From film to tomosynthesis: A short history of screening mammography

INTRODUCTION

One hundred years ago, in 1925, Dr. Malvern B. Clopton described the challenges of diagnosing breast cancer in his Radiology article, “The Difficulty of Diagnosis of Carcinoma of the Breast.”^[1] He noted that “early cancer cannot be diagnosed by palpation or inspection,“ highlighting the limitations of physical examination in distinguishing benign from malignant breast lesions and the reliance on surgical excision as the only definitive diagnostic tool at that time. In the 1950s, it was widely believed that breast cancer was systemic before it could be clinically detected, leading to an emphasis on systemic therapies.^[2,3] The paradigm shifted in the 1960s with the Health Insurance Plan (HIP) of Greater New York study and subsequent randomized controlled trials, which demonstrated that screening mammography reduces breast cancer-related mortality. Since then, breast imaging has become central to early breast cancer detection. This trajectory from surgical diagnosis to imaging-based detection represents one of the most important advances in the history of breast cancer care.

ADVANCES IN BREAST IMAGING TECHNOLOGY

The field of breast imaging has advanced through successive technological innovations: from direct-exposure film mammography, to xeromammography, to screen-film mammography, to full-field digital mammography (FFDM), and most recently to digital breast tomosynthesis (DBT).^[4] Early direct-exposure film studies required relatively high radiation doses to the breast, raising safety concerns, and image quality was limited before dedicated breast equipment was developed. Xeromammography, introduced in the 1970s, provided superior image contrast and edge definition compared with conventional film, but its higher doses, lengthy processing, and costly specialized equipment constrained widespread adoption. It was ultimately supplanted by screen-film mammography, which achieved lower dose with high contrast and spatial resolution.

The development of dedicated mammography systems that provided more complete breast imaging, along with advances in image processing, steadily improved image quality and cancer detection. Screen-film mammography, introduced in the 1960s, became the clinical standard for several decades by offering a practical balance of image quality, lower radiation dose, and reproducibility. Despite these advantages, screen-film mammography had important limitations, including a narrow exposure latitude, reduced sensitivity for the detection of breast cancers in dense breast tissue, and dependence on film processing, which introduced variability in image quality. In addition, maintaining large film libraries for storage and retrieval added logistical complexity. These shortcomings paved the way for the adoption of FFDM, which was approved by the United States Food and Drug Administration (FDA) in 2000. Digital systems allowed images to be stored, transmitted, and manipulated electronically; offered improved contrast resolution, particularly in dense breast tissue; eliminated film processing variability; and enabled the use of computer-aided detection. Still, digital mammography was limited by its two-dimensional nature, as overlapping tissue could obscure or mimic lesions.

DBT, developed in the 1990s and first approved by the FDA in 2011, was designed to address this challenge by creating quasi-three-dimensional reconstructions from multiple low-dose projections, thereby reducing the masking effect of superimposed normal tissues.^[5] Although interpretation time for combined DBT and 2D digital mammography is longer than for 2D mammography alone, and the technology requires additional equipment and storage, DBT markedly reduces the effect of tissue overlap. By providing thin “slices” through the breast, DBT decreases the number of cancers hidden by superimposed tissues and minimizes confusing summation shadows. As a result, DBT detects more cancers—often at an earlier, more treatable stage—while reducing recall rates. It has become the current standard of care for screening mammography in the United States.

EVIDENCE IN SUPPORT OF SCREENING MAMMOGRAPHY

The modern evidence base for screening mammography was established by the landmark HIP of Greater New York randomized controlled trial, which enrolled nearly 62,000 women aged 40–64 years between 1963 and 1966.^[6] At 10 years of follow-up, women assigned to annual screening with mammography and clinical breast examination had a 30% reduction in breast cancer-related mortality compared with those in the control group. The Breast Cancer Detection Demonstration Project, launched in the United States in the 1970s, subsequently provided large-scale observational data in more than 280,000 women and confirmed the feasibility of implementing population-based screening in community practice.^[7] Multiple European randomized trials, including the Swedish Two-County, Malmö, Stockholm, Gothenburg, and Edinburgh studies, further validated and extended these findings, with long-term follow-up of the Swedish Two-County Trial showing a 30% mortality reduction at 29 years.^[8] Randomized controlled trials have conclusively shown that screening saves lives, though by design they likely underestimate the true benefit because of crossover, noncompliance, and limited follow-up durations.

Subsequent trials have evaluated advances in mammographic technology to determine whether newer modalities improve diagnostic performance. The Digital Mammographic Imaging Screening Trial (DMIST), which randomized more than 49,000 women at 33 U.S. and Canadian sites, compared digital mammography with conventional screen-film mammography.^[9] While overall diagnostic accuracy was similar, digital mammography demonstrated significantly higher accuracy among women under 50 years of age, those who were pre- or perimenopausal, and those with dense breast tissue. However, the DMIST trial had important methodological limitations, including restricted access to resources for optimizing digital mammography, which may have affected comparative performance outcomes.^[10] More recently, the Tomosynthesis Mammographic Imaging Screening Trial (TMIST) was designed to compare DBT with FFDM to determine whether combined DBT and digital mammography reduce the incidence of advanced breast cancer. While TMIST represents a major effort to study modern screening technology, concerns have been raised regarding aspects of its design and implementation that may limit its ability to fully capture the clinical advantages of DBT.^[11]

THE QUALITY REVOLUTION

While randomized trials demonstrated the effectiveness of screening mammography, its impact in everyday practice depended on parallel efforts to standardize quality and ensure uniform performance across clinical settings. In 1992, the United States Congress enacted the Mammography Quality Standards Act (MQSA), which established a federal regulatory framework to ensure the safety and reliability of mammography nationwide. MQSA mandated accreditation of facilities, certification by the FDA, annual inspections, and ongoing quality control of equipment and personnel. By setting minimum performance standards, MQSA reduced variability in image quality, strengthened training requirements for technologists and radiologists, and ensured that women across the United States had access to mammography performed at a consistently high level.

Complementing these regulatory measures, the American College of Radiology (ACR) introduced the Breast Imaging Reporting and Data System (BI-RADS) in 1993. BI-RADS provided a standardized lexicon for describing mammographic findings, a structured set of assessment categories, and explicit management recommendations linked to each category. This uniformity enhanced communication between radiologists and referring providers, improved consistency in patient care, and facilitated systematic auditing of performance metrics such as recall rates, cancer detection rates, and positive predictive values. Over time, BI-RADS has expanded beyond mammography to encompass breast ultrasound and breast magnetic resonance imaging (MRI). The forthcoming sixth edition of the BI-RADS Atlas will provide updated terminology and guidance to reflect advances in breast imaging practice and technology.

CONTROVERSIES IN SCREENING

The age at which to initiate screening mammography has long been one of the most debated topics in breast cancer prevention. For many years, the U.S. Preventive Services Task Force (USPSTF) recommended routine screening beginning at age 40. However, in 2009, the USPSTF revised its guidelines—without clear scientific justification—to recommend starting screening at age 50, with individualized decision-making for women in their 40s. These recommendations were not aligned with those of the ACR and the Society of Breast Imaging (SBI), both of which have consistently supported initiating annual screening at age 40. In 2023, the USPSTF issued updated draft guidelines lowering the recommended starting age for biennial screening to 40, reflecting both the accumulated evidence from randomized trials and the rising incidence of breast cancer among women in their 40s.

The frequency of screening remains another point of debate. No randomized controlled trial has directly compared annual versus biennial mammography, leaving modeling and observational studies as the primary sources of evidence. These analyses consistently suggest that annual screening prevents more breast cancer-related deaths than biennial screening. For example, modeling by the National Cancer Institute’s Cancer Intervention and Surveillance Modeling Network (CISNET) estimates that annual screening from ages 40 to 79 reduces breast cancer mortality by approximately 42%, whereas biennial screening from ages 50 to 74 reduces mortality by about 30%.^[12]

Another area of ongoing controversy is “overdiagnosis,” which is often cited as a potential harm of screening. It is defined as the detection of cancers that would not have become clinically significant within a woman’s lifetime. Reported estimates vary considerably, influenced by study design, analytic assumptions, and duration of follow-up. Interpretation is further complicated by the potential conflation of overdiagnosis with the detection of indolent but clinically relevant disease. In addition, many studies evaluating overdiagnosis include cases of ductal carcinoma in situ (DCIS). Although there are ongoing questions regarding the clinical significance and optimal management of DCIS, these issues are distinct from the consideration of overdiagnosis in invasive breast cancer, for which the evidence supporting high rates remains limited. Although rare cases of clinically apparent cancers have regressed without treatment—sometimes even progressing to metastatic disease—no mammographically detected invasive breast cancer has been documented to regress spontaneously.^[13] The most methodologically rigorous analyses estimate that true overdiagnosis from screening mammography is likely less than 10% and possibly as low as 1%.

FUTURE DIRECTIONS

As screening mammography enters its second century, its future will be shaped by advances in artificial intelligence (AI), multimodality imaging, and increasingly personalized approaches to care. AI holds promise for improving quality, enhancing efficiency, and supporting more consistent interpretation, though questions remain about optimal integration into clinical workflows and equitable performance across diverse populations [Figure 1]. Complementary screening strategies, including the combined use of DBT and ultrasound, in addition to MRI and contrast-enhanced mammography for women at increased risk, are also evolving. Efforts to personalize screening by incorporating individual risk factors such as breast density, family history, and genetics continue to expand; however, most breast cancers remain sporadic, occurring in women without identifiable risk factors. Together, these emerging innovations—AI, risk-adapted screening, and multimodality imaging—represent the next stage in the evolution of breast cancer screening, building on a century of progress to further refine early detection and improve outcomes for women.

Close

Figure 1:

A 72-year-old woman diagnosed with grade 2 invasive lobular carcinoma (ILC). (a) The screening mammogram revealed an area of architectural distortion in the medial right breast at posterior depth (white arrow), which was identified by the radiologist. The patient was subsequently diagnosed with grade 2 ILC. (b) On retrospective review, the artificial intelligence (AI) algorithm also detected and correctly localized the same finding, with a score of 83/100 (black circle). (c-d) On retrospective review of the patient’s mammogram from the prior year, which was interpreted as normal by the radiologist, the AI algorithm (Genius AI^® Detection 2.0 software; Hologic, Inc) detected and correctly localized the area of architectural distortion, assigning a score of 81/100 (black circle).

Export to PPT