Computer-Aided Diagnosis in Mammography

Database characteristics can affect significantly the performance of a mammography CAD scheme. Hence adequate performance comparison among different CAD schemes is not suitable since a single scheme could present different results depending on the set of chosen cases. Images in database shou ld follow a set of quality criteria, since the imaging process up to digital file. CAD schemes can not be developed without a database used to test their efficacy, but each database with particular characteristics may in fluence on the processing scheme performance. A possible solution could be using information on the imaging equipment characteristics. This work describes a preprocessing in order to compensateŽ the image degradation during the acquisition steps, assuring a better uniformityŽ relative to the images quality. Thus, poor quality images would be restored, providing therefore some independen ce on the images source to CAD schemes and allowing to reach the better possible performance. Tests performed with mammography images sets reported a 14% increase in sensitivity for microcalcifications detection. Although this result was followed by a little increase in false positive rates, simple changes in techniques pa rameters can provide the same improvement but with a reduction of the false positive detections. Keywords: Computer-aided diagnosis in Mammography, mi crocalcifications detection, mammography images database

As the rapid advance of digital imaging technologies, the content-based image retrieval (CBIR) has became one of the most vivid research areas in computer vision. In the last several years, developing computer-aided detection and/or diagnosis (CAD) schemes that use CBIR to search for the clinically relevant and visually similar medical images (or regions) depicting suspicious lesions has also been attracting research interest. CBIR-based CAD schemes have potential to provide radiologists with "visual aid" and increase their confidence in accepting CAD-cued results in the decision making. The CAD performance and reliability depends on a number of factors including the optimization of lesion segmentation, feature selection, reference database size, computational efficiency, and relationship between the clinical relevance and visual similarity of the CAD results. By presenting and comparing a number of approaches commonly used in previous studies, this article identifies and discusses the optimal approaches in developing CBIR-based CAD schemes and assessing their performance. Although preliminary studies have suggested that using CBIR-based CAD schemes might improve radiologists' performance and/or increase their confidence in the decision making, this technology is still in the early development stage. Much research work is needed before the CBIR-based CAD schemes can be accepted in the clinical practice.

Artificial intelligence (AI) has the potential to transform medical informatics by supporting clinical decision-making, reducing diagnostic errors, and improving workflows and efficiency. However, successful integration of AI-based decision support systems depends on careful consideration of human-AI collaboration, trust, skill maintenance, and automation bias. This work proposes five central questions to guide future research in medical informatics and human-computer interface (HCI). We focus on AI-based clinical decision support systems, including computer vision algorithms for medical imaging (radiology, pathology), natural language processing for structured and unstructured electronic health record (EHR) data, and rule-based systems. Relevant data modalities include clinician-acquired images, EHR text, and increasingly, patient-generated content in telehealth contexts. We review existing evidence regarding diagnostic errors across specialties, the effectiveness and risks of AI tools in reducing perceptual and interpretive errors, and the human factors influencing diagnostic decision-making in AI-enabled contexts. We synthesize insights from medicine, cognitive science, and HCI to identify gaps in knowledge and propose five key questions for continued research. Diagnostic errors remain common across medicine, with AI offering potential to reduce both perceptual and interpretive errors. However, the impact of AI depends critically on how and when information is presented. Studies indicate that delayed or toggleable cues may outperform immediate ones, but attentional capture, overreliance, and bias remain significant risks. Explainable AI provides transparency but can also bias decisions. Long-term reliance on AI may erode clinician skills, particularly for trainees and in low-prevalence contexts. Historical failures of computer-aided diagnosis in mammography highlight these challenges. Effective AI integration requires human-centered and adaptive design. Five central research questions address: (1) what type and format of information AI should provide; (2) when information should be presented; (3) how explainable AI affects diagnostic decisions; (4) how AI influences automation bias and complacency; and (5) the risks of skill decay due to reliance on AI. Each question underscores the importance of balancing efficiency, accuracy, and clinician expertise while mitigating bias and skill degradation. AI holds promise for improving diagnostic accuracy and efficiency, but realizing its potential requires post-deployment evaluation, equitable access, clinician oversight, and targeted training. AI must complement, rather than replace, human expertise, ensuring safe, effective, and sustainable integration into diagnostic decision-making. Addressing these challenges proactively can maximize AI's potential across healthcare and other high-stakes domains.

: In this research we developed a highly sensitive and specific computer-aided diagnosis classifier based on the likelihood ratio (LRb). The classifier is designed to aid physicians to identify mammographic lesions that should not be sent to biopsy. The classifier was developed using a large database of over five thousand breast biopsy cases from several medical centers. As a result of our research, we have developed a likelihood ratio classifier that can predict biopsy outcome for mass lesions. The performance of the classifier has been tested rigorously including testing on data that was not used for training, and also on data that originated from different medical centers. The results suggest that the LRb is a robust classifier for prediction of biopsy outcome. By decreasing the number of benign mass cases sent to biopsy, the classifier could be a valuable tool for physicians and ultimately beneficial to hospitals and patients.

Early detection of breast cancer, a significant public health problem in the world, is the key for improving breast cancer early prognosis. Mammography is considered the most reliable and widely used diagnostic technique for early detection of breast cancer. However, it is difficult for radiologists to perform both accurate and uniform evaluation for the enormous mammograms with widespread screening. Microcalcification clusters is one of the most important clue of the breast cancer, and their automated detection is very helpful for early breast cancer diagnosis. Because of the poor quality of the mammographic images and the small size of the microcalcifications, it is a very difficult task to perform detecting the early breast cancer. In this paper, we propose a novel approach based on hybrid subspace fusion for detection microcalcification clusters, and successfully apply it to detection task in digital mammograms. In such a system, subspace learning algorithms will be selectively fused according to the ability of preserving the classification information. Experimental results show that the proposed method improved the performance and stability of microcalcification cluster detection and could be adapt to the noise environments better. The proposed methods could get satisfactory results on sensitivity and reduce false positive rate, which provide some new ideas and methods for the research and development of computer-aided detection system in the breast cancer detection community.

The aim of this present clinical pilot study is the display of typical perfusion results in patients with solid, non-cystic breast lesions. The lesions were characterized using contrast enhanced ultrasound (CEUS) with (i) time intensity curve analyses (TIC) and (ii) parametric color maps. The 24 asymptomatic patients included were genetically tested for having an elevated risk for breast cancer. At a center of early detection of familial ovary and breast cancer, those patients received annual MRI and grey-scale ultrasound. If lesions remained unclear or appeared even suspicious, those patients also received CEUS. CEUS was performed after intravenous application of sulfur hexafluoride microbubbles. Digital DICOM cine loops were continuously stored for one minute in PACS (picture archiving and communication system). Perfusion images and TIC analyses were calculated off-line with external perfusion software (VueBox). The lesion diameter ranged between 7 and 15 mm (mean 11 ± 3 mm). Five hypoechoic irregular lesions were scars, 6 lesions were benign and 12 lesions were highly suspicious for breast cancer with irregular enhancement at the margins and a partial wash out. In those 12 cases, histopathology confirmed breast cancer. All the suspicious lesions were correctly identified visually. For the perfusion analysis only Peak Enhancement (PE) and Area Under the Curve (AUC) added more information for correctly identifying the lesions. Typical for benign lesions is a prolonged contrast agent enhancement with lower PE and prolonged wash out, while scars are characterized typically by a reduced enhancement in the center. No differences (p = 0.428) were found in PE in the center of benign lesions (64.2 ± 28.9 dB), malignant lesions (88.1 ± 93.6 dB) and a scar (40.0 ± 17.0 dB). No significant differences (p = 0.174) were found for PE values at the margin of benign lesions (96.4 ± 144.9 dB), malignant lesions (54.3 ± 86.2 dB) or scar tissue (203.8 ± 218.9 dB). Significant differences (p < 0.001) were found in PE of the surrounding tissue when comparing benign lesions (33.6 ± 25.2 dB) to malignant lesions (15.7 ± 36.3 dB) and scars (277.2 ± 199.9 dB). No differences (p = 0.821) were found in AUC in the center of benign lesions (391.3 ± 213.7), malignant lesions (314.7 ± 643.9) and a scar (213.1 ± 124.5). No differences (p = 0.601) were found in AUC values of the margin of benign lesions (313.3 ± 372.8), malignant lesions (272.6 ± 566.4) or scar tissue (695.0 ± 360.6). Significant differences (p < 0.01) were found in AUC of the surrounding tissue for benign lesions (151.7 ± 127.8), malignant lesions (177.9 ± 1345.6) and scars (1091 ± 693.3). There were no differences in perfusion evaluation for mean transit time (mTT), rise time (RT) and time to peak (TTP) when comparing the center to the margins and the surrounding tissue. The CEUS perfusion parameters PE and AUC allow a very good assessment of the risk of malignant breast lesions and thus a downgrading of BI-RADS 4 lesions. The use of the external perfusion software (VueBox, Bracco, Milan, Italy) did not lead to any further improvement in the diagnosis of suspicious breast lesions and does appears not to have any additional diagnostic value in breast lesions.

: The objective of the proposed research is to develop computer-aided diagnosis methods for use in mammography in order to increase the diagnostic accuracy of radiologists. Specifically we have developed advanced computerized schemes for the detection spiculated lesions and architectural distortions based on the calculation of the Hough spectrum and for the detection of small, low-contrast early cancers based on gradient and circularity filters. Also, computerized classification schemes for masses using artificial neural networks, rule-based methods, and hybrid systems have been developed. We have investigated the effect of database on feature selection and classifier training. We also investigated a computerized method for including temporal change between mammographic examinations. We have also developed an intelleigent search workstation for aiding radiologist in making diagnostic decisions by providing them with both CAD output and images of known cases that are 'similar' to the case in question. The efficacy and efficiency of the CAD methods for mammography are being evaluated on a clinical workstation. The potential significance of this research project lies in the fact that if the detectability of cancers can be increased by employing a computer to aid the radiologist's diagnosis, then the treatment of patients with cancer can be initiated earlier and their chance of survival improved.

The research presented in this dissertation represents an innovative application of computer‐aided diagnosis and signal detection theory to the specific task of early detection of breast cancer in the context of screening mammography. A number of automated schemes have been developed in our laboratory to detect masses and clustered microcalcifications in digitized mammograms, on the one hand, and to classify known lesions as malignant or benign, on the other. The development of fully automated classification schemes is difficult, because the output of a detection scheme will contain false‐positive detections in addition to detected malignant and benign lesions, resulting in a three‐class classification task. Researchers have so far been unable to extend successful tools for analyzing two‐class classification tasks, such as receiver operating characteristic (ROC) analysis, to three‐class classification tasks. The goals of our research were to use Bayesian artificial neural networks to estimate ideal observer decision variables to both detect and classify clustered microcalcifications and mass lesions in mammograms, and to derive substantial theoretical results indicating potential avenues of approach toward the three‐class classification task. Specifically, we have shown that an ideal observer in an N‐class classification task achieves an optimal ROC hypersurface, just as the two‐class ideal observer achieves an optimal ROC curve; and that an obvious generalization of a well‐known two‐class performance metric, the area under the ROC curve, is not useful as a performance metric in classification tasks with more than two classes. This work is significant for three reasons. First, it involves the explicit estimation of feature‐based (as opposed to image‐based) ideal observer decision variables in the tasks of detecting and classifying mammographic lesions. Second, it directly addresses the three‐class classification task of distinguishing malignant lesions, benign lesions, and false‐positive computer detections. Finally, it develops important theoretical results for N‐class classification tasks that should prove of value in the development of a three‐class extension to ROC analysis methods.

3051 Background: The implementation of mammogram-based screening has significantly improved the early detection of breast cancer in the west. However, the use of screening mammography is less prevalent in Asia partly due to social and cultural reasons. The aim of this study was to determine if a serum microRNA (miRNA) panel could be used as biomarkers to assist in the early detection of breast cancer. Methods: We conducted a multi-center, multi-ethnic study to identify and validate miRNA biomarkers for the early detection of breast cancer. A total of 1070 subjects including 550 breast cancer cases (predominantly stage 1 and 2) and 520 matched controls from 6 independent sources were included in this study. Among these, there were 768 American and European subjects recruited by biobanks and 302 Singaporean Asian Subjects recruited at the National Cancer Centre Singapore and the National University Hospital. The study was conducted in 3 phases. First, 119 European Caucasian serum samples (Discovery Cohort) were interrogated to identify differentially expressed miRNAs between early-stage breast cancer cases and matched controls, among 520 circulating miRNA candidates by quantitative RT-PCR using MiRXES assays. The remaining 951 subjects from 5 independent sources were assigned into two groups for biomarker optimization/algorithm development (Optimization Cohort, n = 451) and validation (Validation Cohort, n = 500). Results: Among the 520 circulating miRNAs measured, 241 were quantified in absolute copy numbers for 119 subjects in the Discovery Cohort. Thirty-two candidate miRNAs were identified. These miRNAs consistently showed differential expression between cancer and control subjects in the Optimization Cohort. A multi-variant panel of 8 miRNAs and an algorithm was developed using the combined Discovery and Optimization Cohorts, with an AUC of &gt; 0.90. When validated in the independent Validation Cohort, the panel demonstrated an AUC of &gt; 0.85. Conclusions: We developed and validated a serum miRNA panel that is both sensitive (~70%) and specific (~90%) in detecting early-stage breast cancer.

To investigate the effect of different breast lesions on exposure parameters in digital mammography and to determine whether the exposure parameters can additively improve diagnostic efficiency. Craniocaudal view and mediolateral view full-field digital mammography images from 982 women with unilateral lesions (341 with malignant lesions, 189 with benign lesions, and 452 healthy women) obtained at Nanfang Hospital were reviewed. Differences in exposure parameters (tube voltage and load, breast thickness (BT), and average glandular dose (AGD)) between breasts were calculated. The relationships between parameter differences and lesion size were explored. A logistic regression model was used based on the AGD and BT differences, and the area under the receiver operating characteristic curve (AUC) was used to assess the performance of these parameters in differentiating malignant from benign and healthy subjects. Independently, data from 129 women (82 with malignant and 47 with benign lesions) treated at Sun Yat-sen Memorial Hospital were collected to validate the model. Differences in tube voltage and load, BT, and AGD between breasts were significantly greater in the malignant subjects than benign (p < 0.05) and healthy subjects (p < 0.05). The AUCs for the comparisons of malignant vs. healthy subjects, malignant vs. benign subjects, and benign vs. healthy subjects were 0.77 ± 0.02, 0.72 ± 0.02, and 0.57 ± 0.02, respectively. The model combining the exposure parameters with the BI-RADS category resulted in a higher AUC (0.910 ± 0.03) compared with physician diagnosis alone (0.820 ± 0.04) for differentiating between malignant and benign lesions. Exposure parameters additively improved diagnostic accuracy for breast cancer and yielded more reliable results. • Differences in kVp, mAs, BT, and AGD between breasts were significantly greater in the malignant subjects than benign and healthy subjects. • The model combining exposure parameters with the BI-RADS category resulted in a higher AUC compared with the physician's diagnosis for differentiating between malignant and benign lesions. • Exposure parameters additively improved diagnostic accuracy for breast cancer.

Background: Breast cancer is one of the most common causes of death among women worldwide. Early detection and diagnosis will be helpful to reduce mortality and improve prognosis. It is urgent to develop efficient detection technology for breast cancer. Mammographic screening is a valuable tool for early detection of breast cancer. However, the increased density of breast tissue significantly reduces the diagnostic accuracy. Aim of Study: To provide an overview of the different reported elasticities of specific breast pathologies based on ultrasound elastography. Material and Methods: A total of 35 articles including 8316 patients and 9057 breast lesions were included in the pooled analysis of which 3060 malignant lesions were included from 40 studies. The median incidence of malignant breast lesion is 37.1% calculated from the incidence of malignant lesions of all included studies. Seven of the included studies assessed VTIQ. Mean age varied along all included studies. Results: The sensitivity and specificity of Emax, Emean and Eratio for the diagnosis of breast cancer varied according to the interpretative criteria used to define a test as positive. The summary estimates of sensitivity and specificity were 82.58% (95% CI 78.32% to 86.16%) and 84.12% (95% CI 79.07% to 87.07%) for Emean, 86.19% (95% CI 81.60% to 89.77%) and 88.56% (95% CI 88.56% to 91.54%) for Emax, and 87.50% (95% CI 77.47% to 93.44%) and 79.30% (95% CI 68.21% to 87.24%) for Eratio respectively. Regarding DOR, Emax achieved the highest value 48.32 (95% CI 28.7 to 67.8) which means there are 48 times the odds of obtaining an Emax positive result in a diseased rather than a non-diseased person. Meta-regression analysis was conducted to assess the impact of two covaries; Emean and Emax using Likelihood ratio test and revealed significant difference existed with higher summary sensitivity (x2=35.04, p < 001) and specificity (x2=18.65, p < 001) in Emax than Emean. SROC curves were used to show the distribution of sensitivity and specificity of Emax, Emean and Eratio in the Roc space as well as the prediction region. Conclusion: Our meta-analysis demonstrates that SWE is an accurate and reliable diagnostic tool in discriminating malignant and benign breast lesions. With wide application, SWE may significantly improve the early diagnostic of breast cancer. SWE can provide additional information on predicting breast cancer prognosis. However, the possibility of false-positive and false-negative results should be considered during interpretation.

Fractal analysis of mammographic lesions: a feasibility study quantifying the difference between benign and malignant masses.

Detection and characterization of microcalcification clusters in mammograms is vital in daily clinical practice. The problem of detecting possible cancer areas is very complex due, on one hand, to the diversity in shape of the ill tissue and, on the other hand, to the poorly defined border between the healthy and the cancerous zone. Even though it has been studied for many years, there are still remaining challenges and directions for future research, such as developing better enhancement and segmentation algorithms. In this paper, we propose a microcalcification detection algorithm using an extended region growing technique in a mammography CAD (computer-aided diagnosis) system.

To evaluate and compare the outcome of stereotactically guided percutaneous, directional, vacuum-assisted breast biopsy (VAB) with the outcome of those performed with 14-gauge automated gun. Between January 2008 and January 2011, altogether 257 stereotactic biopsies were performed in our institute; 193 with the 14-gauge automated needle and 64 with a directional, vacuum-assisted instrument. Patients with malignant or high risk lesions were encouraged to undergo an excisional biopsy. We evaluated and compared concordance between the biopsy results and subsequent examinations (surgical excision or follow up) for both methods. VAB demonstrated 57 (89 %) benign lesions, 2 (3 %) high risk and 5 (8 %) malignant lesions. We obtained post-bioptic mammographic or histological evaluation for altogether 51 lesions, with one (12.5 %) false negative case, one case of underestimation of borderline lesion, and no false positive cases. Clusters of microcalcifiation were completely removed in 18 lesions (28 %). Vacuum-assisted biopsy outperformed the 14-gauge automated needle biopsy with a lower disease underestimation, sensitivity and false-negative rates; however these differences were not statistically significant due to small data files and relatively small number of identified malignant lesions in the vacuum-assisted biopsy group.VAB completely removed significantly more clusters of microcalcifications in comparison to the automated needle biopsy, which significantly reduced the need for surgery in benign lesions (Tab. 7, Fig. 3, Ref. 32).

We review the evidence on computer-aided detection (CAD) as an adjunct to mammography interpretation, and discuss the complexity of its impact on decision-making and potential medico-legal aspects. CAD prompts the reader to review lesions on the mammogram and re-evaluate the decision on whether to recall CAD-prompted findings. Studies show that CAD can improve the sensitivity of a single reader, with an incremental cancer detection rate (from adding CAD to a single read) ranging between 1 and 19%. However, CAD will also substantially increase the recall rate (decrease the reader's specificity) causing additional recall in approximately 6-35% of women. Evidence indicates that CAD does not perform as well as double (human) reading in the context of organized breast screening where double reading is the standard of care. Although CAD can identify and prompt readers to missed cancers, the high number of false-positive prompts (1.5-4 false prompts per case) can have an adverse effect on clinical decision-making, and detracts from CAD's application in screening practice. Refinement in CAD algorithms, in combination with increasing implementation of digital mammography, may improve the potential use of CAD in mammography reading, but will require prospective evaluation.