Revealing artifacts in contrast-enhanced mammography: prevalence and implications for diagnostic accuracy

Introduction

Contrast-enhanced mammography (CEM) is a relatively new imaging technique, approved by the U.S. Food and Drug Administration (FDA) for breast cancer screening in 2011. Literature reports indicate that CEM sensitivity ranges from 93% to 100% (1,2).

Several studies (3,4) have previously compared the diagnostic performance of CEM and magnetic resonance imaging (MRI), both of which have shown relatively high sensitivity in breast cancer evaluation. In fact, CEM may demonstrate even greater diagnostic accuracy than MRI in certain cases (5).

CEM uses iodinated contrast agents, typically at concentrations of 300–370 mg iodine/mL. The contrast is administered intravenously, usually into the antecubital vein using an automatic injector at a rate of 2–3 mL/s, followed by a saline flush. Vascular access is first checked with a saline bolus. The standard dose is 1.5 mL/kg, with a maximum of 120 mL (6). CEM exams take approximately 15–20 minutes. Two minutes after injection, image acquisition begins, and venous access is maintained for 15 minutes to manage possible delayed contrast reactions. Imaging does not require a fixed order. The ideal window is between 2 and 10 minutes post-injection, which allows sufficient time to capture the four standard views—craniocaudal and mediolateral oblique of each breast—and any additional views requested. Each CEM view includes a low-energy (LE) and a high-energy (HE) image. Exposure time varies from 4 to 10 seconds per view, depending on breast size and technical settings. CEM relies on the photoelectric effect of iodine to highlight areas of contrast uptake. Iodine’s absorption k-edge is 33 keV, which falls within the range of mammographic X-ray beams. Its higher attenuation compared to breast tissue allows tumor enhancement to be visualized. LE images (26–30 kVp) are acquired first and are not influenced by iodine, offering image quality similar to standard 2D digital mammography. HE images (44–49 keV) are acquired next. Here, iodine absorbs X-rays via the photoelectric effect, enhancing tumor contrast (7). While HE images are not visually interpretable, they are processed to create a recombined image showing areas of enhancement. The final CEM result includes both LE and recombined images, improving lesion detectability (8).

Normal fibroglandular and adipose tissues generally show minimal uptake of contrast agents, resulting in effective suppression on the subtracted image. However, similar to breast MRI, benign background parenchymal enhancement (BPE) in CEM can reduce sensitivity in detecting enhancing breast tumors (9-11).

Artifacts are defined as any variation in mammographic density not due to genuine attenuation differences in the breast (12). Radiologists and mammography technologists must be familiar with the range of artifacts that can create false lesions or obscure actual abnormalities. They can lower image quality, mimic or obscure abnormalities, and lead to diagnostic errors. Recognizing artifacts improves the accuracy of mammographic interpretation and prevents misinterpreting artifacts as breast disease (13). The literature classifies artifacts into three main categories: patient-related artifacts, contrast agent-related artifacts, and technical artifacts (13,14).

Our study aimed to conduct a comprehensive analysis of the artifacts and pitfalls associated with CEM, evaluating the incidence of each type. Additionally, we sought to compare detection rates between two operators with different experience levels to clarify the influence of expertise on diagnostic accuracy. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-860/rc).

Methods

Study design

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Ethical review and approval were waived for this study by the institutional review board of University of Eastern Piedmont due to the use of fully anonymized data, which were obtained without any identifiers that could trace back to individual participants. Informed consent was obtained from all subjects involved in the study. Between January 2020 to December 2023, we conducted a retrospective study of 209 patients who underwent CEM examinations. The inclusion criteria were as follows: age over 30 years, no prior adverse reactions to iodinated contrast agents, normal renal function verified by recent creatinine and glomerular filtration rate (GFR) results (within 3 months). Exclusion criteria included age under 30 years, pregnancy, impaired renal function and history of adverse reactions to iodinated contrast (Table 1). Eight patients were excluded due to technical issues with the Picture Archiving and Communication System (PACS), which prevented image retrieval. These issues included data corruption during storage or transfer, system malfunctions leading to incomplete or missing files, and patient studies being inadvertently deleted or misfiled. Additionally, network connectivity failures at the time of image acquisition may have resulted in unsuccessful uploads, rendering the data inaccessible. Despite multiple retrieval attempts and technical support intervention, these images could not be restored, necessitating their exclusion from the study. Thus, the final study group included 201 patients (Figure 1). Each CEM examination from these 201 patients was anonymized and independently reviewed by a breast radiologist with 30 years of experience (referred to as “Master: M”) and a radiology resident with 1 year of experience in breast imaging (designated as “Alumn: A”). The experienced radiologist participating in this study had 5 years of dedicated experience in the interpretation of CEM, with an average reporting volume of approximately 10 CEM cases per week. Assuming 52 working weeks per year and subtracting an estimated 20 cases annually to account for leave or non-reporting periods, the total number of CEM examinations interpreted by this radiologist is approximately 2,400. At the time of the study, the radiology resident had completed 12 months of clinical exposure to CEM, during which an average of 10 cases per week were reviewed. After accounting for 20 missed cases, the total number of CEM examinations interpreted by the resident is estimated to be 500. These quantitative estimates offer a more precise assessment of reader experience than years of training alone, particularly given the relatively recent adoption of CEM in clinical practice.

Figure 1 Flowchart showing selected and excluded study cohort. CEM, contrast-enhanced mammography; PACS, Picture Archiving and Communication System.

Both reviewers assessed the CEM images for artifacts and pitfalls.

CEM technique

CEM was performed using a Hologic Selenia Dimension digital mammography system (Marlborough, Massachusetts, USA). After establishing peripheral venous access, a low-osmolar iodinated contrast agent (iomeprol, 714.4 mg/mL, equivalent to 350 mg iodine/mL) was injected with the breast uncompressed to optimize blood flow. The contrast medium was administered intravenously as a single dose using a power injector (Bracco Injeneering EmpowerCTA^®, Milan, Italy) at a rate of 2–3 mL/s, followed by a 20 mL saline flush. The patient remained seated during the injection, and the contrast volume was calculated at 1.5 mL/kg of body weight, with a maximum dose of 110 mL. Approximately 2 minutes after injection, paired LE and HE images were acquired with the breast compressed, first in the craniocaudal view, followed by the mediolateral oblique view using tomosynthesis. Additional delayed acquisitions in craniocaudal and mediolateral oblique views were obtained within 10 minutes post-injection. These were performed in 2D to limit additional radiation exposure to the patient.

CEM artifacts

Artifacts, which can be challenging to detect, diminish the quality of CEM images and may lead to misinterpretation of genuine pathological conditions. Proper calibration of CEM equipment, accurate patient positioning, adequate breast compression, and careful management of contrast agents are crucial to reducing artifact occurrence in CEM. Artifacts can be grouped into three main categories: patient-related artifacts (e.g., superimposed structures), contrast agent-related artifacts (e.g., contrast splatter, transient contrast retention in veins), and technical artifacts (e.g., air artifacts, ripple effects, halo artifacts, skin line artifacts, negative contrast enhancement, misregistration, axillary line artifacts, aborted studies, ghosting, and miscalibration errors). Table 2 provides strategies for minimizing these artefacts.

Patient-related artifacts

Superimposed structures refer to any object, substance on the skin, or anatomical feature that overlaps the field of view during image acquisition, creating artifacts on the resulting images. These can include the shoulder, chin, nose, opposite breast, or clothing. Additionally, traces of deodorant or hair can resemble microcalcifications, potentially leading to image misinterpretation. The recommended approach is to identify the artifact on the image and perform a new exposure, ensuring that the obstructing object or substance is removed from the X-ray beam path.

Contrast agent-related artifacts

• Contrast agent splatter: contrast medium can splatter onto the breast skin or the compression pads because of glove contamination or directly from the injection, in particular when using a port-a-cath.
  When occurring, white lines or dots are found on recombined images, which can mimic linear non-mass enhancements or microcalcifications. Careful contrast agent manipulation and using clean breast compression paddles can help avoid this artifact.
• Transient retention of contrast medium in veins: contrast enhancement of veins in early acquisition but not visible on the later ones. Usually, it is a transient artifact that resolves by itself (Figure 2).
  

  Figure 2 Cranio-caudal view of left breast showing transient retention of contrast in veins (white arrow).

Technical artifacts

• Air artifact: it refers to black lines that are seen around the area of incomplete contact between skin and compression paddle. It can appear in the axillary area and close to scar tissue. To solve the problem, repositioning of the breast and correct compression should be performed (Figure 3).
  

  Figure 3 Medio-lateral-oblique view of right breast highlighting air artifact (white arrow).
• Ripple artifact: it represents thin black and white parallel lines on recombined images. They are due to patient movement in-between the acquisition of low energy and high energy images. They can also be caused by cardiac pulsation transmitted through the chest wall. It is best seen on the mediolateral oblique view. The solution stands on explaining to the patient to stay still during image acquisitions (Figure 4).
  

  Figure 4 Medio-lateral-oblique view of right and left breasts demonstrating ripple artifact (white arrows).
• Halo artifact: it is caused by uneven breast tissue thickness which scatters radiation from the chest to the nipple. The result is a white line that parallels the breast gland profile. To overcome this artifact repositioning of the patient or new equipment software should be taken into consideration (Figure 5).
  

  Figure 5 Medio-lateral-oblique and cranio-caudal views of right breast showing halo artifact (white arrows).
• Skin line artifact: due to scattered radiation from the nonuniform skin thickness across the breast, a thin white line may appear artifactually on recombined images, simulating skin enhancement. This artifact results from image filtration applied to equalize breast thickness. Uneven skin thickness or texture, particularly at the lateral breast edge, can contribute to this effect. To avoid misinterpretation, it is essential to assess skin thickness on LE images without relying solely on recombined images (Figure 6).
  

  Figure 6 Medio-lateral-oblique views of right and left breasts showing skin line artifact (white arrows).
• Negative contrast enhancement: negative contrast enhancement is not strictly an artifact but rather a characteristic feature of CEM that can serve as a diagnostic sign for breast lesions. It occurs when lesions appear darker than the surrounding parenchyma on recombined images. This phenomenon is commonly observed in cysts, where the lack of enhancement results in lower density compared to the background, sometimes creating a thin, smooth enhancement of the cyst wall. Coarse calcifications also typically exhibit negative contrast enhancement. Additionally, post-biopsy hematomas at the lesion site may cause negative enhancement, occasionally obscuring the contrast uptake of breast cancer and leading to the “eclipse” or “crescent” sign. Importantly, malignant lesions can also present with negative contrast enhancement, often with an irregular or thick rim. Therefore, careful interpretation of CEM findings is essential to ensure accurate diagnosis (Figure S1).
• Misregistration artifact: alternating dark and bright thin lines close to surgical clips, markers, devices like pace-makers. Sometimes it can be seen near microcalcifications too. It is due to slight patient movements in-between low energy and high energy acquisitions. Patients should reduce movements to avoid this artifact (Figure S2).
• Axillary line: it refers to horizontal lines in the axilla due to improperly sized compression paddles. Changing the size of the paddles solves the problem (Figure S3).
• Aborted study: a new acquisition must be taken since recombined images are not generated. Only vertical black lines are visible on low energy acquisitions.
• Ghosting artifact: a superimposed image is latent from prior exposure. Recalibration is needed.
• Miscalibration artifact: when equipment is wrongly calibrated rings, lines or streak patterns are detected. Correct equipment calibration must be performed (Figure S4).

CEM pitfalls

Pitfalls include technical challenges and diagnostic limitations related to the CEM technique. False-negative results can arise from lesions that are outside the field of view or hidden by extreme BPE. False-positive findings typically occur when non-malignant findings show contrast enhancement. Thus, pitfalls can be listed as: blind areas on CEM, breast implants, BPE, nipple enhancement, false-negative malignant lesions and false-positive benign lesions. Tricks to avoid pitfalls are listed in Table 3.

• Blind areas on CEM: lesions not included in the field of view. Blind areas refer to axillary tails, areas of breast high-positioned on the chest wall or in the intermammary/inframammary fold. Sometimes it is necessary to acquire additional projections to assess these areas. The radiologist has to be extra careful when reporting CEM, actively searching for lesions visible only in one view and not visible in the second, and palpable masses. Ultrasound and MRI can be requested for studying these particular areas.
• Breast implants: they are generally considered a relative contraindication for CEM but their presence does not entirely rule out the procedure. Special consideration techniques may be required (i.e., Eklund maneuver) to ensure effectiveness of the imaging. Breast implants can obscure a mass on low energy images or create black and white lines artifacts. Free silicone leak may also lead to negative enhancement sign, obscuring a lesion (Figure S5).
• v BPE: enhancement of normal fibroglandular tissue can prevent the radiologist from seeing a tumor, hence causing the reduction of CEM diagnostic accuracy. BPE is classified according to the Breast Imaging Reporting and Data System (BI-RADS) lexicon as per MRI: minimal, mild, moderate, marked. Unlike MRI, it has not been proved any correlation of BPE and menstrual cycle at CEM (11) (Figure S6).
• Nipple enhancement: nipple or Montgomery glands enhancement can be visualized. The tip to understand if it is a normal nipple CE is to check if it is symmetric or not. Eventually inspection of the skin should be performed (15) (Figure S7).
• False-negative malignant lesions: it can occur in low-grade small tumors, lobular carcinomas, mucinous and encapsulated papillary carcinomas, and ductal carcinoma in situ (DCIS). If microcalcifications do not enhance, it does not prove the absence of an eventual tumor. If they do enhance, that should be considered extremely suspicious (15-18).
• False-positive benign lesions: any benign lesion and/or structure (dermal lesions, intramammary lymph nodes, fibroadenoma, phyllodes tumor, papilloma, pseudoangiomatous stromal hyperplasia, fat necrosis, sclerosing adeno-sis, radial scar, infectious, mastitis, abscess) showing CE.

Statistical analyses

The incidence of each artifact type was calculated as a percentage, based on the total number of patients (n=201). For each operator, the number of cases was divided by the total patient count to yield the incidence rate, expressed as a percentage: incidence rate = number of cases/201 ×100%.

The incidence for each artifact type was compared between the two operators to identify differences in artifact detection, both overall and within specific artifact categories.

Cohen’s Kappa was used to assess the level of agreement between the two operators for each artifact category. Kappa values were interpreted according to the following scale:

• 0–0.20: slight agreement;
• 0.21–0.40: fair agreement;
• 0.41–0.60: moderate agreement;
• 0.61–0.80: substantial agreement;
• 0.81–1.00: almost perfect agreement.

Spearman’s rank correlation was used to assess the overall correlation between the operators’ artifact detection rates. Spearman’s correlation was computed across all artifact categories to determine the degree of concordance between the two operators’ rankings.

All statistical analyses were performed using Python, with the scipy. Stats library used for calculating Cohen’s Kappa and Spearman’s rank correlation. Descriptive statistics such as incidence rates were computed using standard formulas in Microsoft Excel. The results were interpreted in line with commonly accepted guidelines for Cohen’s Kappa and Spearman’s rank correlation, and statistical significance was assessed with a threshold of P<0.05.

Results

Patients’ population

The age distribution of the 201 female patients in this cohort exhibits a mean of 58.03 years (standard deviation ±12.29 years).

Artifacts’ incidence

The analysis of the results is divided into several categories: patient-related artifacts, contrast agent-related artifacts, technical artifacts, and CEM pitfalls. The following section provides an in-depth statistical analysis of these findings.

Patient-related artifacts

The incidence of patient-related artifacts was markedly higher for the less experienced operator (A), who recorded 13 instances, representing 6.5% of the patients, compared to only 1 case identified by the more experienced operator (M), an incidence rate of 0.5%.

Contrast agent-related artifacts

In the case of contrast agent-related artifacts, specifically transient retention of contrast medium, M identified 2 instances (1%), while A identified 1 (0.5%). No cases of contrast agent splatter were identified by either operator.

Technical artifacts

This category exhibited the highest incidence of artifacts, with substantial variability between the two operators.

• Air artifact: M identified 175 cases (87.1%), while A identified 167 cases (83.1%).
• Ripple artifact: the incidence of ripple artifacts was higher for M, with 92 cases (45.8%), compared to 82 cases for A (40.8%).
• Halo artifact: A identified slightly more halo artifacts (124 cases, 61.7%) compared to M (119 cases, 59.2%).
• Skin line artifact: A identifying 131 cases (65.2%) compared to only 88 cases (43.8%) by M.
• Negative contrast enhancement: M detected a significantly higher number of negative contrast enhancement artifacts, with 96 cases (47.8%), compared to only 30 cases (14.9%) detected by A.
• Misregistration artifact: A detected 73 cases (36.3%), whereas M identified 51 cases (25.4%).
• Axillary line artifact: both operators detected relatively few instances of axillary line artifacts, with M identifying 11 cases (5.5%) and A identifying 9 cases (4.5%).
• Other technical artifacts: there were no cases of aborted studies or ghosting artifacts reported by either operator, and only 1 case of miscalibration artifact was identified by M (0.5%), with no corresponding case identified by A.

CEM pitfalls

CEM pitfalls include areas of complexity and potential error in CEM. Here, we observe some substantial differences between the operators:

• Blind areas on CEM: A identified a notably higher number of blind areas, with 55 cases (27.4%) compared to only 4 cases (2%) detected by M.
• Breast implants: both operators detected only 1 instance of breast implants.
• BPE: the detection was similar between the two operators, with M identifying 78 cases (38.8%) and A identifying 81 cases (40.3%).
• Nipple enhancement: detected in 38 cases by M (18.9%) and 55 cases by A (27.4%).
• False-negative malignant lesions: M identified 4 cases (2%) and A identifying 3 cases (1.5%).
• False-positive benign lesions: A identified 42 cases of false-positive benign lesions (20.9%), while M detected 27 cases (13.4%).

In Table 4, the incidence counts and the percentage of total patients (n=201) affected by each artifact, as evaluated by both operators (M and A), is presented.

Statistical concordance

To assess the overall concordance between the two operators, we employed the Cohen’s Kappa coefficient to quantify agreement. The resulting Kappa coefficient of 0.59 suggests moderate agreement between the two operators across the different categories of artifacts, particularly in more easily recognizable artifact types like air and halo artifacts. However, substantial variability in other categories (e.g., skin line artifacts, blind areas on CEM).

In addition, Spearman’s rank correlation analysis indicated a moderately strong correlation between the operators (Spearman’s rho =0.64), suggesting that while there is overall agreement in the relative ranking of artifact incidence, specific artifact categories show notable differences in detection.

Discussion

CEM is a relatively new diagnostic imaging technique that gained FDA approval for breast cancer screening in 2011. Its reported sensitivity ranges from 93% to 100%, according to existing literature (1,2). However, like other imaging methods, CEM is not free from artifacts, which can affect diagnostic accuracy. These artifacts are typically categorized into three main types: patient-related artifacts, contrast agent-related artifacts, and technical artifacts. Additionally, there are various technical challenges and diagnostic pitfalls that can lead to false positives or false negatives (16-20).

Recognizing artifacts in CEM has a direct and substantial impact on diagnostic accuracy and patient management. Accurate identification of artifacts—such as motion-related blur, implant-associated distortions, or enhancement due to technical factors—can prevent both false-positive and false-negative interpretations. This is essential to avoid unnecessary recalls, additional imaging, or invasive procedures such as biopsies prompted by misinterpreted findings. Conversely, failure to recognize a true enhancement masked by an artifact may lead to missed malignancies. Furthermore, understanding artifact patterns helps to reduce the need for technical repeats, thus improving workflow efficiency and minimizing patient discomfort and resource utilization. From a reporting perspective, correct recognition of artifacts ensures appropriate BI-RADS classification, contributing to consistent clinical recommendations and standardized patient care. Importantly, this competency enhances diagnostic confidence, particularly among less experienced readers, underscoring the value of dedicated training and education in artifact recognition as CEM continues to be integrated into routine breast imaging practice (20).

In our study, a total of 201 CEM scans were revised by two radiologists: one with 30 years of experience in breast imaging and 5 years of dedicated experience in the interpretation of CEM (M) and a resident with 1 year of breast radiology experience (A). The incidence of patient-related artifacts was notably higher for the A compared to the M, with rates of 6.5% and 0.5%, respectively. This suggests that less experienced operators may be more prone to attributing imaging abnormalities to patient movement or positioning errors, emphasizing the need for better standardization and training in artifact identification.

Regarding contrast agent-related artifacts, there was a high degree of agreement between the M and A, indicating that these artifacts are relatively straightforward to diagnose. On the other hand, technical artifacts were the most frequent, showing considerable variability between the two examiners.

In terms of Air, Halo, and Axillary Line artifacts, both operators showed good concordance in detecting them, indicating that these artifacts are relatively easy to identify for both novice and experienced radiologists. The moderate difference in the detection of Ripple artifacts may be due to some interpretation variability, particularly when distinguishing minor rippling from other types of artifacts.

Although both operators identified a significant number of Misregistration artifacts, the higher detection rate by the A (22 more cases than the M) could suggest overdiagnosis, with the A potentially misinterpreting other technical issues as misregistration artifacts.

The greatest discrepancies in artifact incidence were observed for Skin Line and Negative Contrast Enhancement artifacts. The higher discrepancy in Skin Line artifacts may reflect over-detection by the less experienced operator, who might confuse other irregularities with skin line artifacts. Enhanced training could address this issue. The large gap in the detection of Negative Contrast Enhancement artifacts suggests that recognizing this artifact requires a higher level of expertise.

Neither operator reported any cases of aborted studies or ghosting artifacts, and only one case of miscalibration was identified by the M, with none reported by the A. This suggests that these artifacts are relatively rare in the studied population.

Substantial differences were also noted in the detection of CEM pitfalls. For Blind Areas, the large discrepancy between the operators may reflect overestimation by the less experienced operator or underestimation by the more experienced one, highlighting the need for better guidelines and training in identifying blind spots during imaging. The A’s higher detection rate of nipple enhancement may indicate slight overdiagnosis or a more conservative approach by the M in labeling it as an artifact. The difference in False-Positive incidence between the operators may reflect the tendency of the less experienced radiologist to overestimate benign findings as positive, emphasizing the need for refined interpretation skills through experience.

The interpretation of breast implants, BPE, and false-negative malignant lesions showed relatively low discordance, indicating good agreement between the operators in these critical categories.

Furthermore, as this is a retrospective study, no alterations in patient management were implemented. The CEM examinations had initially been interpreted by breast imaging specialists who correctly identified eventual cancers and artifacts and ensured appropriate patient management from the outset. Importantly, none of the cases in this study were assigned a BI-RADS 0 category, reflecting that all examinations were considered diagnostically conclusive at the time of reporting.

This study has several potential limitations. The primary limitation of this study is the use of only one experienced radiologist and one resident to review the cases. This may not reflect the variability in interpretation that could be observed across a broader group of readers. Therefore, results should not be generalized, and future studies including a larger and more diverse sample of readers are needed to validate our findings. Another limitation is the relatively small sample size, which may not provide consistent statistical power. Additionally, CEM images tend to be more easily interpreted by radiologists familiar with conventional mammography.

Although this study serves as an investigation into CEM artifact characterization and potential pitfalls, the small sample size calls for cautious interpretation of the findings. Nonetheless, it provides valuable insights into CEM artifacts and strategies for their mitigation. Further research is needed to conduct a more comprehensive evaluation of these artifacts’ prevalence and their impact on clinical practice.

Conclusions

This study emphasizes the importance of artifacts recognitions CEM to ensure diagnostic precision and proper patient care. Variances between senior and junior readers reinforce the requirement for standardized education and uniform interpretation criteria to minimize variability and incorrect classification. Despite the limitation of a limited number of samples as well as the use of only two readers, this study offers insightful clues to the prevalence and nature of CEM artifacts. Further multicenter studies with larger populations are necessary to authenticate the findings as well as to calibrate systems of classification. Eventually, enhanced education and awareness about CEM artifacts will improve diagnostic confidence, reduce false assumptions, as well as achieve consistent efficient practice in the field of breast imaging.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-860/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-860/dss

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-860/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Ethical review and approval were waived for this study by the institutional review board of University of Eastern Piedmont due to the use of fully anonymized data, which were obtained without any identifiers that could trace back to individual participants. Informed consent was obtained from all subjects involved in the study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Yakoumakis E, Tzamicha E, Dimitriadis A, Georgiou E, Tsapaki V, Chalazonitis A. Dual-energy contrast-enhanced digital mammography: patient radiation dose estimation using a Monte Carlo code. Radiat Prot Dosimetry 2015;165:369-72. [Crossref] [PubMed]
2. Fallenberg EM, Dromain C, Diekmann F, Engelken F, Krohn M, Singh JM, Ingold-Heppner B, Winzer KJ, Bick U, Renz DM. Contrast-enhanced spectral mammography versus MRI: Initial results in the detection of breast cancer and assessment of tumour size. Eur Radiol 2014;24:256-64. [Crossref] [PubMed]
3. Luna L, Liao L, Germaine P, Tinney E. Abstract P1-02-06: Retrospective comparison of sensitivity and positive predictive value (PPV) of contrast enhanced spectral mammography (CESM) to contrast enhanced breast MRI (BMRI) in 50 malignant breasts. Cancer Res 2015;75:2-6.
4. Łuczyńska E, Heinze-Paluchowska S, Hendrick E, Dyczek S, Ryś J, Herman K, Blecharz P, Jakubowicz J. Comparison between breast MRI and contrast-enhanced spectral mammography. Med Sci Monit 2015;21:1358-67. [Crossref] [PubMed]
5. Tardivel AM, Balleyguier C, Dunant A, Delaloge S, Mazouni C, Mathieu MC, Dromain C. Added Value of Contrast-Enhanced Spectral Mammography in Postscreening Assessment. Breast J 2016;22:520-8. [Crossref] [PubMed]
6. Zanardo M, Cozzi A, Trimboli RM, Labaj O, Monti CB, Schiaffino S, Carbonaro LA, Sardanelli F. Technique, protocols and adverse reactions for contrast-enhanced spectral mammography (CESM): a systematic review. Insights Imaging 2019;10:76. [Crossref] [PubMed]
7. Jochelson MS, Lobbes MBI. Contrast-enhanced Mammography: State of the Art. Radiology 2021;299:36-48. [Crossref] [PubMed]
8. Neeter LMFH, Raat HPJF, Alcantara R, Robbe Q, Smidt ML, Wildberger JE, Lobbes MBI. Contrast-enhanced mammography: what the radiologist needs to know. BJR Open 2021;3:20210034. [Crossref] [PubMed]
9. Xiang W, Rao H, Zhou L. A meta-analysis of contrast-enhanced spectral mammography versus MRI in the diagnosis of breast cancer. Thorac Cancer 2020;11:1423-32. [Crossref] [PubMed]
10. DeMartini WB, Liu F, Peacock S, Eby PR, Gutierrez RL, Lehman CD. Background parenchymal enhancement on breast MRI: impact on diagnostic performance. AJR Am J Roentgenol 2012;198:W373-80. [Crossref] [PubMed]
11. Giess CS, Yeh ED, Raza S, Birdwell RL. Background parenchymal enhancement at breast MR imaging: normal patterns, diagnostic challenges, and potential for false-positive and false-negative interpretation. Radiographics 2014;34:234-47. [Crossref] [PubMed]
12. Sensakovic WF, Carnahan MB, Czaplicki CD, Fahrenholtz S, Panda A, Zhou Y, Pavlicek W, Patel B. Contrast-enhanced Mammography: How Does It Work? Radiographics 2021;41:829-39. [Crossref] [PubMed]
13. Hogge JP, Palmer CH, Muller CC, Little ST, Smith DC, Fatouros PP, de Paredes ES. Quality assurance in mammography: artifact analysis. Radiographics 1999;19:503-22. [Crossref] [PubMed]
14. Geiser WR, Haygood TM, Santiago L, Stephens T, Thames D, Whitman GJ. Challenges in mammography: part 1, artifacts in digital mammography. AJR Am J Roentgenol 2011;197:W1023-30. [Crossref] [PubMed]
15. Sogani J, Morris EA, Kaplan JB, D'Alessio D, Goldman D, Moskowitz CS, Jochelson MS. Comparison of Background Parenchymal Enhancement at Contrast-enhanced Spectral Mammography and Breast MR Imaging. Radiology 2017;282:63-73. [Crossref] [PubMed]
16. Zamora K, Allen E, Hermecz B. Contrast mammography in clinical practice: Current uses and potential diagnostic dilemmas. Clin Imaging 2021;71:126-35. [Crossref] [PubMed]
17. Lalji UC, Houben IP, Prevos R, Gommers S, van Goethem M, Vanwetswinkel S, Pijnappel R, Steeman R, Frotscher C, Mok W, Nelemans P, Smidt ML, Beets-Tan RG, Wildberger JE, Lobbes MB. Contrast-enhanced spectral mammography in recalls from the Dutch breast cancer screening program: validation of results in a large multireader, multicase study. Eur Radiol 2016;26:4371-9. [Crossref] [PubMed]
18. James JJ, Tennant SL. Contrast-enhanced spectral mammography (CESM). Clin Radiol 2018;73:715-23. [Crossref] [PubMed]
19. Jochelson MS, Pinker K, Dershaw DD, Hughes M, Gibbons GF, Rahbar K, Robson ME, Mangino DA, Goldman D, Moskowitz CS, Morris EA, Sung JS. Comparison of screening CEDM and MRI for women at increased risk for breast cancer: A pilot study. Eur J Radiol 2017;97:37-43. [Crossref] [PubMed]
20. Polat DS, Evans WP, Dogan BE. Contrast-Enhanced Digital Mammography: Technique, Clinical Applications, and Pitfalls. AJR Am J Roentgenol 2020;215:1267-78. [Crossref] [PubMed]