A randomised controlled trial of digital breast tomosynthesis vs digital mammography as primary screening tests: Screening results over subsequent episodes of the Proteus Donna study.

Proteus Donna is a randomised controlled trial aimed at prospectively evaluating screening with digital breast tomosynthesis (DBT), including interval cancer detection (ICD) and cancer detection (CD) in the analysis as a cumulative measure over subsequent screening episodes. Consenting women aged 46 to 68 attending the regional Breast Screening Service were randomly assigned to conventional digital mammography (DM, control arm) or DBT in addition to DM (DBT, study arm). At the subsequent round all participants underwent DM. Thirty-six months follow-up allowed for the identification of cancers detected in the subsequent screening and interscreening interval. Relative risk (RR) and 95% confidence interval (95% CI) were computed. Cumulative CD and Nelson-Aalen incidence were analysed over the follow-up period. Between 31 December 2014 and 31 December 2017, 43 022 women were randomised to DM and 30 844 to DBT. At baseline, CD was significantly higher (RR: 1.44, 95% CI: 1.21-1.71) in the study arm. ICD did not differ significantly between the two arms (RR: 0.92, 95% CI: 0.62-1.35). At subsequent screening with DM, the CD was lower (nearly significant) in the study arm (RR: 0.83, 95% CI: 0.65-1.06). Over the follow-up period, the cumulative CD (comprehensive of ICD) was slightly higher in the study arm (RR: 1.15, 95% CI: 1.01-1.31). The Nelson-Aalen cumulative incidence over time remained significantly higher in the study arm for approximately 24 months. Benign lesions detection was higher in the study arm at baseline and lower at subsequent tests. Outcomes are consistent with a lead time gain of DBT compared to DM, with an increase in false positives and moderate overdiagnosis.

AI-Enhanced Diagnosis of Challenging Lesions in Breast MRI: A Methodology and Application Primer.

Computer-aided diagnosis (CAD) systems have become an important tool in the assessment of breast tumors with magnetic resonance imaging (MRI). CAD systems can be used for the detection and diagnosis of breast tumors as a "second opinion" review complementing the radiologist's review. CAD systems have many common parts, such as image preprocessing, tumor feature extraction, and data classification that are mostly based on machine-learning (ML) techniques. In this review article, we describe applications of ML-based CAD systems in MRI covering the detection of diagnostically challenging lesions of the breast such as nonmass enhancing (NME) lesions, and furthermore discuss how multiparametric MRI and radiomics can be applied to the study of NME, including prediction of response to neoadjuvant chemotherapy (NAC). Since ML has been widely used in the medical imaging community, we provide an overview about the state-of-the-art and novel techniques applied as classifiers to CAD systems. The differences in the CAD systems in MRI of the breast for several standard and novel applications for NME are explained in detail to provide important examples, illustrating: 1) CAD for detection and diagnosis, 2) CAD in multiparametric imaging, 3) CAD in NAC, and 4) breast cancer radiomics. We aim to provide a comparison between these CAD applications and to illustrate a global view on intelligent CAD systems based on machine and deep learning in MRI of the breast. LEVEL OF EVIDENCE: 2 TECHNICAL EFFICACY STAGE: 2.

Flexible Sigmoidoscopy and CT Colonography Screening: Patients' Experience with and Factors for Undergoing Screening-Insight from the Proteus Colon Trial.

Purpose To compare the acceptability of computed tomographic (CT) colonography and flexible sigmoidoscopy (FS) screening and the factors predicting CT colonographic screening participation, targeting participants in a randomized screening trial. Materials and Methods Eligible individuals aged 58 years (n = 1984) living in Turin, Italy, were randomly assigned to be invited to screening for colorectal cancer with FS or CT colonography. After individuals who had died or moved away (n = 28) were excluded, 264 of 976 (27.0%) underwent screening with FS and 298 of 980 (30.4%) underwent CT colonography. All attendees and a sample of CT colonography nonattendees (n = 299) were contacted for a telephone interview 3-6 months after invitation for screening, and screening experience and factors affecting participation were investigated. Odds ratios (ORs) were computed by means of multivariable logistic regression. Results For the telephone interviews, 239 of 264 (90.6%) FS attendees, 237 of 298 (79.5%) CT colonography attendees, and 182 of 299 (60.9%) CT colonography nonattendees responded. The percentage of attendees who would recommend the test to friends or relatives was 99.1% among FS and 93.3% among CT colonography attendees. Discomfort associated with bowel preparation was higher among CT colonography than FS attendees (OR, 2.77; 95% confidence interval [CI]: 1.47, 5.24). CT colonography nonattendees were less likely to be men (OR, 0.36; 95% CI: 0.18, 0.71), retired (OR, 0.31; 95% CI: 0.13, 0.75), to report regular physical activity (OR, 0.37; 95% CI: 0.20, 0.70), or to have read the information leaflet (OR, 0.18; 95% CI: 0.08, 0.41). They were more likely to mention screening-related anxiety (mild: OR, 6.30; 95% CI: 2.48, 15.97; moderate or severe: OR, 3.63; 95% CI: 1.87, 7.04), erroneous beliefs about screening (OR, 32.15; 95% CI: 6.26, 165.19), or having undergone a recent fecal occult blood test (OR, 13.69; 95% CI: 3.66, 51.29). Conclusion CT colonography and FS screening are well accepted, but further reducing the discomfort from bowel preparation may increase CT colonography screening acceptability. Negative attitudes, erroneous beliefs about screening, and organizational barriers are limiting screening uptake; all these factors are modifiable and therefore potentially susceptible to interventions. © RSNA, 2017 Online supplemental material is available for this article.

Computer-based self-training for CT colonography with and without CAD.

OBJECTIVES: To determine whether (1) computer-based self-training for CT colonography (CTC) improves interpretation performance of novice readers; (2) computer-aided detection (CAD) use during training affects learning. METHODS: Institutional review board approval and patients' informed consent were obtained for all cases included in this study. Twenty readers (17 radiology residents, 3 radiologists) with no experience in CTC interpretation were recruited in three centres. After an introductory course, readers performed a baseline assessment test (37 cases) using CAD as second reader. Then they were randomized (1:1) to perform either a computer-based self-training (150 cases verified at colonoscopy) with CAD as second reader or the same training without CAD. The same assessment test was repeated after completion of the training programs. Main outcome was per lesion sensitivity (≥ 6 mm). A generalized estimating equation model was applied to evaluate readers' performance and the impact of CAD use during training. RESULTS: After training, there was a significant improvement in average per lesion sensitivity in the unassisted phase, from 74% (356/480) to 83% (396/480) (p < 0.001), and in the CAD-assisted phase, from 83% (399/480) to 87% (417/480) (p = 0.021), but not in average per patient sensitivity, from 93% (390/420) to 94% (395/420) (p = 0.41), and specificity, from 81% (260/320) to 86% (276/320) (p = 0.15). No significant effect of CAD use during training was observed on per patient sensitivity and specificity, nor on per lesion sensitivity. CONCLUSIONS: A computer-based self-training program for CTC improves readers' per lesion sensitivity. CAD as second reader does not have a significant impact on learning if used during training. KEY POINTS: • Computer-based self-training for CT colonography improves per lesion sensitivity of novice readers. • Self-training program does not increase per patient specificity of novice readers. • CAD used during training does not have significant impact on learning.

JOURNAL CLUB: Extracolonic Findings at CT Colonography: Systematic Review and Meta-Analysis.

OBJECTIVE: The purpose of this study was to perform a systematic review and meta-analysis of published studies on CT colonography (CTC) in which extracolonic findings were assessed. MATERIALS AND METHODS: A systematic review of studies of screening CTC and of CTC to evaluate symptoms (1994-June 2017) was conducted to estimate the rate of extra-colonic findings and associated additional workup recommendations. The primary outcome was potentially important extracolonic findings, defined as CT Colonography Imaging Reporting and Data System (C-RADS) category E4 or as having high clinical importance (if C-RADS was not used). Secondary outcomes included likely unimportant extracolonic findings (i.e., C-RADS category E3 or similar). Random-effects and meta-regression analyses were used to generate pooled estimates and to explore risk factors for extracolonic findings related to various cohort characteristics. RESULTS: Primary data were acquired from 44 studies (49,676 patients). The pooled rate of potentially important extracolonic findings was 4.9% (95% CI, 3.7-6.4%) with a high degree of heterogeneity (I2 = 95%). This estimate progressively declined over time (9% per year since 2006) and was significantly related to the reporting system (lower for C-RADS than for low, moderate, high clinical significance reporting). C-RADS-specific meta-analysis (32,746 patients) showed rates of potentially significant extracolonic findings in 2.8% (95% CI, 1.9-3.5%) of the cohort without symptoms and 5.2% (95% CI, 3.5-7.6%) of the cohort with symptoms and in 5.7% (95% CI, 3.3-9.8%) of seniors (≥ 65 years) versus 2.3% (95% CI, 1.2-4.5%) of those younger than 65 years. The overall pooled rates of recommended workup were 8.2% (95% CI, 6.6-10.1%) for all extracolonic findings and 4.0% (95% CI, 2.7-5.9%) for potentially important extracolonic findings. CONCLUSION: With use of the more robust C-RADS classification, potentially important extracolonic findings at CTC occur in less than 3% of cohorts without symptoms. For all extracolonic findings, the rate of suggested or recommended additional workup is approximately 8% but decreases to 4% for potentially important extracolonic findings.

Comparing CT colonography and flexible sigmoidoscopy: a randomised trial within a population-based screening programme.

IMPORTANCE AND AIMS: The role of CT colonography (CTC) as a colorectal cancer (CRC) screening test is uncertain. The aim of our trial was to compare participation and detection rate (DR) with sigmoidoscopy (flexible sigmoidoscopy (FS)) and CTC in a screening setting. DESIGN SETTING AND PARTICIPANTS: We conducted two randomised clinical trials (RCTs). (1) Participation RCT: individuals, aged 58 years, living in Turin (Italy), were randomly assigned to be invited to FS or CTC screening; (2) detection RCT: residents in northern Italy, aged 58-60, giving their consent to recruitment, were randomly allocated to CTC or FS. Polyps ≥6 mm at CTC, or 'high-risk' distal lesions at FS, were referred for colonoscopy (TC). MAIN OUTCOME MEASURES: Participation rate (proportion of invitees examined); DR of advanced adenomas or CRC (advanced neoplasia (AN)). RESULTS: Participation was 30.4% (298/980) for CTC and 27.4% (267/976) for FS (relative risk (RR) 1.1; 95% CI 0.98 to 1.29). Among men, participation was higher with CTC than with FS (34.1% vs 26.5%, p=0.011). In the detection RCT, 2673 subjects had FS and 2595 had CTC: the AN DR was 4.8% (127/2673, including 9 CRCs) with FS and 5.1% (133/2595, including 10 CRCs) with CTC (RR 1.08; 95% CI 0.85 to 1.37). Distal AN DR was 3.9% (109/2673) with FS and 2.9% (76/2595) with CTC (RR 0.72; 95% CI 0.54 to 0.96); proximal AN DR was 1.2% (34/2595) for FS vs 2.7% (69/2595) for CTC (RR 2.06; 95% CI 1.37 to 3.10). CONCLUSIONS AND RELEVANCE: Participation and DR for FS and CTC were comparable. AN DR was twice as high in the proximal colon and lower in the distal colon with CTC than with FS. Men were more likely to participate in CTC screening. TRIAL REGISTRATION NUMBER: NCT01739608; Pre-results.

Mammographic density: Comparison of visual assessment with fully automatic calculation on a multivendor dataset.

OBJECTIVES: To compare breast density (BD) assessment provided by an automated BD evaluator (ABDE) with that provided by a panel of experienced breast radiologists, on a multivendor dataset. METHODS: Twenty-one radiologists assessed 613 screening/diagnostic digital mammograms from nine centers and six different vendors, using the BI-RADS a, b, c, and d density classification. The same mammograms were also evaluated by an ABDE providing the ratio between fibroglandular and total breast area on a continuous scale and, automatically, the BI-RADS score. A panel majority report (PMR) was used as reference standard. Agreement (κ) and accuracy (proportion of cases correctly classified) were calculated for binary (BI-RADS a-b versus c-d) and 4-class classification. RESULTS: While the agreement of individual radiologists with the PMR ranged from κ = 0.483 to κ = 0.885, the ABDE correctly classified 563/613 mammograms (92 %). A substantial agreement for binary classification was found for individual reader pairs (κ = 0.620, standard deviation [SD] = 0.140), individual versus PMR (κ = 0.736, SD = 0.117), and individual versus ABDE (κ = 0.674, SD = 0.095). Agreement between ABDE and PMR was almost perfect (κ = 0.831). CONCLUSIONS: The ABDE showed an almost perfect agreement with a 21-radiologist panel in binary BD classification on a multivendor dataset, earning a chance as a reproducible alternative to visual evaluation. KEY POINTS: Individual BD assessment differs from PMR with κ as low as 0.483. An ABDE correctly classified 92 % of mammograms with almost perfect agreement (κ = 0.831). An ABDE can be a valid alternative to subjective BD assessment.

Breast Cancer: Computer-aided Detection with Digital Breast Tomosynthesis.

PURPOSE: To evaluate a commercial tomosynthesis computer-aided detection (CAD) system in an independent, multicenter dataset. MATERIALS AND METHODS: Diagnostic and screening tomosynthesis mammographic examinations (n = 175; cranial caudal and mediolateral oblique) were randomly selected from a previous institutional review board-approved trial. All subjects gave informed consent. Examinations were performed in three centers and included 123 patients, with 132 biopsy-proven screening-detected cancers, and 52 examinations with negative results at 1-year follow-up. One hundred eleven lesions were masses and/or microcalcifications (72 masses, 22 microcalcifications, 17 masses with microcalcifications) and 21 were architectural distortions. Lesions were annotated by radiologists who were aware of all available reports. CAD performance was assessed as per-lesion sensitivity and false-positive results per volume in patients with negative results. RESULTS: Use of the CAD system showed per-lesion sensitivity of 89% (99 of 111; 95% confidence interval: 81%, 94%), with 2.7 ± 1.8 false-positive rate per view, 62 of 72 lesions detected were masses, 20 of 22 were microcalcification clusters, and 17 of 17 were masses with microcalcifications. Overall, 37 of 39 microcalcification clusters (95% sensitivity, 95% confidence interval: 81%, 99%) and 79 of 89 masses (89% sensitivity, 95% confidence interval: 80%, 94%) were detected with the CAD system. On average, 0.5 false-positive rate per view were microcalcification clusters, 2.1 were masses, and 0.1 were masses and microcalcifications. CONCLUSION: A digital breast tomosynthesis CAD system can allow detection of a large percentage (89%, 99 of 111) of breast cancers manifesting as masses and microcalcification clusters, with an acceptable false-positive rate (2.7 per breast view). Further studies with larger datasets acquired with equipment from multiple vendors are needed to replicate the findings and to study the interaction of radiologists and CAD systems.

AI and machine learning in medical imaging: key points from development to translation.

Innovation in medical imaging artificial intelligence (AI)/machine learning (ML) demands extensive data collection, algorithmic advancements, and rigorous performance assessments encompassing aspects such as generalizability, uncertainty, bias, fairness, trustworthiness, and interpretability. Achieving widespread integration of AI/ML algorithms into diverse clinical tasks will demand a steadfast commitment to overcoming issues in model design, development, and performance assessment. The complexities of AI/ML clinical translation present substantial challenges, requiring engagement with relevant stakeholders, assessment of cost-effectiveness for user and patient benefit, timely dissemination of information relevant to robust functioning throughout the AI/ML lifecycle, consideration of regulatory compliance, and feedback loops for real-world performance evidence. This commentary addresses several hurdles for the development and adoption of AI/ML technologies in medical imaging. Comprehensive attention to these underlying and often subtle factors is critical not only for tackling the challenges but also for exploring novel opportunities for the advancement of AI in radiology.

Artificial intelligence in medicine: mitigating risks and maximizing benefits via quality assurance, quality control, and acceptance testing.

The adoption of artificial intelligence (AI) tools in medicine poses challenges to existing clinical workflows. This commentary discusses the necessity of context-specific quality assurance (QA), emphasizing the need for robust QA measures with quality control (QC) procedures that encompass (1) acceptance testing (AT) before clinical use, (2) continuous QC monitoring, and (3) adequate user training. The discussion also covers essential components of AT and QA, illustrated with real-world examples. We also highlight what we see as the shared responsibility of manufacturers or vendors, regulators, healthcare systems, medical physicists, and clinicians to enact appropriate testing and oversight to ensure a safe and equitable transformation of medicine through AI.

Data augmentation for medical imaging: A systematic literature review.

Recent advances in Deep Learning have largely benefited from larger and more diverse training sets. However, collecting large datasets for medical imaging is still a challenge due to privacy concerns and labeling costs. Data augmentation makes it possible to greatly expand the amount and variety of data available for training without actually collecting new samples. Data augmentation techniques range from simple yet surprisingly effective transformations such as cropping, padding, and flipping, to complex generative models. Depending on the nature of the input and the visual task, different data augmentation strategies are likely to perform differently. For this reason, it is conceivable that medical imaging requires specific augmentation strategies that generate plausible data samples and enable effective regularization of deep neural networks. Data augmentation can also be used to augment specific classes that are underrepresented in the training set, e.g., to generate artificial lesions. The goal of this systematic literature review is to investigate which data augmentation strategies are used in the medical domain and how they affect the performance of clinical tasks such as classification, segmentation, and lesion detection. To this end, a comprehensive analysis of more than 300 articles published in recent years (2018-2022) was conducted. The results highlight the effectiveness of data augmentation across organs, modalities, tasks, and dataset sizes, and suggest potential avenues for future research.

Deep learning-based algorithm for postoperative glioblastoma MRI segmentation: a promising new tool for tumor burden assessment.

OBJECTIVE: Clinical and surgical decisions for glioblastoma patients depend on a tumor imaging-based evaluation. Artificial Intelligence (AI) can be applied to magnetic resonance imaging (MRI) assessment to support clinical practice, surgery planning and prognostic predictions. In a real-world context, the current obstacles for AI are low-quality imaging and postoperative reliability. The aim of this study is to train an automatic algorithm for glioblastoma segmentation on a clinical MRI dataset and to obtain reliable results both pre- and post-operatively. METHODS: The dataset used for this study comprises 237 (71 preoperative and 166 postoperative) MRIs from 71 patients affected by a histologically confirmed Grade IV Glioma. The implemented U-Net architecture was trained by transfer learning to perform the segmentation task on postoperative MRIs. The training was carried out first on BraTS2021 dataset for preoperative segmentation. Performance is evaluated using DICE score (DS) and Hausdorff 95% (H95). RESULTS: In preoperative scenario, overall DS is 91.09 (± 0.60) and H95 is 8.35 (± 1.12), considering tumor core, enhancing tumor and whole tumor (ET and edema). In postoperative context, overall DS is 72.31 (± 2.88) and H95 is 23.43 (± 7.24), considering resection cavity (RC), gross tumor volume (GTV) and whole tumor (WT). Remarkably, the RC segmentation obtained a mean DS of 63.52 (± 8.90) in postoperative MRIs. CONCLUSIONS: The performances achieved by the algorithm are consistent with previous literature for both pre-operative and post-operative glioblastoma's MRI evaluation. Through the proposed algorithm, it is possible to reduce the impact of low-quality images and missing sequences.

AAPM task group report 273: Recommendations on best practices for AI and machine learning for computer-aided diagnosis in medical imaging.

Rapid advances in artificial intelligence (AI) and machine learning, and specifically in deep learning (DL) techniques, have enabled broad application of these methods in health care. The promise of the DL approach has spurred further interest in computer-aided diagnosis (CAD) development and applications using both "traditional" machine learning methods and newer DL-based methods. We use the term CAD-AI to refer to this expanded clinical decision support environment that uses traditional and DL-based AI methods. Numerous studies have been published to date on the development of machine learning tools for computer-aided, or AI-assisted, clinical tasks. However, most of these machine learning models are not ready for clinical deployment. It is of paramount importance to ensure that a clinical decision support tool undergoes proper training and rigorous validation of its generalizability and robustness before adoption for patient care in the clinic. To address these important issues, the American Association of Physicists in Medicine (AAPM) Computer-Aided Image Analysis Subcommittee (CADSC) is charged, in part, to develop recommendations on practices and standards for the development and performance assessment of computer-aided decision support systems. The committee has previously published two opinion papers on the evaluation of CAD systems and issues associated with user training and quality assurance of these systems in the clinic. With machine learning techniques continuing to evolve and CAD applications expanding to new stages of the patient care process, the current task group report considers the broader issues common to the development of most, if not all, CAD-AI applications and their translation from the bench to the clinic. The goal is to bring attention to the proper training and validation of machine learning algorithms that may improve their generalizability and reliability and accelerate the adoption of CAD-AI systems for clinical decision support.

Self-supervised and semi-supervised learning for road condition estimation from distributed road-side cameras.

Monitoring road conditions, e.g., water build-up due to intense rainfall, plays a fundamental role in ensuring road safety while increasing resilience to the effects of climate change. Distributed cameras provide an easy and affordable alternative to instrumented weather stations, enabling diffused and capillary road monitoring. Here, we propose a deep learning-based solution to automatically detect wet road events in continuous video streams acquired by road-side surveillance cameras. Our contribution is two-fold: first, we employ a convolutional Long Short-Term Memory model (convLSTM) to detect subtle changes in the road appearance, introducing a novel temporally consistent data augmentation to increase robustness to outdoor illumination conditions. Second, we present a contrastive self-supervised framework that is uniquely tailored to surveillance camera networks. The proposed technique was validated on a large-scale dataset comprising roughly 2000 full day sequences (roughly 400K video frames, of which 300K unlabelled), acquired from several road-side cameras over a span of two years. Experimental results show the effectiveness of self-supervised and semi-supervised learning, increasing the frame classification performance (measured by the Area under the ROC curve) from 0.86 to 0.92. From the standpoint of event detection, we show that incorporating temporal features through a convLSTM model both improves the detection rate of wet road events (+ 10%) and reduces false positive alarms ([Formula: see text] 45%). The proposed techniques could benefit also other tasks related to weather analysis from road-side and vehicle-mounted cameras.

Performance of a fully automatic lesion detection system for breast DCE-MRI.

PURPOSE: To describe and test a new fully automatic lesion detection system for breast DCE-MRI. MATERIALS AND METHODS: Studies were collected from two institutions adopting different DCE-MRI sequences, one with and the other one without fat-saturation. The detection pipeline consists of (i) breast segmentation, to identify breast size and location; (ii) registration, to correct for patient movements; (iii) lesion detection, to extract contrast-enhanced regions using a new normalization technique based on the contrast-uptake of mammary vessels; (iv) false positive (FP) reduction, to exclude contrast-enhanced regions other than lesions. Detection rate (number of system-detected malignant and benign lesions over the total number of lesions) and sensitivity (system-detected malignant lesions over the total number of malignant lesions) were assessed. The number of FPs was also assessed. RESULTS: Forty-eight studies with 12 benign and 53 malignant lesions were evaluated. Median lesion diameter was 6 mm (range, 5-15 mm) for benign and 26 mm (range, 5-75 mm) for malignant lesions. Detection rate was 58/65 (89%; 95% confidence interval [CI] 79%-95%) and sensitivity was 52/53 (98%; 95% CI 90%-99%). Mammary median FPs per breast was 4 (1st-3rd quartiles 3-7.25). CONCLUSION: The system showed promising results on MR datasets obtained from different scanners producing fat-sat or non-fat-sat images with variable temporal and spatial resolution and could potentially be used for early diagnosis and staging of breast cancer to reduce reading time and to improve lesion detection. Further evaluation is needed before it may be used in clinical practice.

Comparison of three different iodine-based bowel regimens for CT colonography.

OBJECTIVE: The aim of this study was to compare the computed tomographic colonography (CTC) image quality and patient acceptance of three iodine-based faecal tagging bowel preparations in 60 patients undergoing the following regimens: a 2-day regimen of meal-time administration of iodine and phospho-soda (GFPH); a 2-day regimen of meal-time mild laxative, followed by iodine administered 2 h before CTC (SD); and a 2-day regimen of meal-time administration of iodine (GF). METHODS: Two independent radiologists assessed tagging quality; quantitative measures included the tagged stool density, and computer-aided detection (CAD) false-positive rate. RESULTS: The GFPH and SD regimens provided better subjective quality than GF (p < 0.001). The latter regimen resulted in a higher proportion of insufficiently tagged segments: the measured average stool density was less than 200 HU in 10.7% in all segments vs 3.6% for SD and <0.5% for GFPH, respectively. Insufficient tagging occurred mostly in the ascending colon and the caecum. The CAD false-positive rate increased following the trend: GFPH < SD < GF (p = 0.00012). GFPH was worse tolerated than SD (p < 0.05). CONCLUSIONS: Considering preparation quality alone, GFPH was the best regimen, but SD provided the best balance between bowel preparation quality and patient acceptability.

Current Status and Future Perspectives of Artificial Intelligence in Magnetic Resonance Breast Imaging.

Recent advances in artificial intelligence (AI) and deep learning (DL) have impacted many scientific fields including biomedical maging. Magnetic resonance imaging (MRI) is a well-established method in breast imaging with several indications including screening, staging, and therapy monitoring. The rapid development and subsequent implementation of AI into clinical breast MRI has the potential to affect clinical decision-making, guide treatment selection, and improve patient outcomes. The goal of this review is to provide a comprehensive picture of the current status and future perspectives of AI in breast MRI. We will review DL applications and compare them to standard data-driven techniques. We will emphasize the important aspect of developing quantitative imaging biomarkers for precision medicine and the potential of breast MRI and DL in this context. Finally, we will discuss future challenges of DL applications for breast MRI and an AI-augmented clinical decision strategy.

Evaluation of computer-aided detection and diagnosis systems.

Computer-aided detection and diagnosis (CAD) systems are increasingly being used as an aid by clinicians for detection and interpretation of diseases. Computer-aided detection systems mark regions of an image that may reveal specific abnormalities and are used to alert clinicians to these regions during image interpretation. Computer-aided diagnosis systems provide an assessment of a disease using image-based information alone or in combination with other relevant diagnostic data and are used by clinicians as a decision support in developing their diagnoses. While CAD systems are commercially available, standardized approaches for evaluating and reporting their performance have not yet been fully formalized in the literature or in a standardization effort. This deficiency has led to difficulty in the comparison of CAD devices and in understanding how the reported performance might translate into clinical practice. To address these important issues, the American Association of Physicists in Medicine (AAPM) formed the Computer Aided Detection in Diagnostic Imaging Subcommittee (CADSC), in part, to develop recommendations on approaches for assessing CAD system performance. The purpose of this paper is to convey the opinions of the AAPM CADSC members and to stimulate the development of consensus approaches and "best practices" for evaluating CAD systems. Both the assessment of a standalone CAD system and the evaluation of the impact of CAD on end-users are discussed. It is hoped that awareness of these important evaluation elements and the CADSC recommendations will lead to further development of structured guidelines for CAD performance assessment. Proper assessment of CAD system performance is expected to increase the understanding of a CAD system's effectiveness and limitations, which is expected to stimulate further research and development efforts on CAD technologies, reduce problems due to improper use, and eventually improve the utility and efficacy of CAD in clinical practice.

A fully automatic algorithm for segmentation of the breasts in DCE-MR images.

Automatic segmentation of the breast and axillary region is an important preprocessing step for automatic lesion detection in breast MR and dynamic contrast-enhanced-MR studies. In this paper, we present a fully automatic procedure based on the detection of the upper border of the pectoral muscle. Compared with previous methods based on thresholding, this method is more robust to noise and field inhomogeneities. The method was quantitatively evaluated on 31 cases acquired from two centers by comparing the results with a manual segmentation. Results indicate good overall agreement within the reference segmentation (overlap=0.79 ± 0.09, recall=0.95 ± 0.02, precision=0.82 ± 0.1).