How to improve your breast cancer program: Standardized reporting using the new American College of Radiology Breast Imaging-Reporting and Data System

In the USA, the use of the American College of Radiology Breast Imaging-Reporting and Data System (ACR BI-RADS) has served not only as a quality assurance tool and guide to standardizing breast imaging reports but has also improved communication between referring physicians, researchers, and patients. In fact, in the USA, the Mammography Quality Standards Act of 1997 requires that all mammograms be assigned a BI-RADS category based on the ﬁ nding of most concern. In this manuscript, we aim to review the recommendations provided in the 4 th edition of the ACR BI-RADS for mammography, USG


Introduction
The American College of Radiology (ACR) Þ rst developed the Breast Imaging-Reporting and Data System (BI-RADS) in 1993 in an eff ort to provide a quality assurance tool that would standardize mammographic reporting, facilitate outcome monitoring, and reduce the ambiguity surrounding breast imaging reports. [1,2]Before that, reports were not standardized and there was inconsistent use of imaging terminology, resulting in confusion to the referring physicians and patients.There was also no mandate to provide further patient management recommendations based on imaging Þ ndings.In many instances, reports could not be clearly categorized as either positive or negative.This created the additional challenge of the need to track outcomes and perform practice outcome audits.When the mammographic ACR BI-RADS was introduced, it was lauded for providing a standardized lexicon and reporting format.It allowed the radiologist to relate the degree of concern for malignancy through a concise description, using approved terminology, and to give clear management recommendations [Tables 1 and 2].Since then, BI-RADS lexicons have been developed for USG and MRI and these were published in the 4 th edition of ACR BI-RADS. [2]][8] Finally, the new ACR BI-RADS lexicon for breast MRI has been advocated for mainstream use as it facilitates clear and concise interpretation. [9]here are diff erences between the imaging modalities with regard to their technical aspects, all reports are required to include a relevant clinical history, state the indication for the examination, list the pertinent Þ ndings, provide the overall impression, and assign a BI-RADS category.Pertinent Þ ndings are writt en in a succinct manner using standardized terminology from the latest ACR BI-RADS lexicon without embellishment.All Þ ndings are to begin with a comment on the overall composition of the breast, followed by descriptions of any lesions that may be present.Features unique to a particular study, such as calciÞ cations with mammograms or echogenicity with USG, are then described using approved descriptors as per the BI-RADS lexicon.The Þ nal impression needs to clearly state the degree of suspicion for malignancy and off er recommendations for further management based on the BI-RADS classiÞ cation.Although more than one pertinent feature may have been observed, the Þ nal BI-RADS assessment is based on the most suspicious feature.

Indication for examination 3. Comparison with previous studies (if deemed necessary
by the radiologist) 4. Breast composition 5. Finding(s) 6. Overall assessment and management recommendations Indication for the examination could be either 'screening' or 'diagnostic evaluation.'A screening mammographic examination is performed on an asymptomatic woman in order to detect early, clinically unsuspected, breast cancer.Two standard views of each breast -mediolateral oblique and craniocaudal projection -are obtained.A diagnostic mammographic examination is performed on a woman who presents with clinical signs or symptoms.A diagnostic study is tailored by the radiologist to address the presenting complaint and may include additional projections other than the standard views; USG or, in rare cases, breast MRI may also be performed.
The report on the mammographic Þ ndings begins with a description of the breast composition, commenting on background glandularity: fatt y, scatt ered Þ broglandular densities, heterogeneously dense, or extremely dense tissue.When the breast tissue is either heterogeneously dense or extremely dense, mammography has relatively low accuracy and a disclaimer statement can be added to the report regarding the decreased sensitivity of the study.BI-RADS-approved terminology is clearly deÞ ned in detail below.In general, descriptions of masses, calciÞ cations, densities, and associated Þ ndings include details of size, morphology, and location.An example of a standard normal screening mammogram report in a heterogeneously dense breast is presented in Figure 1.

USG report
A USG report is organized similar to a mammography report and includes:   Ojeda-Fournier and Nguyen: The new ACR BI-RADS 4. Background echotexture 5. Finding(s) 6. Correlation with physical Þ ndings, mammography, MRI, etc. 7. Overall assessment and management recommendations USG analysis begins with a brief examination of breast tissue in the area of concern.Any lesion identified is described in terms of size in at least two dimensions (in three dimensions, if the volume of the mass is analyzed in successive studies), a consistent and reproducible assessment of lesion location (o'clock position and distance from the nipple), and characterization of the lesion using ACR-approved descriptors.There should also be a statement correlating USG results with physical Þ ndings or other imaging Þ ndings if relevant.

Breast MRI report
The breast MRI report includes: 1. Clinical history 2. Comparison with old studies 3. Description of the MRI technique used, including postprocessing technique 4. Finding(s) 5. Kinetic curve report 6.Overall assessment and management recommendations Description of the MRI technique includes relevant factors such as the use of a dedicated breast coil, pulse sequences, contrast dose, etc.In the 'findings' section, an overall assessment of the glandular tissues is made, similar to the assessment of background density in mammography or assessment of echotexture in USG studies.Terminology used includes: mild, moderate, or marked background glandularity.These statements can then clarify any questions regarding a lesion being hidden by normal tissue on other studies.Any artifacts that can aff ect interpretation can be described.Breast implants are described to include composition (saline or silicone) and number of lumens.Unique Þ ndings on MRI include enhancement, which is further characterized as focus/foci, mass, non-mass-like enhancement, and/or symmetric/asymmetric.The Þ nal impression should fully assess each lesion as a threedimensional structure along with its kinetic Þ ndings.If an abnormality is deemed suspicious, the report indicates whether a biopsy should be considered and speciÞ es the recommended image guidance for the procedure (i.e., stereotactic, USG, or MRI).
If more than one type of breast imaging modality is being interpreted, a single report that integrates all signiÞ cant Þ ndings is recommended in order to bett er guide clinical management.
An integrated report would include the following:

Lexicon
The ACR Committ ee on Breast Cancer has addressed the widespread confusion caused by evolving terminology, which was neither universal nor clearly defined and, oftentimes, was used redundantly.An approved standardized lexicon [Table 1] was reviewed and adopted into the ACR guidelines with the hope of decreasing confusion and promoting universal utilization among breast imagers.By using the ACR-approved descriptors, breast imaging reports have become more streamlined and clear and concise.Both USG and MRI have unique features not found in mammography but, wherever applicable, common features are described using the terms developed for mammography.BI-RADS assessment categories however are the same for all three imaging modalities.

Mammogram
By deÞ nition, a mass is a space-occupying lesion seen in two diff erent projections and, further, is characterized by shape, margin, and density [Figure 2].
• The shape of a mass is classiÞ ed as round, oval, lobular, or irregular.
• Margins can be obscured, indistinct (ill-defined), spiculated (as demonstrated in Figure 3), microlobulated, or circumscribed (well-deÞ ned or sharply-deÞ ned).Incidentally, the descriptor 'circumscribed' can be used as long as 75% of the margin is clearly visible, with the remainder being no worse than obscured by the overlying tissue.
• Density descriptors include terms like high, equal, low, or fat-containing and are used to describe X-ray att enuation of the lesion relative to the expected att enuation of the surrounding Þ broglandular breast tissue.For example, Figure 4 demonstrates a fat-containing lesion.
• Asymmetry refers to a potential mass that is seen in only a single projection and is planar, lacks convex borders, and usually contains interspersed fat.Asymmetries lack the conspicuity of a three-dimensional mass.The term density has also been used for this same Þ nding but is discouraged by the ACR as it has led to confusion because the term is also used to describe a mass' att enuation characteristics.A review of previous Þ lms is critical for accurate interpretation of asymmetries and, if none are available, a full diagnostic evaluation is warranted.
• 'Global asymmetry' involves a large portion of the breast (at least a quadrant) and is judged in comparison with a corresponding area in the contralateral breast.There is no mass, distorted architecture, or associated suspicious calciÞ cations.
• 'Focal asymmetry' involves a smaller portion of the breast than does global asymmetry and does not Þ t the criteria of a mass [Figure 5].It is a conÞ ned asymmetry, with a similar shape on two views, and it lacks the borders and the conspicuity of a true mass.• Calcifications need to be described by their size, morphology, and distribution.
• Amorphous calcifications are small and hazy and warrant further investigation with biopsy.Suspicious calcifications are described as fine linear, linear branching, and pleomorphic [Figure 6].
• Descriptors for distribution patt erns include grouped or clustered, segmental, regional, or diff use/scatt ered.

USG
Given the success of the standardized mammogram lexicon, the ACR developed a similar lexicon for USG in 2003.Research has shown the success of the USG BI-RADS lexicon in improving communication, diminishing confusion, and facilitating reporting and data tracking. [5]ckground echotexture is unique to USG and is characterized as heterogeneous or homogeneous.Homogeneous echotextures can be further described as Þ broglandular or fat.
The shape and margins of masses are described using the same descriptors as are used for masses in mammography.Pertinent descriptions of masses should also include distinctive USG features, such as: • Posterior acoustic qualities, which may be described as "none," "enhancement," or "shadowing" (as seen in Figure 8B), or may be a combination of these • Orientation (parallel or not parallel to the skin edge) • Echo patt ern [hyperechoic, anechoic (as seen in Figure 2C), hypoechoic, isoechoic, or a complex combination of these] • Lesion boundary describes the transition zone between the mass and the surrounding tissue and may be an abrupt interface or an echogenic halo • Eff ects of the mass on surrounding tissue (compression, obliteration, eff ects on Cooper ligaments, echogenic halo, edema, etc.) • USG is not ideal for characterizing calciÞ cations; they are seen as echogenic foci, particularly when within a mass.
• There are several special cases described by the USG lexicon: • Clustered microcysts consist of tiny anechoic foci, individually smaller than 2-3 mm, with thin (<0.5 mm) intervening septae and no discrete solid component.
• Complicated cysts may contain brightly echogenic foci that scintillate as they shift through ß uid-debris levels.These cysts do not contain solid mural nodules.If a discrete solid component is identiÞ ed, the lesion should be classiÞ ed as a complex mass requiring aspiration or other intervention.
• Masses in or on the skin, such as epidermal inclusion cysts, keloids, moles, etc., can also be identiÞ ed.
• Foreign bodies include marker clips, coils, wires, catheter sleeves, silicone (with a characteristic 'snowstorm' appearance), and metal or glass related to trauma.
• Intramammary and axillary lymph nodes.• Vascularity should be described in terms of whether it is present or not, location in relation to lesion, and extent.

MRI
The MRI lexicon has evolved over the last 4 years and remains work-in-progress.The current ACR BI-RADS MRI lexicon reß ects current technology and will be periodically updated as new sequences and imaging techniques develop.The ACR has emphasized that breast MRI images are representations of three-dimensional objects and that multiplanar reconstructions and three-dimensional images are appropriate for review; therefore, use of a soft ware package for breast MRI interpretation is recommended.Kinetic information is as important as morphologic descriptions of signiÞ cant lesions and the availability of a soft ware package to interpret breast MRI aids in the display of color overlay and curve assessment.
The terms 'mass' and 'regional enhancements' have been a source of confusion.A mass has well-deÞ ned margins, with distinct edges separating it from the surrounding glandular tissue.This is demonstrated on USG and MRI in Figure 9.It is usually related to a pathological process in a ball-like, three-dimensional structure that may or may not displace or otherwise aff ect normal breast tissue.Masses can be further described using characteristics common to mammography and USG, such as margins, basic geometry, size, location, and associated Þ ndings, in addition to features unique to MRI, such as enhancement patt ern and kinetics.
Ojeda-Fournier and Nguyen: The new ACR BI-RADS Figure 9 (A-C): A 55-year-old woman with a palpable abnormality.USG (A) shows a hypoechoic, irregular mass (arrow) with angular margins, classifi ed as BI-RADS 5. Biopsy of the mass showed stage II invasive ductal carcinoma.A pre-operative breast MRI with high-resolution delayed post-contrast (B) and subtracted (C) images, demonstrates a heterogeneously enhancing, irregular mass (arrows) with spiculated margins, classifi ed as BI-RADS 6 • Shapes can be round, oval, lobulated, or irregular • Margin analysis is dependent on spatial resolution and can be described as smooth, spiculated, or irregular (it is strongly recommended that the descriptor 'irregular' be used to characterize either the margin or the shape, but not both, to avoid confusion).Figure 9 shows a heterogeneously enhancing mass with an irregular shape and spiculated margins • As with other breast imaging modalities, size and location should be reported in a consistent, reproducible manner • Enhancement is defined as higher signal intensity compared with the surrounding normal glandular tissue aft er contrast administration.The size and location should be reported along with any associated Þ ndings • If a mass exhibits internal enhancement, it is described as homogeneous or heterogeneous.Homogeneous enhancement is conß uent and uniform.Heterogeneous enhancement is non-uniform, with areas of variable signal intensity.Additional descriptors of patterns of enhancement of a mass are rim enhancement, dark internal septae, enhancing septae, or central enhancement.
• A 'focus,' on the other hand, is a tiny punctate enhancement that is non-speciÞ c and too small (<5 mm) for morphological characterization.There is usually no corresponding Þ nding on the pre-contrast scan.Foci can be multiple spots or dots of enhancement, separated widely in the breast by normal tissue or fat, and are not a conglomerate of dots in one small area.
• An area of non-mass-like enhancement is neither a focus nor a mass and its patt ern can extend over small or large regions in a patt ern that can be focal, linear, ductal, segmental, regional, multiple regions, or diff use.A focal area of abnormal enhancement is <25% of a breast quadrant volume and comprises a single abnormal enhancement patt ern.Ductal enhancement can be linear or branching and corresponds to one or more ducts, usually radiating toward the nipple.Segmental enhancement refers to a triangular or cone-like shape with the apex at the nipple.Regional enhancement may be geographic and lacks convex borders, as demonstrated in Figure 10.To qualify as multiple regions of enhancement, there must be at least two broad areas separated by normal tissue or fat.Lastly, diff use non-mass-like enhancement is widely scatt ered and evenly distributed • The internal characteristics of non-mass-like enhancements are further described as homogeneous, heterogeneous, stippled/punctate, clumped, or reticular/ dendritic.Stippled refers to multiple, oft en innumerable, punctate foci that are approximately 1-2 mm in size and appear scatt ered throughout an area of the breast.It does not conform to a duct system.The clumped patt ern refers to an aggregate of enhancing masses or foci in a cobblestone patt ern that may occasionally be conß uent.Reticular/dendritic enhancement is a spider web-like patt ern found among women who have undergone at least partial involution of the glandular tissue, leaving strands of breast tissue among strands of fat.It appears as bright thickening, distortion, and foreshortening of normal Þ broglandular trabeculae and tissue, with loss of the normal scalloped edges of the fat/breast tissue interface at its edges • 'Symmetric' and 'asymmetric' are used when bilateral breast MRI is performed.Symmetric refers to mirrorimage enhancement in the right and left breasts.Asymmetric enhancement [Figure 11] is a more  concerning Þ nding and should be further modiÞ ed using the lexicon described above.
• Kinetic techniques analyze dynamic lesion enhancement and generate signal intensity/time curves.The shapes of these curves off er a wealth of information to help diff erentiate between benign and malignant lesions • The initial enhancement phase is analyzed within the Þ rst 1-2 min aft er injection.The descriptors are slow, medium, or rapid • The delayed phase refers to the enhancement patt ern aft er 2 min or aft er the curve starts to change.This curve is described as persistent, washout, or plateau.Persistent refers to continued increase in signal intensity over time.Washout describes a curve that shows decreasing signal intensity aft er peak enhancement.A plateau curve reaches the maximum signal intensity and then remains constant In Figure 12, a biopsy-proven malignant mass is preoperatively evaluated with MRI, including kinetic assessment.There is a rapid initial enhancement, followed by a washout signal intensity curve.

Overall assessment and recommendations
The key component of the breast-imaging report is the overall assessment and recommendations [Table 2].A single BI-RADS category is assigned based on the most suspicious Þ nding on imaging.If more than one imaging study is performed, the Þ nal assessment incorporates Þ ndings from all these studies.
There are seven assessment categories: ACR BI-RADS categories 0 through 6.These assessment categories are divided into incomplete (category 0) and Þ nal (categories 1-6).

Category 0: Assessment is incomplete
These examinations are incomplete until some further evaluation is performed.This can be in the form of additional mammographic views, comparison films, USG, or, less commonly, breast MRI.At times, in order to accurately assign a Þ nal BI-RADS classiÞ cation, comparison with old films is required.The radiologist should use judgment on how vigorously to att empt to Þ nd old Þ lms for comparison and whether the report truly needs a category 0 coding vs category 1 or 2. Tracking old Þ lms can be time consuming and expensive and therefore 'category 0 (for comparison)' should only be used when such comparison is absolutely required for making a Þ nal assessment.Once comparisons or additional imaging studies are completed, a Þ nal assessment can then be rendered.Category 0 is oft en used in a screening situation.

Category 1: Negative
This is a completely negative exam with no signiÞ cant Þ ndings.There are no masses, no architectural distortion, no suspicious calciÞ cations, and no asymmetries.
Figure 12 (A-E): A 66-year-old woman with a remote history of bilateral breast cancer and radiation therapy presented with a new rapidly growing mass.Craniocaudal (A) and mediolateral oblique (B) mammograms show a round mass.USG (C) demonstrates a hypoechoic oval mass with no posterior enhancement, classifi ed as BI-RADS 5. Biopsy showed sarcoma.Pre-operative MRI (D) shows a round mass with a lobulated margin, with no chest wall invasion.Kinetic assessment (E) demonstrates rapid initial enhancement followed by washout in the delayed portion of the curve, classifi ed as BI-RADS 6

Category 3: Probably benign
Category 3 remains a source of confusion and is sometimes controversial.This category is to be used in the presence of a Þ nding that is almost certainly benign but for which a short follow-up is desired.It is never to be used as an indeterminate malignancy category or in lieu of a diagnostic work-up.In fact, this category has a <2% risk of malignancy and is unlikely to require biopsy.The Þ nding is not expected to change quickly over time and therefore recommended follow-up involves a series of short-interval (6-month) follow-ups over a period of 24-36 months.Aft er the Þ nding has maintained a stable appearance for 2-3 years, it can be considered benign and be coded as a category 2 (benign) Þ nding.In some instances, due to patient and/or clinician concern, some category 3 findings may end up being biopsied rather than followed.This category is not to be used when there is a palpable lesion.Examples of category 3 lesions include non-palpable, incidental, complicated cysts [Figure 13].

Category 4: Suspicious abnormality
This category includes Þ ndings that do not have a classic appearance for malignancy, but have a higher probability for malignancy than Þ ndings classiÞ ed as category 3. Here, some form of intervention (preferably image-guided needle core biopsy) is recommended to establish a diagnosis.The wide range of outcome probabilities has stimulated a subdivision of category 4.An optional form of subdivision suggested by the ACR is 4A, 4B, and 4C.A 4A coding can be used for a Þ nding with a low suspicion for malignancy but requiring some type of intervention.Examples include a palpable complicated cyst or probable abscess.Category 4B includes lesions with intermediate probability for malignancy.Such a lesion requires close radiologic and pathologic follow-up and is only truly benign if the results of both concur.Examples include fat necrosis and papilloma.Category 4C identifies findings that are of moderate concern but do not exhibit the classic signs of malignancy.A malignant result is expected in this category and it should alert pathologists and clinicians to carefully follow-up on these biopsies.Included here are ill-deÞ ned, irregular solid masses or new clusters of Þ ne pleomorphic calciÞ cations.

Category 5: Highly suggestive of malignancy
These lesions are almost certainly malignant, carrying a >95% probability of malignancy.Imaging Þ ndings exhibit the classic characteristics of malignancy and percutaneous tissue sampling may be required for oncologic management (i.e., neoadjuvant chemotherapy) or to plan a one-stage deÞ nitive surgical intervention that may include lymph node sampling.A spiculated, irregular, high-density mass would be a classic example of category 5 [Figure 3].

Category 6: Known biopsy-proven malignancy
This category is new and was added to accommodate breast Þ ndings that have been proven to be cancer by biopsy but for which deÞ nitive treatment (surgical excision, radiation, chemotherapy, or mastectomy) has not yet been executed.It is appropriate for patients seeking a second opinion, for monitoring responses to neoadjuvant chemotherapy, or for patients who require further staging.
The Þ nal assessment is always based on the most immediate action required.For instance, a patient with known cancer in one breast may be sent for an outside imaging consultation.In the event that there are no other signiÞ cant Þ ndings apart from the cancer or if there is a benign Þ nding that requires no intervention, the report can be classiÞ ed as category 6.If another abnormality is found that requires further evaluation, the Þ nal assessment would then be category 0. If there is an additional lesion requiring biopsy, then the report would be coded category 4.
In terms of the mammography audit, BI-RADS have helped to more clearly deÞ ne positive and negative studies.Any screening mammogram coded BI-RADS categories 0, 4, or 5 is considered positive.Categories 1, 2, and 3 are negative.In the scenario of a patient with negative imaging but a clinically signiÞ cant palpable Þ nding, the report should still be coded based on the imaging Þ ndings alone.However, the Þ nal assessment should take into consideration the clinical Þ ndings and make appropriate recommendations.All breast imaging centers in the USA are required by the Mammography Quality Standards Act (MQSA) to perform a basic medical audit to track sensitivity, speciÞ city, positive and negative predictive values, cancer detection rate, and abnormal interpretation rates.A chapter in the 4 th edition of the ACR BI-RADS manual provides guidance for follow-up and outcome monitoring.

Controversies
Multiple studies have shown that the ACR BI-RADS lexicon is eff ective and that the positive predictive value for each of the BI-RADS assessment categories is for the most part consistent among diff erent radiologists. [7,8,10]In the beginning, several weaknesses were noted, including confusion regarding the system's guidelines.This was partly att ributed to the fact that the guidelines do not explicitly state how to make a Þ nal assessment based on mammographic features.Also, the approved terminology was initially thought to be restrictive or unsuitable. [8]espite initial drawbacks, BI-RADS has been an excellent system and, for the most part, has accomplished its goal of clearing up ambiguity surrounding breast imaging reporting.Acceptance, since its introduction in 1993, has universally increased.Since the 1997 MQSA was passed, assignment of the ACR BI-RADS category based on the Þ nding of most concern is now required in the USA for all mammograms. [1]Since 1993, ACR BI-RADS has undergone several revisions, including the addition of USG and MRI lexicons and reÞ nement of the lexicons and assessment categories to bett er accommodate clinical needs.
Revisions have been guided in part by studies demonstrating intra-and interobserver variability.It has been consistently noted that intraobserver agreement is better than interobserver agreement. [10]The reason seems to be that radiologists have their own personal interpretations of BI-RADS, varying thresholds, and diff erent cut-off points in determining the best-Þ t descriptors and categories. [10]aplin, Lehman, and others have shown that interobserver variability is not signiÞ cant in the negative and benign assessment categories but is statistically significant in assessments and management recommendations associated with BI-RADS category 3 and, to a lesser extent, categories 4 and 5. [3,8,11,12] Although lesion description may vary from observer to observer depending on personal preferences and experience, it is critical that the Þ nal assessment category result in the correct action.
With regard to BIRADS category 3, the 'probably benign' finding, there is high variability in its usage, in both academic and community practices.This category is reserved for Þ ndings where the risk of malignancy is low (<2%) and short-interval follow-up is preferred over biopsy in almost all cases.Lehman and colleagues, however, demonstrated that category 3 lesions in a community practice had an actual cancer detection rate of 8.8%. [6]Upon further investigation, the criterion for a probably benign Þ nding was not strictly adhered to in at least 80% of the examinations.The factors cited for this misuse included misclassiÞ cation of morphologic features, failure to take into account available prior images, or failure to properly track signiÞ cant changes in the lesion over time. [6]Taplin observed similar improper usage of category 3-coded examinations, where 37% had recommendations for 'additional imaging' and another 19% were referred for 'normal-interval followup.' [12]In Taplin's study, improper usage was also seen in category 4 (38%) and category 5 (7%), for which additional imaging was recommended instead of the appropriate recommendation of a biopsy.To summarize, although category 3 remains controversial, it is to be used only aft er complete problem-solving imaging has been performed and only then is the lesion in question deemed probably benign.
Category 0 also has limitations as it covers a broad range of results, ranging from those needing comparisons to assess stability to those recommending further evaluation of benign or highly suspicious lesions.Taplin suggests a modiÞ ed category 0 to bett er describe the level of concern. [12]ModiÞ ers such as 'incomplete assessment of a probably benign Þ nding,' where imaging would be performed to clarify an ambiguity and 'incomplete assessment of a suspicious Þ nding,' where additional imaging would lead to a category 4 or 5 coding, may be useful in the future.
In the 4 th edition of the ACR BI-RADS, a subcategorization of category 4 lesions is recommended by the ACR, although it is not explicitly required.Before this, BI-RADS category 4 was a heterogeneous collection of lesions with a wide range of malignancy possibilities and thus could not clearly describe the radiologist's degree of concern. [13]By dividing category 4 into 4a (low), 4b (moderate), and 4c (high), the pre-biopsy risk of malignancy is bett er communicated to the referring physician, pathologist, and radiologist who may be involved in performing the follow-up biopsy.Berg found that it is the lesions near the threshold for biopsy that are the most problematic and were probably responsible for the high variability in actual practice. [14]Subdivision of category 4 is expected to address some of this variability.
Berg has found that specialized BI-RADS training improves the interobserver agreement among breast imagers with respect to lesion description and the Þ nal assessment. [3]ther researchers agree with Berg's Þ ndings and also feel that periodic performance assessments and self-auditing practice tests will aid consistent use of the ACR BI-RADS system. [10,12,14]plicit in the ACR BI-RADS system is the fact that the lexicon and assessment categories are to be consistently used among radiologists, even if it requires additional training to become familiar with the system.For example, the MRI and USG lexicons are new additions to BI-RADS that still need optimization and consistent use among radiologists. [5,15]This is recognized by the ACR and it is clearly stated in their guidelines that the BI-RADS lexicon remains a 'living document' that will certainly undergo future revision.As new imaging modalities emerge, the BI-RADS lexicon will also evolve.

Conclusion
By consistently and appropriately utilizing the standard lexicon and assigning the appropriate ACR BI-RADS Þ nal assessment categories, radiologists specializing in breast imaging will eff ectively communicate Þ ndings, degree of concern for malignancy, and recommendations to both clinicians and patients.Although problems persist with the system, including the issues of inter-and intraobserver variabilities, multiple studies have validated the effi cacy of the descriptors and assessment categories.There is no other system available currently that accomplishes what BI-RADS

Figure 1 :
Figure 1: Example of a normal mammogram report in a patient with heterogenous breast density

Figure 2 (Figure 4 :
Figure2 (A-C): A 37-year-old female with a palpable breast mass marked with a BB at the time of diagnostic evaluation.Mammogram (A,B) demonstrates a partly obscured, partly circumscribed, oval mass (arrow).Follow-up USG (C) demonstrates an anechoic structure with imperceptible margins and posterior enhancement consistent with a simple cyst.This was classifi ed as BI-RADS 2

Figure 3 :Figure 5 :
Figure 3: Screening mammography in a 67-year-old woman demonstrates a spiculated mass.Biopsy showed invasive ductal carcinoma, nuclear grade II.This was classifi ed as BI-RADS 5

Figure 7 :
Figure 7: A 52-year-old woman presented with a palpable abnormality following breast reduction surgery.Mammography shows coarse and eggshell calcifi cation consistent with fat necrosis, classifi ed as BI-RADS 2

Figure 10 :
Figure 10: A 61-year-old woman with a history of ductal carcinoma in situ.Breast MRI demonstrates an area of non-mass-like enhancement (arrow), classifi ed as BI-RADS 4. MRI-guided biopsy revealed benign fi brocystic changes

Figure 11 :
Figure 11: Pre-operative breast MRI in a 62-year-old woman with biopsy-proven high-grade ductal carcinoma in situ demonstrates diffuse non-mass-like enhancement (arrows) in the right breast, classifi ed as BI-RADS 6

Figure 13 :
Figure 13: Incidental, non-palpable complicated cyst on USG, classifi ed as BI-RADS 3. A short interval, 6-month follow-up is requested to assess stability

Table 2 : American College of Radiology BI-RADS Final Assessment Categories BI-RADS categories Assessment Clinical management
6Known biopsy-proven malignancy breast findings already proven by biopsy to be cancer but pending definitive treatment appropriate for patients seeking a second opinion, monitoring responses to neoadjuvant chemotherapy, or for patients who require further staging is necessary to split Þ ndings for two or more imaging modalities into separate reports for coding and reimbursement reasons, but it is recommended that the Þ nal assessment integrate Þ ndings from all current imaging studies.