Deep learning performance for detection and classification of microcalcifications on mammography.

BACKGROUND: Breast cancer screening through mammography is crucial for early detection, yet the demand for mammography services surpasses the capacity of radiologists. Artificial intelligence (AI) can assist in evaluating microcalcifications on mammography. We developed and tested an AI model for localizing and characterizing microcalcifications. METHODS: Three expert radiologists annotated a dataset of mammograms using histology-based ground truth. The dataset was partitioned for training, validation, and testing. Three neural networks (AlexNet, ResNet18, and ResNet34) were trained and evaluated using specific metrics including receiver operating characteristics area under the curve (AUC), sensitivity, and specificity. The reported metrics were computed on the test set (10% of the whole dataset). RESULTS: The dataset included 1,000 patients aged 21-73 years and 1,986 mammograms (180 density A, 220 density B, 380 density C, and 220 density D), with 389 malignant and 611 benign groups of microcalcifications. AlexNet achieved the best performance with 0.98 sensitivity, 0.89 specificity of, and 0.98 AUC for microcalcifications detection and 0.85 sensitivity, 0.89 specificity, and 0.94 AUC of for microcalcifications classification. For microcalcifications detection, ResNet18 and ResNet34 achieved 0.96 and 0.97 sensitivity, 0.91 and 0.90 specificity and 0.98 and 0.98 AUC, retrospectively. For microcalcifications classification, ResNet18 and ResNet34 exhibited 0.75 and 0.84 sensitivity, 0.85 and 0.84 specificity, and 0.88 and 0.92 AUC, respectively. CONCLUSIONS: The developed AI models accurately detect and characterize microcalcifications on mammography. RELEVANCE STATEMENT: AI-based systems have the potential to assist radiologists in interpreting microcalcifications on mammograms. The study highlights the importance of developing reliable deep learning models possibly applied to breast cancer screening. KEY POINTS: • A novel AI tool was developed and tested to aid radiologists in the interpretation of mammography by accurately detecting and characterizing microcalcifications. • Three neural networks (AlexNet, ResNet18, and ResNet34) were trained, validated, and tested using an annotated dataset of 1,000 patients and 1,986 mammograms. • The AI tool demonstrated high accuracy in detecting/localizing and characterizing microcalcifications on mammography, highlighting the potential of AI-based systems to assist radiologists in the interpretation of mammograms.

Inter-reader agreement of breast magnetic resonance imaging and contrast-enhanced mammography in breast cancer diagnosis: a multi-reader retrospective study.

OBJECTIVE: Breast magnetic resonance imaging (MRI) and contrast-enhanced mammography (CEM) are nowadays used in breast imaging but studies about their inter-reader agreement are lacking. Therefore, we compared the inter-reader agreement of CEM and MRI in breast cancer diagnosis in the same patients. METHODS: Breast MRI and CEM exams performed in a single center (09/2020-09/2021) for an IRB-approved study were retrospectively and independently evaluated by four radiologists of two different centers with different levels of experience who were blinded to the clinical and other imaging data. The reference standard was the histological diagnosis or at least 1-year negative imaging follow-up. Inter-reader agreement was examined using Cohen's and Fleiss' kappa (κ) statistics and compared with the Wald test. RESULTS: Of the 750 patients, 395 met inclusion criteria (44.5 ± 14 years old), with 752 breasts available for CEM and MRI. Overall agreement was moderate (κ = 0.60) for MRI and substantial (κ = 0.74) for CEM. For expert readers, the agreement was substantial (κ = 0.77) for MRI and almost perfect (κ = 0.82) for CEM; for non-expert readers was fair (κ = 0.39); and for MRI and moderate (κ = 0.57) for CEM. Pairwise agreement between expert readers and non-expert readers was moderate (κ = 0.50) for breast MRI and substantial (κ = 0.74) for CEM and it showed a statistically superior agreement of the expert over the non-expert readers only for MRI (p = 0.011) and not for CEM (p = 0.062). CONCLUSIONS: The agreement of CEM was superior to that of MRI (p = 0.012), including for both expert (p = 0.031) and non-expert readers (p = 0.005).

Advancements in Standardizing Radiological Reports: A Comprehensive Review.

Standardized radiological reports stimulate debate in the medical imaging field. This review paper explores the advantages and challenges of standardized reporting. Standardized reporting can offer improved clarity and efficiency of communication among radiologists and the multidisciplinary team. However, challenges include limited flexibility, initially increased time and effort, and potential user experience issues. The efforts toward standardization are examined, encompassing the establishment of reporting templates, use of common imaging lexicons, and integration of clinical decision support tools. Recent technological advancements, including multimedia-enhanced reporting and AI-driven solutions, are discussed for their potential to improve the standardization process. Organizations such as the ACR, ESUR, RSNA, and ESR have developed standardized reporting systems, templates, and platforms to promote uniformity and collaboration. However, challenges remain in terms of workflow adjustments, language and format variability, and the need for validation. The review concludes by presenting a set of ten essential rules for creating standardized radiology reports, emphasizing clarity, consistency, and adherence to structured formats.

Percutaneous Cholecystostomy to Manage a Hot Gallbladder: A Single Center Experience.

Objective A percutaneous cholecystostomy (PC) is a suitable option for treating acutely inflamed gallbladders. Its use has been postulated before for treating acute cholecystitis (AC), especially in elderly populations. The primary aim of our study is to analyze and present the positive results of PC as a bridge to laparoscopic cholecystectomy. Methods All patients who underwent PC at our hospital, Princess Royal University Hospital, King's College Hospital NHS Foundation Trust, London, GBR, from October 2020 were reviewed using a retrospective approach. Results Our study comprises 123 patients, with 72 females (58.5%) and 51 males (41.4%). In our study, many patients had significant comorbidities, and some of them were categorized as high-risk due to their frailty and medical conditions. The majority of the patients were in American Society of Anaesthesiologists' (ASA) groups II and III (45, 61), respectively. Though hospital stays can depend on variable factors, in our experience, the mean hospital length of stay was 12.7 days. In our study, 119 patients (96.8%) had the procedure through the interventional radiological approach, while only four patients had it through the laparoscopic approach. The transhepatic route for drainage was more commonly practiced at our center and was used in 108 patients. At the time of writing this article, 54 patients have already had a laparoscopic cholecystectomy (LC) done as an interval procedure after surpassing the acute attack of cholecystitis, while 42 patients are still awaiting their surgical procedure. Conclusion Our results show that PC is a viable option, especially in cases of AC that are not responding to conservative treatments. Our study has shown low complications and conversion rates after PC. We believe PC is a safe and effective tool for managing severe and refractory cases of AC.

How Radiomics Can Improve Breast Cancer Diagnosis and Treatment.

Recent technological advances in the field of artificial intelligence hold promise in addressing medical challenges in breast cancer care, such as early diagnosis, cancer subtype determination and molecular profiling, prediction of lymph node metastases, and prognostication of treatment response and probability of recurrence. Radiomics is a quantitative approach to medical imaging, which aims to enhance the existing data available to clinicians by means of advanced mathematical analysis using artificial intelligence. Various published studies from different fields in imaging have highlighted the potential of radiomics to enhance clinical decision making. In this review, we describe the evolution of AI in breast imaging and its frontiers, focusing on handcrafted and deep learning radiomics. We present a typical workflow of a radiomics analysis and a practical "how-to" guide. Finally, we summarize the methodology and implementation of radiomics in breast cancer, based on the most recent scientific literature to help researchers and clinicians gain fundamental knowledge of this emerging technology. Alongside this, we discuss the current limitations of radiomics and challenges of integration into clinical practice with conceptual consistency, data curation, technical reproducibility, adequate accuracy, and clinical translation. The incorporation of radiomics with clinical, histopathological, and genomic information will enable physicians to move forward to a higher level of personalized management of patients with breast cancer.

Disparities in Breast Cancer Diagnostics: How Radiologists Can Level the Inequalities.

Access to medical imaging is pivotal in healthcare, playing a crucial role in the prevention, diagnosis, and management of diseases. However, disparities persist in this scenario, disproportionately affecting marginalized communities, racial and ethnic minorities, and individuals facing linguistic or cultural barriers. This paper critically assesses methods to mitigate these disparities, with a focus on breast cancer screening. We underscore scientific mobility as a vital tool for radiologists to advocate for healthcare policy changes: it not only enhances diversity and cultural competence within the radiology community but also fosters international cooperation and knowledge exchange among healthcare institutions. Efforts to ensure cultural competency among radiologists are discussed, including ongoing cultural education, sensitivity training, and workforce diversification. These initiatives are key to improving patient communication and reducing healthcare disparities. This paper also highlights the crucial role of policy changes and legislation in promoting equal access to essential screening services like mammography. We explore the challenges and potential of teleradiology in improving access to medical imaging in remote and underserved areas. In the era of artificial intelligence, this paper emphasizes the necessity of validating its models across a spectrum of populations to prevent bias and achieve equitable healthcare outcomes. Finally, the importance of international collaboration is illustrated, showcasing its role in sharing insights and strategies to overcome global access barriers in medical imaging. Overall, this paper offers a comprehensive overview of the challenges related to disparities in medical imaging access and proposes actionable strategies to address these challenges, aiming for equitable healthcare delivery.

Comparison of Sensitivity and Specificity of Biparametric versus Multiparametric Prostate MRI in the Detection of Prostate Cancer in 431 Men with Elevated Prostate-Specific Antigen Levels.

(1) Background: the study of dynamic contrast enhancement (DCE) has a limited role in the detection of prostate cancer (PCa), and there is a growing interest in performing unenhanced biparametric prostate-MRI (bpMRI) instead of the conventional multiparametric-MRI (mpMRI). In this study, we aimed to retrospectively compare the performance of the mpMRI, which includes DCE study, and the unenhanced bpMRI, composed of only T2-weighted imaging and diffusion-weighted imaging (DWI), in PCa detection in men with elevated prostate-specific-antigen (PSA) levels. (2) Methods: a 1.5 T MRI, with an endorectal-coil, was performed on 431 men (aged 61.5 ± 8.3 years) with a PSA ≥4.0 ng/mL. The bpMRI and mpMRI tests were independently assessed in separate sessions by two readers with 5 (R1) and 3 (R2) years of experience. The histopathology or ≥2 years follow-up served as a reference standard. The sensitivity and specificity were calculated with their 95% CI, and McNemar's and Cohen's κ statistics were used. (3) Results: in 195/431 (45%) of histopathologically proven PCa cases, 62/195 (32%) were high-grade PCa (GS ≥ 7b) and 133/195 (68%) were low-grade PCa (GS ≤ 7a). The PCa could be excluded by histopathology in 58/431 (14%) and by follow-up in 178/431 (41%) of patients. For bpMRI, the sensitivity was 164/195 (84%, 95% CI: 79-89%) for R1 and 156/195 (80%, 95% CI: 74-86%) for R2; while specificity was 182/236 (77%, 95% CI: 72-82%) for R1 and 175/236 (74%, 95% CI: 68-80%) for R2. For mpMRI, sensitivity was 168/195 (86%, 95% CI: 81-91%) for R1 and 160/195 (82%, 95% CI: 77-87%) for R2; while specificity was 184/236 (78%, 95% CI: 73-83%) for R1 and 177/236 (75%, 95% CI: 69-81%) for R2. Interobserver agreement was substantial for both bpMRI (κ = 0.802) and mpMRI (κ = 0.787). (4) Conclusions: the diagnostic performance of bpMRI and mpMRI were similar, and no high-grade PCa was missed with bpMRI.

Myths and facts about artificial intelligence: why machine- and deep-learning will not replace interventional radiologists.

Artificial intelligence (AI) is revolutionizing healthcare and transforming the clinical practice of physicians across the world. Radiology has a strong affinity for machine learning and is at the forefront of the paradigm shift, as machines compete with humans for cognitive abilities. AI is a computer science simulation of the human mind that utilizes algorithms based on collective human knowledge and the best available evidence to process various forms of inputs and deliver desired outcomes, such as clinical diagnoses and optimal treatment options. Despite the overwhelmingly positive uptake of the technology, warnings have been published about the potential dangers of AI. Concerns have been expressed reflecting opinions that future medicine based on AI will render radiologists irrelevant. Thus, how much of this is based on reality? To answer these questions, it is important to examine the facts, clarify where AI really stands and why many of these speculations are untrue. We aim to debunk the 6 top myths regarding AI in the future of radiologists.

Double reading of diffusion-weighted magnetic resonance imaging for breast cancer detection.

PURPOSE: To estimate the performance of diffusion-weighted imaging (DWI) for breast cancer detection. METHODS: Consecutive breast magnetic resonance imaging examinations performed from January to September 2016 were retrospectively evaluated. Examinations performed before/after neoadjuvant therapy, lacking DWI sequences or reference standard were excluded; breasts after mastectomy were also excluded. Two experienced breast radiologists (R1, R2) independently evaluated only DWI. Final pathology or > 1-year follow-up served as reference standard. Mc Nemar, χ2, and κ statistics were applied. RESULTS: Of 1,131 examinations, 672 (59.4%) lacked DWI sequence, 41 (3.6%) had no reference standard, 30 (2.7%) were performed before/after neoadjuvant therapy, and 10 (0.9%) had undergone bilateral mastectomy. Thus, 378 women aged 49 ± 11 years (mean ± standard deviation) were included, 51 (13%) with unilateral mastectomy, totaling 705 breasts. Per-breast cancer prevalence was 96/705 (13.6%). Per-breast sensitivity was 83/96 (87%, 95% confidence interval 78-93%) for both R1 and R2, 89/96 (93%, 86-97%) for double reading (DR) (p = 0.031); per-lesion DR sensitivity for cancers ≤ 10 mm was 22/31 (71%, 52-86%). Per-breast specificity was 562/609 (93%, 90-94%) for R1, 538/609 (88%, 86-91%) for R2, and 526/609 (86%¸ 83-89%) for DR (p < 0.001). Inter-observer agreement was substantial (κ = 0.736). Acquisition time varied from 3:00 to 6:22 min:s. Per-patient median interpretation time was 46 s (R1) and 51 s (R2). CONCLUSIONS: DR DWI showed a 93% sensitivity and 88% specificity, with 71% sensitivity for cancers ≤ 10 mm, pointing out a potential for DWI as stand-alone screening method.

Recurrent giant juvenile fibroadenomas with hemihypertrophy.

Giant juvenile fibroadenomas in patients with hemihypertrophy are exceptionally rare. We present a very interesting case of a 13 year old girl with hemihypertrophy of the left side presenting with recurrent giant juvenile fibroadenomas of the left breast. The giant fibroadenomas occurred twice in the left breast over two years. The first had a diameter of 12 cm and was excised through an inframammary incision. The second occurred a year later, had a diameter of 11 cm, and was associated with three smaller fibroadenomas. These lesions were removed through a single periareolar incision. The procedures were complicated by keloid scarring but the results were improved with steroid impregnated tape dressing and local methylprednisolone injection. This report adds to our experience in managing patients with recurrent giant juvenile fibroadenomas complicated by hemihypertrophy and raises awareness to anticipate keloid scarring.