHO) and AI-based methods enable automated continuous quality monitoring and earlier detection of performance degradation. CDMAM remains a central tool for perceptual image quality evaluation and quality control in digital mammography and provides a practical bridge between physical metrics and detectability-oriented performance assessment. Current evidence supports the preferential use of CDMAM 4.0 in modern quality control programs, emphasizes the need for software harmonization in longitudinal analyses, and highlights the growing role of hybrid CDMAM-model observer-AI pipelines for automated, metrologically consistent quality assurance across emerging breast imaging modalities.