Hubbry Logo
BI-RADSBI-RADSMain
Open search
BI-RADS
Community hub
BI-RADS
logo
7 pages, 0 posts
0 subscribers
Be the first to start a discussion here.
Be the first to start a discussion here.
BI-RADS
BI-RADS
BI-RADS
View on Wikipedia
from Wikipedia

The Breast Imaging-Reporting and Data System (BI-RADS) is a quality assurance tool originally designed for use with mammography. The system is a collaborative effort of many health groups but is published and trademarked by the American College of Radiology (ACR).

The system is designed to standardize reporting and is used by medical professionals to communicate a patient's risk of developing breast cancer, particularly for patients with dense breast tissue. The document focuses on patient reports used by medical professionals, not "lay reports" that are provided to patients.

Published documents

[edit]

The BI-RADS is published by ACR in the form of the BI-RADS Atlas. As of 2013 the Atlas is divided into three publications:

  • Mammography, Fifth Edition
  • Ultrasound, Second Edition
  • MRI, Second Edition

Assessment categories

[edit]

While BI-RADS is a quality control system, in day-to-day usage the term BI-RADS refers to the mammography assessment categories. These are standardized numerical codes typically assigned by a radiologist after interpreting a mammogram. This allows for concise and unambiguous understanding of patient records between multiple doctors and medical facilities.[1]

The assessment categories were initially developed for mammography and later adapted for use with MRI and ultrasound findings. The summary of each category, given below, is nearly identical for all three modalities.

Category 6 was added in the 4th edition of the BI-RADS.

BI-RADS assessment categories are:[2]

  • 0: Incomplete
  • 1: Negative
  • 2: Benign
  • 3: Probably benign
  • 4: Suspicious
  • 5: Highly suggestive of malignancy
  • 6: Known biopsy-proven malignancy

An incomplete (BI-RADS 0) classification warrants either an effort to ascertain prior imaging for comparison, or to call the patient back for additional views and/or higher quality films. A BI-RADS classification of 4 or 5 warrants biopsy to further evaluate the offending lesion.[3] Some experts believe that the single BI-RADS 4 classification does not adequately communicate the risk of cancer to doctors and recommend a subclassification scheme:[4]

  • 4A: low suspicion of malignancy, about > 2% to ≤ 10% likelihood of malignancy
  • 4B: intermediate suspicion of malignancy, about > 10% to ≤ 50% likelihood of malignancy
  • 4C: moderate concern, but not classic for malignancy, about > 50% to < 95% likelihood of malignancy

Breast composition categories

[edit]

As of the BI-RADS 5th edition:[5]

  • a. The breasts are almost entirely fatty
  • b. There are scattered areas of fibroglandular density
  • c. The breasts are heterogeneously dense, which may obscure small masses
  • d. The breasts are extremely dense, which lowers the sensitivity of mammography

Automated extraction

[edit]

Automatic parsers have been developed to automatically extract BI-RADS features,[6][7] categories[8] and breast composition[9] from textual mammography reports.

There is also an automatic parser available for BI-RADS final category inference by parsing only the semi-formatted finding section of the textual mammography report.[10]

References

[edit]
[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia

BI-RADS

View on Grokipedia
from Grokipedia
The Breast Imaging Reporting and Data System (BI-RADS) is a standardized framework developed by the American College of Radiology (ACR) to provide consistent terminology, structured reporting, and risk assessment for breast imaging examinations, including , , and (MRI), with the goal of enhancing communication among healthcare providers, reducing interpretive variability, and guiding clinical management decisions to improve patient outcomes. Initiated in 1986 as a response to inconsistencies in reporting and , BI-RADS was first released in 1993 and has since evolved through multiple editions, with the fifth edition published in 2013 incorporating updates for emerging technologies like digital breast tomosynthesis and contrast-enhanced . A sixth edition is currently under development as of 2025. The system extends beyond to include dedicated lexicons for and MRI (both introduced in 2003), ensuring uniform descriptors across modalities for features such as breast density, mass characteristics (e.g., , margins, orientation), calcifications, and associated findings like asymmetries or distortions. At its core, BI-RADS employs a final assessment category scale from 0 to 6 to convey the likelihood of and recommend next steps: category 0 indicates an incomplete assessment requiring additional ; 1 signifies a negative ; 2 denotes benign findings; 3 suggests probably benign lesions with less than 2% risk, typically followed by short-interval ; 4 indicates suspicious abnormalities subdivided into 4A (2–10% risk), 4B (10–50% risk), and 4C (50–95% risk), prompting ; 5 represents highly suggestive of with over 95% risk; and 6 confirms known , often post-. This categorical structure, combined with over 700 illustrative clinical images and frequently asked questions in the BI-RADS Atlas, supports , facilitates data collection for research, and promotes equitable and diagnosis worldwide, with translations available in eight languages.

Overview

Definition and Purpose

The Breast Imaging Reporting and Data System (BI-RADS) is a and reporting tool developed by the American College of Radiology (ACR) to standardize the terminology, structure, and classification of findings across multiple modalities. It provides a framework for organizing reports, including descriptors for lesions and tissues, assessment categories indicating the likelihood of , and recommendations for management, thereby promoting uniformity in how radiologists document and interpret results. The primary purpose of BI-RADS is to minimize ambiguity in reports, ensuring clear and consistent communication of findings among radiologists, referring clinicians, and patients. It facilitates standardized management recommendations based on imaging features, supports the collection and analysis of demographic and outcome data for , and enables through and auditing of practices. By standardizing assessments, BI-RADS enhances overall diagnostic accuracy, reduces the performance of unnecessary biopsies for low-risk findings, and aids in population-based auditing to track cancer detection rates and improve patient care quality. Its scope encompasses , , and (MRI) of the breast, incorporating risk stratification through probabilistic categories rather than providing definitive diagnoses.

History and Development

The Breast Imaging Reporting and Data System (BI-RADS) was initiated by the American College of Radiology (ACR) in the late 1980s to address significant variability in reporting practices, which included inconsistent terminology, ambiguous recommendations, and indecisive interpretations that hindered effective communication and patient care. This effort began in with the formation of an ACR committee to develop a voluntary program for facilities, responding to rising concerns over quality amid increasing screening volumes. A pilot program and formal BI-RADS committee were established by 1988, leading to the first edition's publication in 1993 as an initial atlas focused solely on , marking its formal adoption as a standardized framework. Driving factors included widespread inconsistencies in radiologist reports that contributed to litigation, as well as the need for uniform data collection to support and in . Subsequent developments expanded BI-RADS to accommodate evolving imaging technologies and clinical needs. The second edition, released in 1995, refined mammography guidelines based on early implementation feedback. By the fourth edition in 2003, BI-RADS incorporated dedicated sections for and (MRI), reflecting the growing role of these modalities in and addressing gaps in standardized reporting beyond . The 2013 fifth edition consolidated updates across all modalities, introducing refined assessment categories and lexicon improvements derived from accumulated evidence. These iterations were heavily influenced by the Quality Standards Act (MQSA) of 1992, a federal mandate that required accredited facilities to use standardized reporting and assessment categories to ensure consistent quality and reduce interpretive errors. Early challenges, such as high inter-observer variability in assessments, were systematically addressed through iterative validation studies that demonstrated improved agreement among radiologists following BI-RADS , with values indicating substantial concordance in key descriptors and final categories. Post-2013, the ACR's ongoing committee oversight has focused on evidence-based refinements, including supplements for like contrast-enhanced (CEM), with dedicated BI-RADS guidelines for CEM published in 2022 to integrate its capabilities into standardized reporting. As of 2025, the sixth edition is anticipated for release later in the year, incorporating further updates. These updates ensure BI-RADS remains adaptable to technological advances while maintaining its core goal of reducing reporting discrepancies and enhancing detection outcomes.

Published Documents

Editions and Updates

The Breast Imaging Reporting and Data System (BI-RADS) has evolved through several editions since its inception, with each iteration expanding its scope and refining its components to incorporate advances in technology and clinical evidence. The first edition, published in 1993, focused exclusively on , introducing standardized terminology, reporting structures, and assessment categories to reduce variability in interpretations. Subsequent refinements appeared in a 1995 update, which enhanced descriptor clarity, followed by the third edition in 1998 that included an illustrated atlas for visual reference. The fourth edition in 2003 marked a significant expansion by integrating and (MRI) modalities, along with refinements such as subdivided category 4 assessments (4A, 4B, 4C) for better risk stratification. A supplement to the fourth edition was also released in 2003 to address specific updates. The fifth edition, released in 2013, represented the last major revision of the BI-RADS atlas, providing a comprehensive update across all modalities with over 700 clinical images and enhanced guidance for follow-up and outcome monitoring. Key updates included streamlining the lexicon by removing redundant terms (e.g., "lobular" mass shape and "eggshell" calcifications) and adding new descriptors, such as "developing asymmetry" for , elasticity assessments (soft, intermediate, hard) for , and "clustered ring" for non-mass enhancement in MRI. Background parenchymal enhancement levels (minimal, mild, moderate, marked) were newly categorized for MRI, and composition descriptors were revised to eliminate percentage-based classifications. Final assessment categories were refined to reserve BI-RADS 3, 4, and 5 primarily for diagnostic contexts rather than screening, with an increased emphasis on data tracking, auditing, and location syntax (e.g., and distance from ) to support research and . Following the 2013 edition, the American College of Radiology (ACR) has issued targeted supplements rather than a full sixth edition, which remains unreleased as of 2025 despite annual reviews by the BI-RADS Committee. Notable post-2013 developments include the 2022 supplement for contrast-enhanced mammography (CEM), which introduced a dedicated for CEM findings, including mass and non-mass enhancement descriptors aligned with existing BI-RADS frameworks. These supplements reflect ongoing adaptations to without overhauling the core atlas. The BI-RADS materials are developed by the ACR BI-RADS Committee, comprising breast imaging radiologists, oncologists, epidemiologists, and other stakeholders, through an evidence-based process involving clinical trials, multi-institutional , and expert consensus to ensure reliability and applicability. Revisions prioritize input from diverse clinical perspectives to maintain while addressing real-world implementation challenges. The fifth edition atlas is available for purchase from the ACR (ISBN 978-1559030168) in print and digital formats for licensed users, such as through approved software vendors. Free resources, including reference cards, posters, and summaries of key sections, are accessible on the ACR to support broad adoption.

Core Content and Structure

The BI-RADS Atlas, published by the American College of Radiology (ACR), is structured to provide a comprehensive framework for standardized reporting and data management. The document is divided into key sections, including a of descriptive terms for findings, reporting templates to guide consistent documentation, assessment categories for classifying results, management recommendations tied to those assessments, and auditing tools for . Central elements of the Atlas include illustrated examples featuring over 700 clinical images to demonstrate lexicon terms across modalities, as well as data forms designed for tracking practice outcomes such as recall rates and cancer detection rates. Appendices offer detailed follow-up protocols and frequently asked questions to support implementation. The organization emphasizes modality-specific chapters for , , and (MRI), while incorporating a shared where applicable to promote integration of multimodal findings. Supporting materials within the Atlas encompass guidelines to ensure consistent application and statistical methods for practice audits, such as calculations for positive predictive value to evaluate performance metrics. The fifth edition, released in , introduced a unified reporting across modalities to minimize redundancy and enhance in reports. This structure facilitates a logical progression from descriptive findings to actionable recommendations, with assessment categories serving as the pivotal framework for risk stratification.

Key Components

Standardized Lexicon

The standardized lexicon of the Breast Imaging Reporting and Data System (BI-RADS) serves as a comprehensive dictionary of descriptors designed to promote precise, reproducible language in reports, thereby reducing variability and subjectivity among radiologists. Developed by the American College of Radiology (ACR), this lexicon standardizes the characterization of findings across , , and (MRI), facilitating consistent communication, , and for and auditing purposes. By mandating the use of these specific terms, the lexicon minimizes ambiguous phrases such as "suspicious density" and ensures that all reports exclusively employ lexicon-approved descriptors for imaging findings. The categorizes terms by the type of finding observed. For masses, descriptors include (oval, round, or irregular), margins (circumscribed, indistinct, angular, microlobulated, or spiculated), and internal characteristics such as density (high, equal, low, or fat-containing for ), echo pattern (anechoic, hyperechoic, hypoechoic, isoechoic, heterogeneous, or complex cystic and solid for ), or enhancement patterns (homogeneous, heterogeneous, or rim for MRI). Calcifications are described by morphology (e.g., fine pleomorphic or coarse heterogeneous for suspicious types versus benign forms like large rod-like or vascular) and distribution (diffuse, regional, grouped, linear, or segmental). Asymmetries are classified as focal, global, or developing, while architectural distortions refer to or indentation of tissue without a defined . These terms form the building blocks for structured reporting, integrating seamlessly into the overall BI-RADS framework. Adaptations within the lexicon account for modality-specific features to enhance diagnostic accuracy. In mammography, terms like grouped coarse calcifications highlight clustered benign-appearing deposits, while incorporates orientation (parallel or not) and posterior features (enhancement, shadowing, or mixed). For MRI, descriptors emphasize kinetic curves, including initial enhancement (slow, medium, or fast) and delayed phase patterns (persistent for continuous increase, plateau for stabilization, or washout for signal decrease), alongside non-mass enhancement distributions (focal, linear, segmental, or regional). These tailored terms allow for modality-appropriate nuance without compromising uniformity. The fifth edition of the BI-RADS Atlas, released in 2013, refined the lexicon for greater harmonization across modalities, adding terms such as "focus" to describe small enhancing areas less than 1 cm in diameter on MRI without pre-contrast visibility, and "heterogeneous" echo pattern for ultrasound masses. Obsolete descriptors were removed, including "lobular" shape for masses, "eggshell" and "lucent-centered" for calcifications, and several MRI-specific terms like "enhancing internal septation," "ductal" distribution, and "reticular/dendritic" patterns, to streamline usage and reflect evolving evidence on malignancy risk. This edition also replaced artistic renderings with over 700 actual clinical illustrations to aid interpretation and training. The sixth edition is forthcoming as of 2025, with updates including refinements to descriptors.

Reporting Guidelines

The BI-RADS reporting guidelines establish a standardized format for breast imaging reports across , , and modalities to facilitate clear communication between radiologists and referring clinicians. Required sections in a BI-RADS-compliant report include the indication for the examination (such as screening or diagnostic evaluation), a description of findings using the standardized , an impression that synthesizes the key observations, assignment of a final assessment category, and a management recommendation tailored to the assessment. Procedural rules emphasize that reports must be concise and objective, incorporating comparisons to prior when available to contextualize changes or stability, and noting technical quality factors such as the adequacy of views or compression to ensure interpretability. Where applicable, reports should employ clear, accessible to support understanding in accompanying lay summaries, while maintaining professional precision for clinical use. The American College of Radiology provides fillable template forms for each modality, designed to promote completeness by prompting of elements like breast composition and to streamline adherence to BI-RADS standards. These templates serve as practical tools for radiologists, ensuring all essential components are addressed without unnecessary elaboration. Communication standards require that the final assessment category be prominently indicated, often bolded for emphasis, to highlight its significance in guiding follow-up. For urgent findings, such as those warranting immediate , direct notification to the referring is mandated, typically via telephone followed by a written report within specified timelines to expedite care. In the United States, BI-RADS compliance is mandated under the Mammography Quality Standards Act (MQSA) for all facilities to ensure uniform reporting and , with global adoption encouraged to promote consistency in practices worldwide.

Breast Composition Categories

The Breast Imaging Reporting and Data System (BI-RADS) includes a standardized assessment of breast composition, also known as , to describe the relative amounts of fatty and fibroglandular tissue visible on mammograms. This evaluation is essential for contextualizing the overall interpretation of imaging findings, as denser breast tissue can affect the visibility of potential abnormalities. The categories are determined qualitatively based on the mammographic appearance and are required in every report to ensure consistent communication among radiologists and clinicians. BI-RADS defines four composition categories, ranging from predominantly fatty to extremely dense:
CategoryDescriptionApproximate Glandular Tissue ProportionImplications for Detection
AThe s are almost entirely fatty<25%High mammographic sensitivity, as lesions are easily distinguishable against fatty tissue.
BThere are scattered areas of fibroglandular density25-50%Generally good visibility, though scattered dense areas may occasionally mask small lesions.
CThe s are heterogeneously dense, which may obscure small masses51-75%Reduced sensitivity, with heterogeneous dense regions potentially hiding noncalcified masses.
DThe s are extremely dense, which lowers the sensitivity of mammography>75%Lowest detection rates, as dense tissue significantly obscures both small and larger lesions.
These categories are assessed through subjective visual estimation by the radiologist, reviewing the cranio-caudal and mediolateral oblique views to gauge the overall proportion and distribution of fibroglandular tissue, though efforts have been made to standardize this process for greater interobserver reproducibility. Clinically, higher density categories (C and D) increase the risk of masking, where cancers are hidden within dense tissue, leading to false-negative mammograms. Additionally, dense breasts represent an independent for , with women in category D facing approximately 4- to 6-fold higher risk compared to those in category A, independent of other factors like age or family history. This elevated risk underscores the importance of density assessment in risk stratification and supports recommendations for supplemental screening modalities, such as or MRI, in dense breasts. The fifth edition of the BI-RADS Atlas, published in , refined the category descriptors to emphasize their impact on mammographic sensitivity and to enhance reporting consistency, shifting toward a more qualitative focus while linking to evidence-based supplemental screening guidelines. In reporting, breast composition must be explicitly stated in the "Findings" section and referenced in the "Impression" if relevant to the assessment; for example: "The breasts are heterogeneously , which may obscure small masses (BI-RADS category C)." This requirement ensures that information informs patient management discussions and complies with regulatory standards in many regions.

Assessment Categories

The BI-RADS assessment categories provide a standardized framework for radiologists to convey the degree of suspicion for in findings, facilitating consistent reporting, risk stratification, and clinical decision-making. These categories, numbered 0 through 6, incorporate evidence-based estimates of likelihood derived from large-scale data in the American College of Radiology (ACR) BI-RADS Atlas, enabling auditing of positive predictive values (PPV) in programs. The risk percentages reflect the probability that a in that category represents invasive cancer or (DCIS), based on outcomes from biopsied or followed cases. Management recommendations are tied to each category to guide follow-up, though the primary focus remains on the assessment itself.
CategoryDescriptionMalignancy RiskExample Findings
0Incomplete: Additional imaging evaluation and/or prior mammograms are needed.No risk estimate provided.Inconclusive initial views requiring spot compression or .
1Negative: There is nothing to comment on; the breasts are symmetric with no suspicious findings.<2%.Routine symmetric tissue without masses, distortions, or calcifications.
2Benign: A definite benign finding or entity is identified.<2%.Simple cysts or vascular calcifications with no suspicion of cancer.
3Probably benign: A finding placed in this category should have a very low suspicion for malignancy, warranting short-interval follow-up.<2%.Concordant fibroadenoma or focal asymmetry stable on prior imaging.
4Suspicious abnormality: Biopsy should be considered. Subdivided based on level of suspicion.2% to <95% overall; 4A: >2% to ≤10%; 4B: >10% to ≤50%; 4C: >50% to <95%.Irregular mass or grouped amorphous microcalcifications prompting tissue sampling.
5Highly suggestive of : Appropriate action should be taken, typically tissue diagnosis.≥95%.Spiculated mass with associated microcalcifications.
6Known biopsy-proven : Appropriate action should be taken, such as surgical excision or neoadjuvant therapy planning.100% (prior histologic confirmation).Treated lesion with confirmed invasive cancer on prior biopsy.
These categories emphasize conceptual risk levels rather than modality-specific applications, with the subdivisions in Category 4 introduced in the fifth edition to refine biopsy recommendations and improve PPV calculations in auditing. For the BI-RADS 4A subcategory, the standard malignancy risk is 2%–10%, corresponding to a 90%–98% probability of being benign (e.g., fibroadenoma, cyst, hyperplasia); clinical studies report rates around 4%–9%, rarely exceeding 10%. The risk estimates are periodically updated based on ACR-collected data to reflect evolving outcomes in diverse populations. The sixth edition is forthcoming as of 2025, with updates including refinements to auditing recommendations.

Applications by Modality

The applications of BI-RADS described below are based on the 5th edition of the Atlas, released in 2013, which remains the current standard as of November 2025. The 6th edition is anticipated for release later in 2025, with updates to the lexicons and reporting for mammography, ultrasound, and MRI, among other enhancements.

Mammography

The Breast Imaging Reporting and Data System (BI-RADS) provides a standardized framework for interpreting and reporting mammographic findings, emphasizing modality-specific descriptors to enhance diagnostic accuracy and communication. In mammography, BI-RADS integrates a lexicon tailored to the detection of subtle parenchymal changes and calcifications visible on two-dimensional or three-dimensional X-ray projections, facilitating the differentiation between benign and malignant features. Key elements of the mammography-specific lexicon include descriptors for calcifications, asymmetries, and architectural distortions. Calcifications are classified by morphology and distribution; for instance, amorphous calcifications appear as small, faint clusters without discernible shapes, often warranting further evaluation due to their potential association with ductal carcinoma in situ, while vascular calcifications are typically benign, following the course of blood vessels. Asymmetries are categorized as focal (limited to one quadrant) or global (involving a larger portion of breast tissue without mass-like features), with the addition of "developing asymmetry" in the fifth edition to describe a focal asymmetry that is new or increased in size or conspicuity compared to prior examinations. Architectural distortions manifest as tethering or retraction of tissue planes, often subtle and requiring orthogonal views for confirmation. Standard mammographic views, such as craniocaudal (CC) and mediolateral oblique (MLO), are essential for localizing these findings and assessing their extent across the breast. BI-RADS protocols distinguish between screening and diagnostic mammography to optimize detection. Screening mammography employs routine bilateral two-view imaging (CC and MLO) for asymptomatic women, while diagnostic mammography incorporates additional targeted views, such as spot compression or magnification, for symptomatic patients or abnormal screening results. Every mammographic report must include a breast density assessment using the fifth edition's categories—A (almost entirely fatty), B (scattered areas of fibroglandular density), C (heterogeneously dense), or D (extremely dense)—to contextualize sensitivity limitations. Comparison with prior mammograms is mandatory when available, as it informs the stability or evolution of findings like asymmetries or distortions. Common BI-RADS assessment categories in mammography reflect the probability of malignancy based on these descriptors. Category 3 (probably benign) is frequently assigned to isolated clusters of amorphous or fine pleomorphic calcifications with a malignancy risk of less than 2%, recommending short-interval follow-up imaging at 6 months. Category 4 (suspicious) applies to new or unexplained architectural distortions, subdivided into 4A (2–10% risk), 4B (10–50% risk), and 4C (50–95% risk), prompting biopsy to exclude invasive carcinoma. These categories align with the uniform BI-RADS scale but are illustrated through mammographic examples, such as evolving calcifications or distortions. The fifth edition of BI-RADS, published in 2013, introduced clarifications specific to mammography, including explicit guidance on the "developing asymmetry" descriptor to address its association with a 12–15% malignancy risk when progressing over time. It also emphasized differences between digital and film-screen mammography, noting improved conspicuity of microcalcifications and reduced artifacts in digital systems, which enhance overall interpretive reliability. A primary limitation of mammography within BI-RADS is its reduced sensitivity in dense breasts, where glandular tissue obscures lesions, dropping detection rates from approximately 93% in fatty breasts to as low as 30% in extremely dense ones; BI-RADS mandates explicit notation of density category C or D to guide supplemental screening recommendations.

Ultrasound

The BI-RADS system for ultrasound provides a standardized framework for describing and categorizing breast lesions detected via sonography, emphasizing features that differentiate benign from malignant findings. Developed by the American College of Radiology (ACR), it integrates ultrasound as a complementary modality to mammography, particularly for characterizing masses in dense breast tissue. The modality-specific lexicon focuses on detailed sonographic descriptors to ensure consistent reporting. For masses, key features include shape (oval, round, irregular), orientation (parallel to the skin surface, suggesting benign etiology like fibroadenomas, versus non-parallel, indicating potential malignancy), margins (circumscribed versus not, with the latter subdivided into indistinct, angular, microlobulated, or spiculated), echo pattern (anechoic, hyperechoic, isoechoic, hypoechoic, or complex cystic and solid), and posterior features (no posterior features, enhancement, shadowing, or combined patterns, where shadowing often correlates with invasive carcinoma). Non-mass lesions are described by distribution (focal, linear, regional, multiple regions, or diffuse) and internal echo pattern (anechoic, hyperechoic, isoechoic, heterogeneous, or mixed), with associated features such as ductal changes (dilated ducts or intraductal masses) or architectural distortion. Ultrasound protocols under BI-RADS are tailored for targeted evaluation following abnormal mammography findings or for supplemental whole-breast screening in women with dense breasts, where it improves lesion detection. Examinations typically involve high-frequency linear transducers for radial and anti-radial scanning, with color Doppler incorporated to assess vascularity, aiding in the differentiation of benign hypervascular lesions from malignancies. BI-RADS mandates correlation with mammography results in composite reports to refine assessments. Assessment categories in ultrasound follow the same 0-6 scale as other modalities, with examples including Category 2 (benign) for simple anechoic cysts with posterior enhancement and Category 5 (highly suggestive of malignancy) for irregular, non-parallel hypoechoic masses with spiculated margins and posterior shadowing. The 5th edition (2013) introduced refinements, such as expanded margin descriptors (adding angular and microlobulated types specific to ultrasound) and the "complex cystic and solid" echo pattern, which typically upgrades lesions to Category 4 or higher due to increased malignancy risk (e.g., 23-31% for such lesions). A primary advantage of BI-RADS ultrasound is its ability to distinguish solid from cystic lesions in real-time, without ionizing radiation, though its operator-dependent nature requires adherence to standardized techniques for reproducibility. All ultrasound assessments must integrate with mammographic findings to guide management, such as short-interval follow-up for Category 3 lesions (≤2% malignancy risk) or biopsy for Category 4 (2-95% risk).

Magnetic Resonance Imaging

The Breast Imaging Reporting and Data System (BI-RADS) for magnetic resonance imaging (MRI) provides a standardized framework for interpreting contrast-enhanced breast MRI examinations, focusing on morphologic features and kinetic enhancement patterns to assess lesion suspicion. Breast MRI protocols typically involve dynamic contrast-enhanced sequences, including pre- and post-contrast T1-weighted imaging to evaluate enhancement kinetics, along with T2-weighted sequences to identify fluid or edema characteristics. These examinations are primarily indicated for high-risk screening in women with elevated lifetime breast cancer risk, evaluating the extent of known disease, or problem-solving ambiguous findings from other modalities. The forthcoming 6th edition (2025) includes new recommendations to audit preoperative breast MRI examinations for extent-of-disease assessment in newly diagnosed patients. The modality-specific lexicon emphasizes morphologic and kinetic descriptors to classify findings. Masses are categorized as focus (small enhancing area <5 mm), non-focus mass (larger discrete enhancing lesion), with further descriptors for shape (oval, round, irregular), margins (circumscribed, irregular, spiculated), and internal enhancement (homogeneous, heterogeneous, rim). Non-mass enhancement (NME) is described by distribution patterns such as focal, linear, segmental, regional, multiple regions, or diffuse, and internal patterns including homogeneous, heterogeneous, clumped, or clustered ring. Kinetic assessment evaluates the initial phase of contrast uptake (slow, medium, fast) and delayed phase (persistent—increasing enhancement; plateau—stable; washout—decreasing), which helps differentiate benign from malignant processes. The fifth edition of the BI-RADS Atlas, released in 2013, introduced standardized kinetic curve terminology to improve reproducibility and added categories for background parenchymal enhancement (BPE)—minimal, mild, moderate, marked—to account for physiologic enhancement that can mimic or obscure pathology, with symmetry/asymmetry noted. Assessment categories are tailored to MRI findings; for example, a BI-RADS category 3 (probably benign, <2% malignancy risk) may be assigned to a focus or mass with persistent enhancement suggesting benign etiology, warranting short-interval follow-up. Conversely, a BI-RADS category 5 (highly suggestive of malignancy, ≥95% risk) is appropriate for an irregular mass with spiculated margins and washout kinetics, prompting biopsy. Despite its high sensitivity for detecting breast cancer (ranging from 90% to 100%), breast MRI under demonstrates lower specificity (approximately 80% to 90%), leading to frequent false positives that necessitate additional evaluation. The system underscores the importance of multidisciplinary correlation with clinical history, prior imaging, and pathology to mitigate overdiagnosis and guide management.

Implementation and Tools

Clinical Use and Management

The BI-RADS system translates assessment categories into specific management recommendations to guide clinical decision-making in breast imaging. For category 0 (incomplete assessment), additional imaging evaluation or comparison with prior studies is required to clarify findings. Categories 1 (negative) and 2 (benign finding) indicate routine annual screening mammography without further immediate action. Category 3 (probably benign finding) warrants short-interval follow-up imaging at 6, 12, and 24 months to monitor stability, with risk typically ≤2%. Categories 4 (suspicious abnormality) and 5 (highly suggestive of ) prompt tissue diagnosis via biopsy, with core needle biopsy preferred over fine-needle aspiration for its higher accuracy in characterizing lesions; subcategory risks range from >2% to ≤95% for category 4 (A: 2-10%, B: 10-50%, C: 50-95%) and ≥95% for category 5. Category 6 (known biopsy-proven ) focuses on treatment planning, including surgical, radiation, or oncologic interventions. BI-RADS facilitates multidisciplinary integration by providing radiologists with a structured framework to recommend actions, while the referring makes the final decision based on . This process includes communication to using category-specific probabilities, such as informing a category 3 of a low (<2%) cancer to support shared decision-making on follow-up. In special scenarios, management may be adjusted for patient risk factors. High-risk individuals, such as those with /2 mutations, may warrant category upgrades or more aggressive evaluation (e.g., for a category 3 finding) due to elevated baseline cancer risk and distinct imaging features like higher rates of masses or asymmetries. During , is preferred over to minimize while maintaining diagnostic efficacy for palpable or suspicious findings. Implementation of BI-RADS has demonstrated positive outcomes, including reduced interobserver variability in assessments and biopsy recommendations, leading to more consistent recall rates across radiologists, typically in the 7-9% range with digital breast tomosynthesis (DBT) as of 2025. This standardization supports personalized screening intervals, particularly for high-risk patients, by aligning follow-up with individual risk profiles rather than uniform protocols. Beyond the , BI-RADS has achieved global adoption, with the European Society of Breast Imaging (EUSOBI) endorsing its lexicon and categories in guidelines for , , and MRI across European countries. Adaptations include a dedicated supplement for digital breast , which integrates BI-RADS descriptors to address layered imaging challenges and improve detection.

Quality Assurance and Auditing

The BI-RADS system incorporates robust auditing mechanisms to monitor and enhance the performance of breast imaging practices, tracking key metrics such as the abnormal interpretation rate (AIR), cancer detection rate (CDR), and positive predictive value (PPV). The AIR, defined as the percentage of examinations receiving a positive final assessment (BI-RADS categories 0, 4, or 5), has an acceptable benchmark of 5% to 12% for screening mammography. The CDR measures cancers detected per 1,000 examinations and targets at least 2.5 per 1,000 for screening, rising to 20 per 1,000 for diagnostic workups of abnormal findings and 40 per 1,000 for palpable lumps. PPV1, the proportion of abnormal interpretations leading to a cancer diagnosis within one year, aims for 3% to 8% in screening contexts, while PPV2 for biopsy recommendations targets 20% to 40%. These metrics enable facilities to evaluate radiologist performance against national standards, identifying variations in sensitivity (≥75%) and specificity (88% to 95%). Data collection for BI-RADS audits relies on standardized forms provided in the BI-RADS Atlas, which facilities must complete annually to compile outcomes . These forms capture details on positive examinations, results from biopsies, and follow-up for recommended actions, including interval cancer rates—cancers diagnosed within 12 months of a negative or benign assessment—and false-negative analyses to assess missed detections. The audit process uses BI-RADS assessment categories to categorize interpretations, ensuring consistent tracking across modalities. Compliance with these requirements supports ongoing quality monitoring and regulatory adherence under the Quality Standards Act (MQSA). Quality measures within BI-RADS auditing emphasize of assessment category assignments to promote interpretive consistency among radiologists, alongside mandatory equipment calibration and protocols mandated by MQSA to uphold image quality standards. Radiologists interpreting mammograms must meet MQSA training requirements, including initial qualifications such as in and documented proficiency in interpretation, plus annual focused on updates like BI-RADS lexicon revisions. These elements collectively ensure technical and interpretive reliability in practices. Auditing through BI-RADS benefits practices by pinpointing tendencies toward under-calling or over-calling abnormalities, thereby refining diagnostic accuracy and reducing unnecessary recalls or missed cancers. It also facilitates processes, such as designation as an ACR , which recognizes facilities demonstrating superior performance via audit compliance and outcome monitoring. The sixth edition of the BI-RADS Atlas (released in 2025) builds on prior enhancements with further updates to auditing, including modality-neutral approaches for consistent benchmarking across , , and MRI; redefinition of screening versus diagnostic exams; addition of BI-RADS category 3 auditing to basic audits; and new recommendations for auditing preoperative breast MRI examinations to evaluate extent of disease in newly diagnosed cases. These revisions improve practicality, align metrics with evolving clinical data, and enhance cross-modality comparability.

Automated Extraction

Automated extraction of BI-RADS data leverages computational methods to process unstructured radiology reports and images, enabling efficient parsing of assessment categories and lexicon terms without manual intervention. (NLP) techniques are commonly employed to identify and extract BI-RADS final assessment categories, such as 0 through 6, from free-text reports, achieving high in clinical datasets. For instance, rule-based and machine learning-based NLP systems can detect lexicon terms like mass margins or calcifications, facilitating automated categorization with reported precision rates up to 96.6%. Complementing NLP, (AI) models, particularly convolutional neural networks, automate breast density estimation aligned with BI-RADS categories A through D, demonstrating accuracies exceeding 90% in mammographic assessments. Several software tools support structured BI-RADS reporting to streamline extraction. The American College of Radiology (ACR) endorses platforms like Nuance's PowerScribe 360, which integrates BI-RADS coding templates to generate standardized reports that can be automatically parsed for and compliance tracking. Additionally, FDA-cleared AI solutions, such as iCAD's ProFound Detection software, enhance lesion detection in by providing case scores and BI-RADS-aligned risk assessments, with recent versions cleared for use in digital breast tomosynthesis to improve sensitivity for invasive cancers. These methods find practical applications in automated audits and optimization. In audits, NLP-driven extraction from electronic records (EHRs) enables calculation of performance metrics like positive predictive value (PPV) for BI-RADS categories, supporting quality improvement without manual review. For efficiency, systems prioritize reports flagged as Category 4 or 5, reducing turnaround times by integrating extracted data into decision support tools that alert clinicians to high-risk findings. As of 2025, emerging machine learning approaches focus on multimodal integration, combining mammography and ultrasound data to refine BI-RADS assessments through fused models that improve diagnostic specificity for lesion characterization. However, challenges persist due to variability in lexicon usage across institutions, which can degrade NLP performance and necessitate ongoing model retraining. Validation studies confirm the reliability of these automated methods, with NLP extraction of BI-RADS categories showing accuracies ranging from 85% to 95% across diverse report corpora, depending on the algorithm and dataset size. Despite these advances, regulatory hurdles remain, including the need for robust clinical validation under FDA guidelines and addressing biases in training data to ensure equitable performance in diverse populations for diagnostic AI tools.

References

  1. https://www.acr.org/Clinical-Resources/Clinical-Tools-and-Reference/Reporting-and-Data-Systems/BI-RADS
  2. https://www.ncbi.nlm.nih.gov/books/NBK459169/
  3. https://acsjournals.onlinelibrary.wiley.com/doi/10.1002/cncr.24156
  4. https://pmc.ncbi.nlm.nih.gov/articles/PMC5600516/
  5. https://pmc.ncbi.nlm.nih.gov/articles/PMC3099247/
  6. https://www.sciencedirect.com/science/article/pii/S2211568417300013
  7. https://pmc.ncbi.nlm.nih.gov/articles/PMC3500788/
  8. https://pmc.ncbi.nlm.nih.gov/articles/PMC11674923/
  9. https://www.scielo.br/j/rbgo/a/6CT3Ln4QNrG8xLBFPNtYmcF/
  10. https://doi.org/10.1016/j.diii.2017.01.001
  11. https://msrads.web.unc.edu/wp-content/uploads/sites/15695/2018/04/BIRADS-Reference-Card_web_F.pdf
  12. https://pmc.ncbi.nlm.nih.gov/articles/PMC5040135/
  13. https://pmc.ncbi.nlm.nih.gov/articles/PMC2748779/
  14. https://www.radiologyinfo.org/en/info/article-breast-imaging-report
  15. https://accreditationsupport.acr.org/support/solutions/articles/11000067043-reporting-breast-ultrasound-revised-12-12-19-
  16. https://www.acr.org/Clinical-Resources/Clinical-Tools-and-Reference/Screening-Resources/Breast-Imaging-Resources/Lay-Report-Letters
  17. http://www.gdqc-us.com/upload/files/2022/5/ae9b4027d33cd0cc.pdf
  18. https://www.fda.gov/radiation-emitting-products/mammography-quality-standards-act-mqsa-and-mqsa-program
  19. https://pmc.ncbi.nlm.nih.gov/articles/PMC10146726/
  20. https://pmc.ncbi.nlm.nih.gov/articles/PMC10091988/
  21. https://radiopaedia.org/articles/breast-imaging-reporting-and-data-system-bi-rads-assessment-category-4?lang=us
  22. https://pmc.ncbi.nlm.nih.gov/articles/PMC4453135/
  23. https://pmc.ncbi.nlm.nih.gov/articles/PMC8774686/
  24. https://www.arrs.org/ARRSLIVE/ARRSLIVE/Education/Meetings/Symposia/VBR25/Home.aspx
  25. https://www.uclahealth.org/departments/radiology/education/breast-imaging-teaching-resources/birads/mammography-calcifications
  26. https://radiologyassistant.nl/breast/bi-rads/bi-rads-for-mammography-and-ultrasound-2013
  27. https://pubs.rsna.org/doi/abs/10.1148/radiol.14132759
  28. https://www.uclahealth.org/departments/radiology/education/breast-imaging-teaching-resources/birads/architectural-distortion
  29. https://www.cancer.org/cancer/types/breast-cancer/screening-tests-and-early-detection/mammograms/understanding-your-mammogram-report.html
  30. https://pubs.rsna.org/doi/abs/10.1148/rg.240169
  31. https://ajronline.org/doi/10.2214/AJR.15.14628
  32. https://pubs.rsna.org/doi/full/10.1148/rg.230024
  33. https://www.acog.org/clinical/clinical-guidance/committee-opinion/articles/2015/03/management-of-women-with-dense-breasts-diagnosed-by-mammography
  34. https://pmc.ncbi.nlm.nih.gov/articles/PMC5207351/
  35. https://www.uclahealth.org/departments/radiology/education/breast-imaging-teaching-resources/birads/mri-sequences-and-how-read-breast-mri-and-bi-rads
  36. https://pubs.rsna.org/doi/10.1148/radiol.243803
  37. https://pmc.ncbi.nlm.nih.gov/articles/PMC9549780/
  38. https://www.uclahealth.org/departments/radiology/education/breast-imaging-teaching-resources/birads/breast-mri-findings-focus-and-mass
  39. https://pmc.ncbi.nlm.nih.gov/articles/PMC6767440/
  40. https://radiopaedia.org/articles/breast-imaging-reporting-and-data-system-bi-rads-2?lang=us
  41. https://www.breastcancer.org/screening-testing/mammograms/bi-rads-results
  42. https://pmc.ncbi.nlm.nih.gov/articles/PMC6413875/
  43. https://www.jacr.org/article/S1546-1440(09)00390-1/fulltext
  44. https://ajronline.org/doi/10.2214/AJR.16.16957
  45. https://pubs.rsna.org/doi/abs/10.1148/rg.230014
  46. https://pubs.rsna.org/doi/10.1148/radiol.2533082308
  47. https://pmc.ncbi.nlm.nih.gov/articles/PMC8087168/
  48. https://be-accepted.com/wp-content/uploads/2023/02/Mammography_Update_Recommendations-EUSOBI.pdf
  49. https://ajronline.org/doi/10.2214/AJR.20.24894
  50. https://cme.lww.com/ovidfiles/00219246-202104150-00001.pdf
  51. https://www.acr.org/Accreditation/Designations/Designated-Comprehensive-Breast-Imaging-Center
  52. https://ceus.med.br/wp-content/uploads/2024/10/2606307-2187608-001.pdf
  53. https://pmc.ncbi.nlm.nih.gov/articles/PMC3782591/
  54. https://pubs.rsna.org/doi/full/10.1148/radiol.2018180694
  55. https://isupportcontent.nuance.com/healthcare/documents/ats/resource/p36/P36A084.pdf
  56. https://www.accessdata.fda.gov/cdrh_docs/pdf24/K240417.pdf
  57. https://www.icadmed.com/about/science/
  58. https://pmc.ncbi.nlm.nih.gov/articles/PMC5108519/
  59. https://www.clinicalradiologyonline.net/article/S0009-9260%2823%2900423-3/fulltext
  60. https://www.nature.com/articles/s41598-025-99535-2
  61. https://www.mdpi.com/2075-4418/13/8/1420
  62. https://journals.lww.com/cmj/fulltext/2019/07200/automatic_extraction_of_imaging_observation_and.6.aspx
  63. https://www.jmir.org/2025/1/e62941
  64. https://pmc.ncbi.nlm.nih.gov/articles/PMC11075018/
Add your contribution
Related Hubs
User Avatar
No comments yet.