Optimal reference mammography: a comparison of mammograms obtained 1 and 2 years before the present examination

Revolutionizing Breast Cancer Detection With Artificial Intelligence (AI) in Radiology and Radiation Oncology: A Systematic Review

The number one cause of cancer in women worldwide is breast cancer. Over the last three decades, the use of traditional screen-film mammography has increased, but in recent years, digital mammography and 3D tomosynthesis have become standard procedures for breast cancer screening. With the advancement of technology, the interpretation of images using automated algorithms has become a subject of interest. Initially, computer-aided detection (CAD) was introduced; however, it did not show any long-term benefit in clinical practice. With recent advances in artificial intelligence (AI) methods, these technologies are showing promising potential for more accurate and efficient automated breast cancer detection and treatment. While AI promises widespread integration in breast cancer detection and treatment, challenges such as data quality, regulatory, ethical implications, and algorithm validation are crucial. Addressing these is essential for fully realizing AI's potential in enhancing early diagnosis and improving patient outcomes in breast cancer management. In this review article, we aim to provide an overview of the latest developments and applications of AI in breast cancer screening and treatment. While the existing literature primarily consists of retrospective studies, ongoing and future prospective research is poised to offer deeper insights. Artificial intelligence is on the verge of widespread integration into breast cancer detection and treatment, holding the potential to enhance early diagnosis and improve patient outcomes.

A CNN Hyperparameters Optimization Based on Particle Swarm Optimization for Mammography Breast Cancer Classification

Breast cancer is considered one of the most-common types of cancers among females in the world, with a high mortality rate. Medical imaging is still one of the most-reliable tools to detect breast cancer. Unfortunately, manual image detection takes much time. This paper proposes a new deep learning method based on Convolutional Neural Networks (CNNs). Convolutional Neural Networks are widely used for image classification. However, the determination process for accurate hyperparameters and architectures is still a challenging task. In this work, a highly accurate CNN model to detect breast cancer by mammography was developed. The proposed method is based on the Particle Swarm Optimization (PSO) algorithm in order to look for suitable hyperparameters and the architecture for the CNN model. The CNN model using PSO achieved success rates of 98.23% and 97.98% on the DDSM and MIAS datasets, respectively. The experimental results proved that the proposed CNN model gave the best accuracy values in comparison with other studies in the field. As a result, CNN models for mammography classification can now be created automatically. The proposed method can be considered as a powerful technique for breast cancer prediction.

Does the availability of prior mammograms improve radiologists' observer performance?-a scoping review

Objective

The objective of this review was to examine the impact of previous mammogram availability on radiologists' performance from screening populations and experimental studies.

Materials and Methods

A search of the literature was conducted using five databases: MEDLINE, PubMed, Web of Science, ScienceDirect, and CINAHL as well as Google and reference lists of articles. Keywords were combined with "AND" or "OR" or "WITH" and included "prior mammograms, diagnostic performance, initial images, diagnostic efficacy, subsequent images, previous imaging, and radiologist's performance". Studies that assessed the impact of previous mammogram availability on radiologists' performance were reviewed. The Standard for Reporting Diagnostic Accuracy guidelines was used to critically appraise individual sources of evidence.

Results

A total of 15 articles were reviewed. The sample of mammogram cases used across these studies varied from 36 to 1,208,051. Prior mammograms did not affect sensitivity [with priors: 62-86% (mean = 73.3%); without priors: 69.4-87.4% (mean = 75.8%)] and cancer detection rate, but increased specificity [with priors: 72-96% (mean = 87.5%); without priors: 63-87% (mean = 80.5%)] and reduced false-positive rates [with priors: 3.7 to 36% (mean = 19.9%); without priors 13.3-49% (mean = 31.4%)], recall rates [with priors: 3.8-57% (mean = 26.6%); without priors: [4.9%-67.5% (mean = 37.9%)], and abnormal interpretation rate decreased by 4% with priors. Evidence for the associations between the availability of prior mammograms and positive-predictive value, area under the curve (AUC) from the receiver operating characteristic curve (ROC) and localisation ROC AUC, and positive-predictive value of recall is limited and unclear.

Conclusion

Availability of prior mammograms reduces recall rates, false-positive rates, abnormal interpretation rates, and increases specificity without affecting sensitivity and cancer detection rate.

Repeat Breast Ultrasound Demonstrates Utility with Added Cancer Detection in Patients following Breast Imaging Second Opinion Recommendations

Purpose

Second opinion consultation for patients with suspicious findings on breast imaging and patients with known breast cancer is not uncommon. We sought to determine the frequency of second opinion breast and axillary ultrasound imaging review and the subsequent impact on clinical management.

Materials and Methods

An IRB-approved retrospective chart review was conducted on 400 consecutive patients with second opinion radiology interpretations performed by subspecialized breast radiologists at a designated cancer center, including mammogram and ultrasound review. The outside institution imaging reports were compared with second opinion reports to categorize ultrasound review discrepancies which were defined as any BI-RADS category change. The discrepancy frequency, relevant alterations in patient management, and added cancer detection were measured.

Results

The second opinion imaging review resulted in discrepant findings in 108/400 patients (27%). Patients with heterogeneously or extremely dense breasts had higher discrepancy frequency (36% discrepancy, 68/187) than those with almost entirely fatty or scattered fibroglandular breast tissue (19% discrepancy, 40/213) with P = 0.0001. Discrepancies resulted in the following changes in impression/recommendations: 70 repeat ultrasounds for better characterization of a breast lesion, 11 repeat ultrasounds of a negative region, 20 repeat ultrasounds for benign axillary lymph nodes, 5 downgrades from probably benign to benign, and 2 upgrades from benign to suspicious. Repeat ultrasounds of the axilla in 19 patients resulted in 13 biopsy recommendations, and 4 were metastatic (PPV3 31%). In the breast, repeat ultrasounds in 81 patients resulted in 14 upgrades to suspicious. Of these, 5 yielded malignancy. In addition, one patient was upgraded from benign to suspicious based on the outside image, with pathology revealing malignancy (breast PPV3 40%). Breast lesion BI-RADS category downgrades in 27 patients resulted in 10 avoided biopsies. Ultimately, second opinion ultrasound review resulted in altered management in 12% of patients (47/400). This included discovery of additional breast malignancies in 6 patients, metastatic lymph nodes in 4 patients, excisional biopsy for atypia in 1 patient, 4 patients proceeding to mastectomy, 10 patients who avoided biopsies, and 22 patients who avoided follow-up of benign findings.

Conclusions

In this study, subspecialized second opinion ultrasound review had an impact on preventing unnecessary procedures and follow-up exams in 8% of patients while detecting additional cancer in 2.5%.

Artificial intelligence for breast cancer detection in mammography and digital breast tomosynthesis: State of the art

Screening for breast cancer with mammography has been introduced in various countries over the last 30 years, initially using analog screen-film-based systems and, over the last 20 years, transitioning to the use of fully digital systems. With the introduction of digitization, the computer interpretation of images has been a subject of intense interest, resulting in the introduction of computer-aided detection (CADe) and diagnosis (CADx) algorithms in the early 2000's. Although they were introduced with high expectations, the potential improvement in the clinical realm failed to materialize, mostly due to the high number of false positive marks per analyzed image. In the last five years, the artificial intelligence (AI) revolution in computing, driven mostly by deep learning and convolutional neural networks, has also pervaded the field of automated breast cancer detection in digital mammography and digital breast tomosynthesis. Research in this area first involved comparison of its capabilities to that of conventional CADe/CADx methods, which quickly demonstrated the potential of this new technology. In the last couple of years, more mature and some commercial products have been developed, and studies of their performance compared to that of experienced breast radiologists are showing that these algorithms are on par with human-performance levels in retrospective data sets. Although additional studies, especially prospective evaluations performed in the real screening environment, are needed, it is becoming clear that AI will have an important role in the future breast cancer screening realm. Exactly how this new player will shape this field remains to be determined, but recent studies are already evaluating different options for implementation of this technology. The aim of this review is to provide an overview of the basic concepts and developments in the field AI for breast cancer detection in digital mammography and digital breast tomosynthesis. The pitfalls of conventional methods, and how these are, for the most part, avoided by this new technology, will be discussed. Importantly, studies that have evaluated the current capabilities of AI and proposals for how these capabilities should be leveraged in the clinical realm will be reviewed, while the questions that need to be answered before this vision becomes a reality are posed.

Availability of prior mammograms affects incomplete report rates in mobile screening mammography

PURPOSE

Mobile mammography can improve access to screening mammography in rural areas and underserved populations. We evaluated the frequency of incomplete reports in mobile mammography screening and the relationships between prior mammograms and recall rates.

METHODS

The frequency of incomplete mammogram reports, the subgroups of those needing prior comparison mammograms, recalls for additional imaging, and availability of prior mammograms of a mobile screening mammography unit were compared with fixed site mammography from January 1, 2007 through December 31, 2009. All mobile unit mammograms were full field digital mammography (FFDM). Differences between rates of recall, incomplete reports, and availability of prior mammograms were calculated using the Chi-Square statistic.

RESULTS

Of 2640 mobile mammography cases, 21.9% (578) reports were incomplete, versus 15.2% (7653) (p ≤ 0.001) of 50325 fixed site reports. Of incomplete cases, recall for additional imaging occurred among 8.3% (218) of mobile mammography reports versus 11.3% (5708) (p ≤ 0.001) of fixed site reports. Prior mammograms were needed among 13.6% (360) of mobile mammography versus 3.9% (1945) (p ≤ 0.001) of fixed site reports. Mobile mammography recall rate varied with availability of prior mammograms: 16.0% (54) when no prior mammograms, 7.6% (127) when prior mammograms were elsewhere but unavailable and 5.9% (37) when prior FFDM were immediately available (p ≤ 0.001).

CONCLUSIONS

Incomplete reports were more frequent in mobile mammography than the fixed site. The availability of prior comparison mammograms at time of interpretation decreased the rate of incomplete mammogram reports. Recall rates were higher without prior comparison mammograms and lowest when comparison FFDM mammograms were available.

Impact of prior mammograms on combined reading of digital mammography and digital breast tomosynthesis

Background Although digital breast tomosynthesis (DBT) is an emerging technique yielding higher sensitivity and specificity compared to digital mammography (DM) alone, relative contribution of prior mammograms on the interpretation of DBT combined with DM has not been investigated. Purpose To retrospectively compare the diagnostic performances of DM, DM + DBT, and DM + DBT with prior mammograms. Material and Methods Three breast radiologists independently reviewed images of 116 patients with 24 cancers in the sequential order of DM, DM + DBT, and DM + DBT with prior mammograms using Breast Imaging Reporting and Data System (BI-RADS) assessment categories. Results The average areas under the receiver operating characteristic curve (AUC) of DM, DM + DBT, and DM + DBT with prior mammograms were 0.712, 0.777, and 0.816, respectively. Adding prior mammograms did not significantly affect the AUC of DM + DBT ( P = 0.108), whereas adding DBT significantly increased the AUC of DM ( P = 0.009). Sensitivity for DM, DM + DBT, and DM + DBT with prior mammograms was 58.3%, 69.4%, and 69.4%, and specificities were 84.1%, 85.9%, and 93.8%, respectively. Addition of DBT significantly increased the sensitivity ( P = 0.0090) of DM. Prior mammograms significantly improved the specificity of DM + DBT ( P = 0.0004), whereas adding prior mammogram did not affect sensitivity of DM + DBT ( P = 1.000). Conclusion DBT significantly increases the overall sensitivity and diagnostic performance of DM. Prior mammograms significantly increase the specificity of DM + DBT but have no significant effect on sensitivity and overall diagnostic performance.

Improving Screening Mammography Outcomes Through Comparison With Multiple Prior Mammograms

OBJECTIVE

The objective of the present study is to evaluate the effect of comparison with multiple prior mammograms on the outcomes of screening mammography relative to comparison with a single prior mammogram.

MATERIALS AND METHODS

We retrospectively analyzed 46,288 consecutive screening mammograms performed at our institution for 22,792 women. We divided these examinations into three groups: those interpreted without comparison with prior mammograms, those interpreted in comparison with one prior examination, and those interpreted in comparison with two or more prior examinations. For each group, we determined the rate of examination recall. We also calculated the positive predictive value of recall (i.e., positive predictive value level 1 [PPV1]) and the cancer detection rate (CDR) for both the group of examinations compared with a single prior mammogram and the group compared with multiple prior mammograms. Generalized estimating equations with the logistic link function were used to determine the relative odds ratio of recall as a function of the number of comparisons, with adjustment made for age as a confounding variable. The Fisher exact test was performed to compare the PPV1 and the CDR in the different cohorts.

RESULTS

The recall rate for mammograms interpreted without comparison with prior examinations was 16.6%, whereas that for mammograms compared with one prior examination was 7.8% and that for mammograms compared with two or more prior examinations was 6.3%. After adjustment was made for age, the odds ratio of recall for the group with multiple prior examinations relative to the group with a single prior examination was 0.864 (95% CI, 0.776-0.962; p = 0.0074). Statistically significant increases in the PPV1 of 0.05 (p = 0.0009) and in the CDR of 2.3 cases per 1000 examinations (p = 0.0481) were also noted for mammograms compared with multiple prior examinations relative to those compared with a single prior examination.

CONCLUSION

Comparison with two or more prior mammograms resulted in a statistically significant reduction in the screening mammography recall rate and increases in the CDR and PPV1 relative to comparison with a single prior mammogram.

Effect of the Availability of Prior Full-Field Digital Mammography and Digital Breast Tomosynthesis Images on the Interpretation of Mammograms

PURPOSE

To assess the effect of and interaction between the availability of prior images and digital breast tomosynthesis (DBT) images in decisions to recall women during mammogram interpretation.

MATERIALS AND METHODS

Verbal informed consent was obtained for this HIPAA-compliant institutional review board-approved protocol. Eight radiologists independently interpreted twice deidentified mammograms obtained in 153 women (age range, 37-83 years; mean age, 53.7 years ± 9.3 [standard deviation]) in a mode by reader by case-balanced fully crossed study. Each study consisted of current and prior full-field digital mammography (FFDM) images and DBT images that were acquired in our facility between June 2009 and January 2013. For one reading, sequential ratings were provided by using (a) current FFDM images only, (b) current FFDM and DBT images, and (c) current FFDM, DBT, and prior FFDM images. The other reading consisted of (a) current FFDM images only, (b) current and prior FFDM images, and (c) current FFDM, prior FFDM, and DBT images. Fifty verified cancer cases, 60 negative and benign cases (clinically not recalled), and 43 benign cases (clinically recalled) were included. Recall recommendations and interaction between the effect of prior FFDM and DBT images were assessed by using a generalized linear model accounting for case and reader variability.

RESULTS

Average recall rates in noncancer cases were significantly reduced with the addition of prior FFDM images by 34% (145 of 421) and 32% (106 of 333) without and with DBT images, respectively (P < .001). However, this recall reduction was achieved at the cost of a corresponding 7% (23 of 345) and 4% (14 of 353) reduction in sensitivity (P = .006). In contrast, availability of DBT images resulted in a smaller reduction in recall rates (false-positive interpretations) of 19% (76 of 409) and 26% (71 of 276) without and with prior FFDM images, respectively (P = .001). Availability of DBT images resulted in 4% (15 of 338) and 8% (25 of 322) increases in sensitivity, respectively (P = .007). The effects of the availability of prior FFDM images or DBT images did not significantly change regardless of the sequence in presentation (P = .81 and P = .47 for specificity and sensitivity, respectively).

CONCLUSION

The availability of prior FFDM or DBT images is a largely independent contributing factor in reducing recall recommendations during mammographic interpretation.

Impact of and interaction between the availability of prior examinations and DBT on the interpretation of negative and benign mammograms

RATIONALE AND OBJECTIVES

To assess the interaction between the availability of prior examinations and digital breast tomosynthesis (DBT) in decisions to recall a woman during interpretation of mammograms.

MATERIALS AND METHODS

Eight radiologists independently interpreted twice 36 mammography examinations, each of which had current and prior full-field digital mammography images (FFDM) and DBT under a Health Insurance Portability and Accountability Act-compliant, institutional review board-approved protocol (written consent waived). During the first reading, three sequential ratings were provided using FFDM only, followed by FFDM + DBT, and then followed by FFDM + DBT + priors. The second reading included FFDM only, then FFDM + priors, and then FFDM + priors + DBT. Twenty-two benign cases clinically recalled, 12 negative/benign examinations (not recalled), and two verified cancer cases were included. Recall recommendations and interaction between the effect of priors and DBT on decisions were assessed (P = .05 significance level) using generalized linear model (PROC GLIMMIX, SAS, version 9.3; SAS Institute, Cary, NC) accounting for case and reader variability.

RESULTS

Average recall rates in noncancer cases were significantly reduced (51%; P < .001) with the addition of DBT and with addition of priors (23%; P = .01). In absolute terms, the addition of DBT to FFDM reduced the recall rates from 0.67 to 0.42 and from 0.54 to 0.27 when DBT was available before and after priors, respectively. Recall reductions were from 0.64 to 0.54 and from 0.42 to 0.33 when priors were available before and after DBT, respectively. Regardless of the sequence in presentation, there were no statistically significant interactions between the effect of availability of DBT and priors (P = .80).

CONCLUSIONS

Availability of both priors and DBT are independent primary factors in reducing recall recommendations during mammographic interpretations.

Mammography test sets: reading location and prior images do not affect group performance

AIM

To examine how the location where reading takes place and the availability of prior images can affect performance in breast test-set reading.

MATERIALS AND METHODS

Under optimized viewing conditions, 10 expert screen readers each interpreted a reader-specific set of images containing 200 mammographic cases. Readers, randomly divided into two groups read images under one of two pairs of conditions: clinical read with prior images and laboratory read with prior images; laboratory read with prior images and laboratory read without prior images. Region-of-interest (ROI) figure-of-merit (FOM) was analysed using JAFROC software. Breast side-specific sensitivity and specificity were tested using Wilcoxon matched-pairs signed rank tests. Agreement between pairs of readings was measured using Kendall's coefficient of concordance.

RESULTS

Group performances between test-set readings demonstrated similar ROI FOMs, sensitivity and specificity median values, and acceptable levels of agreement between pairs of readings were shown (W = 0.75-0.79, p < 0.001) for both pairs of reading conditions. On an individual reader level, two readers demonstrated significant decreases (p < 0.05) in ROI FOMs when prior images were unavailable. Reading location had an inconsistent impact on individual performance.

CONCLUSION

Reading location and availability of prior images did not significantly alter group performance.

Measures of angular spread and entropy for the detection of architectural distortion in prior mammograms

PURPOSE

Architectural distortion is an important sign of early breast cancer. We present methods for computer-aided detection of architectural distortion in mammograms acquired prior to the diagnosis of breast cancer in the interval between scheduled screening sessions.

METHODS

Potential sites of architectural distortion were detected using node maps obtained through the application of a bank of Gabor filters and linear phase portrait modeling. A total of 4,224 regions of interest (ROIs) were automatically obtained from 106 prior mammograms of 56 interval-cancer cases, including 301 true-positive ROIs, and from 52 mammograms of 13 normal cases. Each ROI was represented by three types of entropy measures of angular histograms composed with the Gabor magnitude response, angle, coherence, orientation strength, and the angular spread of power in the Fourier spectrum, including Shannon's entropy, Tsallis entropy for nonextensive systems, and Rényi entropy for extensive systems.

RESULTS

Using the entropy measures with stepwise logistic regression and the leave-one-patient-out method for feature selection and cross-validation, an artificial neural network resulted in an area under the receiver operating characteristic curve of 0.75. Free-response receiver operating characteristics indicated a sensitivity of 0.80 at 5.2 false positives (FPs) per patient.

CONCLUSION

The proposed methods can detect architectural distortion in prior mammograms taken 15 months (on the average) before clinical diagnosis of breast cancer, with a high sensitivity and a moderate number of FPs per patient. The results are promising and may be improved with additional features to characterize subtle abnormalities and larger databases including prior mammograms.

Effect of observing change from comparison mammograms on performance of screening mammography in a large community-based population

PURPOSE

To evaluate the effect of comparison mammograms on accuracy, sensitivity, specificity, positive predictive value (PPV(1)), and cancer detection rate (CDR) of screening mammography to determine the role played by identification of change on comparison mammograms.

MATERIALS AND METHODS

This HIPAA-compliant and institutional review board-approved prospective study was performed with waiver of patient informed consent. A total of 1,157,980 screening mammograms obtained between 1994 and 2008 in 435,183 women aged at least 40 years were included. Radiologists recorded presence of comparison mammograms and change, if seen. Women were followed for 1 year to monitor cancer occurrence. Performance measurements were calculated for screening with comparison mammograms versus screening without comparison mammograms and for screening with comparison mammograms that showed a change versus screening with comparison mammograms that did not show a change while controlling for age, breast density, and data clustering.

RESULTS

Comparison mammograms were available in 93% of examinations. For screening with comparison mammograms versus screening without comparison mammograms, CDR per 1000 women was 3.7 versus 7.1; recall rate, 6.9% versus 14.9%; sensitivity, 78.9% versus 87.4%; specificity, 93.5% versus 85.7%; and PPV(1), 5.4% versus 4.8%. For screening with comparison mammograms that showed a change versus screening with comparison mammograms that did not show a change, CDR per 1000 women was 25.4 versus 0.8; recall rate, 41.4% versus 2.0%; sensitivity, 96.6% versus 43.5%; specificity, 60.4% versus 98.1%; and PPV(1), 6.0% versus 3.9%. Detected cancers with change were 21.1% ductal carcinoma in situ and 78.9% invasive carcinoma. Detected cancers with no change were 19.3% ductal carcinoma in situ and 80.7% invasive carcinoma.

CONCLUSION

Performance is affected when change from comparison mammograms is noted. Without change, sensitivity is low and specificity is high. With change, sensitivity is high, with a high false-positive rate (low specificity). Further work is needed to appreciate changes that might indicate cancer and to identify changes that are likely not indicative of cancer.

Detection of architectural distortion in prior mammograms

We present methods for the detection of sites of architectural distortion in prior mammograms of interval-cancer cases. We hypothesize that screening mammograms obtained prior to the detection of cancer could contain subtle signs of early stages of breast cancer, in particular, architectural distortion. The methods are based upon Gabor filters, phase portrait analysis, a novel method for the analysis of the angular spread of power, fractal analysis, Laws' texture energy measures derived from geometrically transformed regions of interest (ROIs), and Haralick's texture features. With Gabor filters and phase portrait analysis, 4224 ROIs were automatically obtained from 106 prior mammograms of 56 interval-cancer cases, including 301 true-positive ROIs related to architectural distortion, and from 52 mammograms of 13 normal cases. For each ROI, the fractal dimension, the entropy of the angular spread of power, 10 Laws' measures, and Haralick's 14 features were computed. The areas under the receiver operating characteristic curves obtained using the features selected by stepwise logistic regression and the leave-one-ROI-out method are 0.76 with the Bayesian classifier, 0.75 with Fisher linear discriminant analysis, and 0.78 with a single-layer feed-forward neural network. Free-response receiver operating characteristics indicated sensitivities of 0.80 and 0.90 at 5.8 and 8.1 false positives per image, respectively, with the Bayesian classifier and the leave-one-image-out method.

Computer-aided detection of architectural distortion in prior mammograms of interval cancer

Architectural distortion is an important sign of breast cancer, but because of its subtlety, it is a common cause of false-negative findings on screening mammograms. This paper presents methods for the detection of architectural distortion in mammograms of interval cancer cases taken prior to the detection of breast cancer using Gabor filters, phase portrait analysis, fractal analysis, and texture analysis. The methods were used to detect initial candidates for sites of architectural distortion in prior mammograms of interval cancer and also normal control cases. A total of 4,224 regions of interest (ROIs) were automatically obtained from 106 prior mammograms of 56 interval cancer cases, including 301 ROIs related to architectural distortion, and from 52 prior mammograms of 13 normal cases. For each ROI, the fractal dimension and Haralick's texture features were computed. Feature selection was performed separately using stepwise logistic regression and stepwise regression. The best results achieved, in terms of the area under the receiver operating characteristics curve, with the features selected by stepwise logistic regression are 0.76 with the Bayesian classifier, 0.73 with Fisher linear discriminant analysis, 0.77 with an artificial neural network based on radial basis functions, and 0.77 with a support vector machine. Analysis of the performance of the methods with free-response receiver operating characteristics indicated a sensitivity of 0.80 at 7.6 false positives per image. The methods have good potential in detecting architectural distortion in mammograms of interval cancer cases.

Development and validation of queries using structured query language (SQL) to determine the utilization of comparison imaging in radiology reports stored on PACS

The purpose of this research was to develop queries that quantify the utilization of comparison imaging in free-text radiology reports. The queries searched for common phrases that indicate whether comparison imaging was utilized, not available, or not mentioned. The queries were iteratively refined and tested on random samples of 100 reports with human review as a reference standard until the precision and recall of the queries did not improve significantly between iterations. Then, query accuracy was assessed on a new random sample of 200 reports. Overall accuracy of the queries was 95.6%. The queries were then applied to a database of 1.8 million reports. Comparisons were made to prior images in 38.69% of the reports (693,955/1,793,754), were unavailable in 18.79% (337,028/1,793,754), and were not mentioned in 42.52% (762,771/1,793,754). The results show that queries of text reports can achieve greater than 95% accuracy in determining the utilization of prior images.

Breast masses: computer-aided diagnosis with serial mammograms

PURPOSE

To retrospectively evaluate effects of computer-aided diagnosis (CAD) involving an interval change classifier (which uses interval change information extracted from prior and current mammograms and estimates a malignancy rating) on radiologists' accuracy in characterizing masses on two-view serial mammograms as malignant or benign.

MATERIALS AND METHODS

The data collection protocol had institutional review board approval. Patient informed consent was waived for this HIPAA-compliant retrospective study. Ninety temporal pairs of two-view serial mammograms (depicting 47 malignant and 43 benign biopsy-proved masses) were obtained from 68 patient files and were digitized. Biopsy was the reference standard. Eight Mammography Quality Standards Act of 1992-accredited radiologists and two breast imaging fellows assessed digitized two-view temporal pairs (in preselected regions of interest only) by estimating likelihood of malignancy and Breast Imaging Reporting and Data System (BI-RADS) category without and with CAD. Observers' rating data were analyzed with Dorfman-Berbaum-Metz (DBM) multireader multicase method. Statistical significance of differences was estimated with the DBM method and Student two-tailed paired t test.

RESULTS

Average area under the receiver operating characteristic curve for likelihood of malignancy across the 10 observers was 0.83 (range, 0.74-0.88) without CAD and improved to 0.87 (range, 0.80-0.92) with CAD (P < .05). The average partial area index above a sensitivity of 0.90 for likelihood of malignancy was 0.35 (range, 0.13-0.54) without CAD and 0.49 (range, 0.18-0.73) with CAD--a nonsignificant improvement (P = .11). For BI-RADS assessment, it was estimated that with CAD, six radiologists would correctly recommend additional biopsies for malignant masses (range, 4.3%-10.6%) and five would correctly recommend reduction of biopsy (ie, fewer biopsies) for benign masses (range, 2.3%-9.3%). However, five radiologists would incorrectly recommend additional biopsy for benign masses (range, 2.3%-14.0%), and one would incorrectly recommend reduction of biopsy (4.3%).

CONCLUSION

CAD involving interval change analysis of preselected regions of interest can significantly improve radiologists' accuracy in classifying masses on digitized screen-film mammograms as malignant or benign.

Use of prior mammograms in the classification of benign and malignant masses

The purpose of this study was to determine the importance of using prior mammograms for classification of benign and malignant masses. Five radiologists and one resident classified mass lesions in 198 mammograms obtained from a population-based screening program. Cases were interpreted twice, once without and once with comparison of previous mammograms, in a sequential reading order using soft copy image display. The radiologists' performances in classifying benign and malignant masses without and with previous mammograms were evaluated with receiver operating characteristic (ROC) analysis. The statistical significance of the difference in performances was calculated using analysis of variance. The use of prior mammograms improved the classification performance of all participants in the study. The mean area under the ROC curve of the readers increased from 0.763 to 0.796. This difference in performance was statistically significant (P = 0.008).

A method to test the reproducibility and to improve performance of computer-aided detection schemes for digitized mammograms

The purpose of this study is to develop a new method for assessment of the reproducibility of computer-aided detection (CAD) schemes for digitized mammograms and to evaluate the possibility of using the implemented approach for improving CAD performance. Two thousand digitized mammograms (representing 500 cases) with 300 depicted verified masses were selected in the study. Series of images were generated for each digitized image by resampling after a series of slight image rotations. A CAD scheme developed in our laboratory was applied to all images to detect suspicious mass regions. We evaluated the reproducibility of the scheme using the detection sensitivity and false-positive rates for the original and resampled images. We also explored the possibility of improving CAD performance using three methods of combining results from the original and resampled images, including simple grouping, averaging output scores, and averaging output scores after grouping. The CAD scheme generated a detection score (from 0 to 1) for each identified suspicious region. A region with a detection score >0.5 was considered as positive. The CAD scheme detected 238 masses (79.3% case-based sensitivity) and identified 1093 false-positive regions (average 0.55 per image) in the original image dataset. In eleven repeated tests using original and ten sets of rotated and resampled images, the scheme detected a maximum of 271 masses and identified as many as 2359 false-positive regions. Two hundred and eighteen masses (80.4%) and 618 false-positive regions (26.2%) were detected in all 11 sets of images. Combining detection results improved reproducibility and the overall CAD performance. In the range of an average false-positive detection rate between 0.5 and 1 per image, the sensitivity of the scheme could be increased approximately 5% after averaging the scores of the regions detected in at least four images. At low false-positive rate (e.g., < or =average 0.3 per image), the grouping method alone could increase CAD sensitivity by 7%. The study demonstrated that reproducibility of a CAD scheme can be tested using a set of slightly rotated and resampled images. Because the reproducibility of true-positive detections is generally higher than that of false-positive detections, combining detection results generated from subsets of rotated and resampled images could improve both reproducibility and overall performance of CAD schemes.

Screening mammography practice essentials

A successful screening mammography practice has three directives. The first directive is quality mammography interpretation, which results in detection of a high percentage of early stage breast cancers, an acceptable recall rate, and an acceptable biopsy rate and yield. The second directive is providing a cost-efficient service. The third directive is access for as many eligible women as possible. Strategies that have helped improve screening mammography access for underserved women are discussed in this article.