[Computer-aided diagnosis of breast cancer]

Computer-aided diagnosis (CAD) is generally defined as a diagnosis made by a radiologist who takes into account computer output as a "second opinion". The computer output usually indicates the location of suspicious lesions such as microcalcifications or masses in mammograms, and also provides quantitative analysis of potential lesions, yielding measures such as the likelihood of malignancy. The purpose of CAD is to improve diagnostic accuracy and the consistency of radiologists' image interpretation by using such computer output as a guide. This article provides a brief overview of some CAD schemes used for breast cancer.

News media coverage of screening mammography for women in their 40s and tamoxifen for primary prevention of breast cancer

CONTEXT

In the late 1990s, 3 events pertaining to breast cancer prevention received considerable attention in the US news media: a National Institutes of Health (NIH) consensus panel recommended against routine screening mammography for women in their 40s (January 1997), the National Cancer Institute (NCI) subsequently reversed the recommendation (March 1997), and an NCI-sponsored study demonstrated the efficacy of tamoxifen in the primary prevention of breast cancer (April 1998).

OBJECTIVE

To examine how the major US news media covered the potential benefits and harms of 2 breast cancer preventive strategies.

DESIGN AND SETTING

Content analysis of US news stories reporting on the breast cancer prevention events. We used Lexis-Nexis to search for print news stories in the 10 highest-circulation US newspapers and requested transcripts from 3 major television networks to obtain all relevant news coverage in the 2 weeks following each event.

MAIN OUTCOME MEASURES

Attitude toward preventive strategy (encourage, neutral, discourage); level of uncertainty about benefit and how benefits and harms were presented.

RESULTS

Twenty-seven stories about the NIH consensus panel, 24 about the NCI reversal, and 34 about tamoxifen appeared in high-profile news media within 2 weeks of each event. Sixty-seven percent of NIH consensus panel stories left the impression that there was a lot of uncertainty about whether women aged 40 to 49 years should undergo screening, but 59% suggested that women should probably or definitely be screened. Only 4 stories suggested that women faced a genuine decision about what to do. The level of uncertainty reported was substantially lower following the NCI reversal (21% suggested a lot of uncertainty), and most stories (96%) suggested that women should be screened. In contrast, tamoxifen stories highlighted uncertainty about what women at high risk should do (62% suggested there was a lot of uncertainty), and none left the impression that women should definitely take the drug (24% suggested they probably should). Sixty-five percent of these stories suggested that women faced a genuine choice and would have to weigh the risks and benefits themselves.

CONCLUSIONS

Most news stories favored routine use of screening mammography and urged caution about using tamoxifen. Almost all noted the potential harms of each preventive strategy; however, the negative aspects of tamoxifen received greater emphasis. Whereas taking tamoxifen was presented as a difficult decision, having a mammogram was presented as something women ought to do.

Technical quality control practices in mammography screening programs in 22 countries

OBJECTIVE

To assess current technical quality control (QC) practices within breast cancer screening or surveillance programs internationally.

MATERIALS AND METHODS

The International Breast Cancer Screening Network (IBSN) conducted an extensive survey of quality assurance (QA) activities in developed countries known to have population-based breast cancer screening or surveillance programs in place. Twenty-three countries were sent questionnaires that included items about QA and QC requirements at screening sites, the minimum frequencies of QC test performance, and the personnel responsible for performing QC tests.

RESULTS

All 23 countries in the IBSN completed general information on their QA practices. Twenty-two countries responded with complete details on their technical QC practices. The responses indicated a pattern of consistently high-quality control practices among population-based breast cancer screening and surveillance programs. Most programs performed the great majority of QC tests. Variations were observed in the performance frequencies of QC tests and in the personnel responsible for performing QC tests.

CONCLUSION

QC practices among population-based breast cancer screening and surveillance programs are highly evolved, with the great majority of responding countries following prescribed QC protocols. Further research is needed on appropriate performance frequencies for mammography QC tests.

[Clear-cut beneficial effect of breast cancer screening on the trend in breast-conserving surgery]

Since the introduction of mammographic screening, controversies have emerged concerning its effect on mortality as well as on the management of breast cancer. A Dutch population-based study demonstrates a significant difference between screening-detected breast cancers and non-screening-detected breast cancers in terms of smaller tumour size, more favourable nodal status and a 14% increase (difference between percentages) in breast-conserving surgery in favour of the screening-detected group. However, since no overall increase in the rate of breast-conserving therapy was found, the authors concluded that the impact of screening on surgical management was limited. An alternative explanation is that during the 1990s, the indications for breast-conserving surgery have become more restrictive in terms of tumour-free margins. Other investigators have demonstrated similar increases in the rate of breast-conserving surgery in screening-detected breast cancer patients. Although the debate on mammographic screening has definitely not reached its conclusion, Swedish data available clearly demonstrate a decrease in breast cancer mortality (21%) due to mammographic screening, which is contrary to recent criticism.

[The biased information given to women about population screening for breast cancer makes a well considered informed choice to participate unlikely]

The 'Wet op de Geneeskundige Behandelingsovereenkomst' (Act on Agreement Concerning Medical Treatment) requires that patients are clearly informed about 'the risks and consequences that can be expected from an intended treatment or investigation'. The information received by women aged 50-74 years when invited for a biannual investigation for breast cancer does not satisfy the requirements: the available knowledge is presented too favourably and it is therefore not possible to make a well-considered informed choice to participate. For example, in the group to be screened, 4000 cases of breast cancer are diagnosed each year, yet in the information leaflet a figure of 9000 is stated, the number for all age categories. Furthermore, these 9000 cases include 1500 cases detected only as a result of screening. In the information leaflet no mention is made of the fact that 1 in 5 women will die despite screening and the risk of interval carcinomas is not examined; 40% of malignant mammary tumours are still found outside of the screening programme.

Factors affecting radiologist inconsistency in screening mammography

RATIONALE AND OBJECTIVES

Although research has successfully documented variability in radiologists' interpretation of mammograms, it has failed to determine the relative contributions of case-specific features and reader inconsistency. Training interventions to improve consistency will be ineffectual if they do not target the principal determinants of disagreement among radiologists. The current study assessed the relative contributions of the case and the interpreter to the problem of inconsistent interpretation.

MATERIALS AND METHODS

One hundred ten radiologists independently interpreted mammograms from the same 148 screening cases (43% with biopsy-proved cancers) and reported the presence or absence of calcifications, mass, architectural distortion, and asymmetric density in each of 296 breasts. The radiologists were blinded to disease status (established at biopsy or follow-up).

RESULTS

Case-related differences accounted for a greater proportion of interpretation disagreement than did differences between interpreters. The presence of cancer was associated with increased disagreement, perhaps because of the multiplicity of findings. Patient age was also associated with increased disagreement in the reporting of calcifications.

CONCLUSION

For screening mammography, increased consistency between radiologists in their recognition and reporting of clinically important findings will best be achieved by reducing disagreement in difficult cases. Current training in the United States addresses difficult cases only as they have been defined intuitively or experientially. The authors' population-based method provides an objective metric to measure case difficulty and basis from which to identify difficult cases for targeted training.

Agreement or prediction: asking and answering the right question

INTRODUCTION

Measuring agreement and measuring predictive ability are similar but distinct problems. Failure to appreciate the conceptual and practical differences may lead clinical researchers to give the right answer to the wrong question.

METHODS

We illustrate the relation and difference between measuring agreement and predictive ability in a non-technical way. We provide a real example investigating the feasibility of using preoperative breast cancer tumour size measurements to estimate postoperative histological size. The intraclass correlation and R-squared are calculated to ascertain the level of agreement and predictive ability respectively.

RESULTS

Analysis of agreement and analysis of predictive ability serve different purposes. The optimal solution found in terms of agreement may be different from that found for prediction.

CONCLUSIONS

A careful clarification of the goal of an investigation is important. Using an inappropriate analysis can lead to misleading results, or to results that do not really answer the research question of interest.

Breast imaging reporting and data system (BI-RADS)

The Breast Imaging Reporting and Data System (BI-RADS) lexicon was developed by the American College of Radiology to standardize mammographic reporting. The BI-RADS lexicon defines terms to describe abnormalities on mammograms, and it defines final assessment categories that are predictive of the likelihood of malignancy. Although the lexicon is clinically useful and facilitates communication and research, there is still substantial interobserver variability in its application. Lexicons for breast sonography and breast MRI are in progress.

Screening mammography: clinical image quality and the risk of interval breast cancer

OBJECTIVE

We evaluated the association between clinical image quality and breast cancer occurrence within 24 months of a negative mammogram.

MATERIALS AND METHODS

We identified women with breast cancer who were younger than 40 years old and older and screened from January 1, 1988, through December 31, 1993. We retrospectively assigned Breast Imaging Reporting and Data System (BI-RADS) assessments to their screening mammogram. We classified cancers (invasive or ductal in situ) as "screen-detected" when found after positive assessments (BI-RADS codes 3, 4, and 5) and "interval-detected" when found after negative assessments (BI-RADS codes 1 and 2). One reviewer evaluated mediolateral oblique and craniocaudal views for all cancer cases using a 3-point scale (failure, borderline, pass) for each measure of clinical image quality (positioning, breast compression, contrast, exposure, noise, sharpness, artifacts, overall quality). We used separate logistic regression models and evaluated the odds of interval invasive cancer or invasive plus in situ cancer as a function of each measure of quality using "pass" as the referent group.

RESULTS

We found 492 screen-detected and 164 interval-detected cancers that met study criteria. Cancer detection (sensitivity) was highest (84%) among patients with proper breast positioning, but when images failed this measure (33.4%), sensitivity fell to 66.3%. After adjustment for age, film date, and breast density, interval-detected invasive cancers were more likely after images failing positioning (odds ratio, 2.57; 95% confidence interval, 1.28-5.52%). Failures in overall quality were also associated with interval cancers when cases of ductal carcinoma in situ (p = 0.037) were included.

CONCLUSION

Invasive breast cancer detection by mammography may be improved through attention to correct positioning.

Improving the accuracy of mammography: volume and outcome relationships

BACKGROUND

Countries with centralized, high-volume mammography screening programs, such as the U.K. and Sweden, emphasize high specificity (low percentage of false positives) and high sensitivity (high percentage of true positives). By contrast, the United States does not have centralized, high-volume screening programs, emphasizes high sensitivity, and has lower average specificity. We investigated whether high sensitivity can be achieved in the context of high specificity and whether the number of mammograms read per radiologist (reader volume) drives both sensitivity and specificity.

METHODS

The U.K.'s National Health Service Breast Screening Programme uses the PERFORMS 2 test as a teaching and assessment tool for radiologists. The same 60-film PERFORMS 2 test was given to 194 high-volume U.K. radiologists and to 60 U.S. radiologists, who were assigned to low-, medium-, or high-volume groups on the basis of the number of mammograms read per month. The standard binormal receiver-operating characteristic (ROC) model was fitted to the data of individual readers. Detection accuracy was measured by the sensitivity at specificity = 0.90, and differences among sensitivities were determined by analysis of variance.

RESULTS

The average sensitivity at specificity = 0.90 was 0.785 for U.K. radiologists, 0.756 for high-volume U.S. radiologists, 0.702 for medium-volume U.S. radiologists, and 0.648 for low-volume U.S. radiologists. At this specificity, low-volume U.S. radiologists had statistically significantly lower sensitivity than either high-volume U.S. radiologists or U.K. radiologists, and medium-volume U.S. radiologists had statistically significantly lower sensitivity than U.K. radiologists (P<.001, for all comparisons).

CONCLUSIONS

Reader volume is an important determinant of mammogram sensitivity and specificity. High sensitivity (high cancer detection rate) can be achieved with high specificity (low false-positive rate) in high-volume centers. This study suggests that there is great potential for optimizing mammography screening.

Compliance testing of medical diagnostic x-ray equipment: three years' experience at a major teaching hospital in Western Australia

The impact of a formal compliance testing program has been evaluated three years post-implementation on a major teaching hospital (Sir Charles Gairdner Hospital) with 46 x-ray tubes located throughout 37 rooms. The mandatory program, implemented by the statutory authority in January 1997, called for all medical (including chiropractic and dental) equipment used in human diagnosis to be tested at prescribed frequencies using established protocols. The application of the required test methods demonstrated various non-compliance issues. Notices of non-compliance were received for approximately 60% of the equipment in the hospital following the equipment's first annual test. The reasons for, and the significance of, failure varied according to equipment category, test category, equipment use and equipment age. However, at the end of the third year of testing, approximately 75% of the tested x-ray units satisfied the compliance criteria. The main reasons for non-compliance were found to be design limitations of old technology and the current radiation legislation that makes it difficult for some older equipment to meet the relatively stringent criteria.

Assessment of a breast cancer screening programme in Shiraz, Islamic Republic of Iran

A breast cancer screening programme was evaluated for approximately 10,000 women aged 35 years and older. There were 67 cases of breast cancer. Highest rates of attendance were seen among younger women (35-44 years) and middle socioeconomic groups. Lowest rates were among those aged over 65 years and low socioeconomic groups. The rate of detection by self-examination was similar to that by health personnel examination. At all stages of screening, positive findings were most common among the high socioeconomic class. Attendance decreased steadily from first to last stages of serial screening. Although mammography is the most sensitive method of detection, because of its high cost we suggest establishing breast self-examination education programmes and encouraging women to self-examine.