Background and Objectives: Breast cancer is a leading global health challenge, where early detection is essential for improving survival outcomes. Two-dimensional (2D) mammography is the established standard for breast cancer screening; however, its diagnostic accuracy is limited by factors such as breast density and inter-reader variability. Recent advances in artificial intelligence (AI) have shown promise in enhancing radiological interpretation. This study aimed to assess the utility of AI in improving lesion detection and classification in 2D mammography. Materials and Methods: A retrospective analysis was performed on a dataset of 578 mammographic images obtained from a single radiology center. The dataset consisted of 36% pathologic and 64% normal cases, and was partitioned into training (403 images), validation (87 images), and test (88 images) sets. Image preprocessing involved grayscale conversion, contrast-limited adaptive histogram equalization (CLAHE), noise reduction, and sharpening. A convolutional neural network (CNN) model was developed using transfer learning with ResNet50. Model performance was evaluated using sensitivity, specificity, accuracy, and area under the receiver operating characteristic (AUC-ROC) curve. Results: The AI model achieved an overall classification accuracy of 88.5% and an AUC-ROC of 0.93, demonstrating strong discriminative capability between normal and pathologic cases. Notably, the model exhibited a high specificity of 92.7%, contributing to a reduction in false positives and improved screening efficiency. Conclusions: AI-assisted 2D mammography holds potential to enhance breast cancer detection by improving lesion classification and reducing false-positive findings. Although the model achieved high specificity, further optimization is required to minimize false negatives. Future efforts should aim to improve model sensitivity, incorporate multimodal imaging techniques, and validate results across larger, multicenter prospective cohorts to ensure effective integration into clinical radiology workflows.

Background/Objectives: Postpartum depression and anxiety affect up to 20% of women worldwide, yet remain understudied in Eastern Europe. Romania, one of Europe’s most religious countries, provides a unique context for examining how coping strategies and religiosity influence perinatal mental health. This cross-sectional study characterized coping styles, depressive and anxiety symptoms, and religiosity among postpartum women from two western Romanian counties (Bihor and Timiș) and examined associations between coping dimensions and psychological outcomes. Methods: A cross-sectional study was conducted among 201 postpartum women recruited from two public maternity hospitals between 2024 and 2025. Sociodemographic, obstetric, neonatal, coping (COPE Inventory), depressive (EPDS; PHQ-9), anxiety (GAD-7), and religiosity (RCI-10) data were collected through questionnaires and medical record extraction. Associations between coping dimensions and psychological symptoms were examined using Pearson correlations. Analyses were performed in RStudio. Results: Prevalence of possible depression (EPDS ≥ 10) was 32.8% overall, with no regional difference (Bihor 32.0% vs. Timiș 33.7%, p = 0.920). EPDS demonstrated strong convergent validity with PHQ-9 (r = 0.58, p < 0.001) and GAD-7 (r = 0.61, p < 0.001). In bivariate analyses, avoidant coping showed the strongest association with depressive symptoms (r = 0.28, p < 0.001), particularly in Bihor (r = 0.35, p < 0.001). Multiple regression analysis (R² = 0.196, p < 0.001) identified avoidant coping as the strongest independent predictor (β = 2.82, 95% CI [1.16, 4.48], p < 0.001), followed by social support coping (β = 2.46, 95% CI [1.09, 3.83], p < 0.001). Emotion-focused coping showed an unexpected protective effect (β = −2.97, p = 0.004). Problem-focused coping and religiosity were not significant predictors. Critically, county was non-significant after controlling for coping strategies (p = 0.732), indicating regional differences are mediated by coping patterns rather than geographic location. Conclusions: Postpartum depression prevalence in Romania aligns with international estimates. Avoidant coping emerged as the primary modifiable risk factor. Findings support integrating coping assessment into postpartum screening and developing interventions targeting avoidant strategies in Romanian perinatal care.

Intrauterine contraception has changed over the past decade through longer validated duration, lower-dose hormonal devices, redesigned copper devices, new indications, and better safety surveillance. This targeted scoping review (OSF registration 10.17605/OSF.IO/3GUQF) mapped English-language primary human studies published from 1 January 2016 to 5 September 2026 in journals indexed in the Science Citation Index Expanded (SCIE). A reproducible search targeted to the major device programs, randomized trials, and large surveillance cohorts identified 766 PubMed/MEDLINE records, with no additional unique records from Scopus or Web of Science; of the 624 unique records screened, 123 reports were assessed, and 114, from 66 distinct trials, cohorts, or data sources, were included and appraised with design-specific Joanna Briggs Institute (JBI) tools. Two single-arm programs and a 7-year randomized trial supported the efficacy of 52 mg levonorgestrel intrauterine devices (LNG-IUDs) through 8 years. Lower-dose LNG-IUDs eased placement but were associated with higher ectopic pregnancy rates in four Scandinavian analyses and with less antidepressant use than the 52 mg device. Smaller or size-matched copper devices improved continuation and reduced expulsion in randomized comparisons, whereas a low-copper nitinol device had a higher 5-year pregnancy rate than expected with TCu380A. Perforation was uncommon but more frequent after postpartum insertion and during breastfeeding; immediate postpartum insertion carried the highest expulsion risk. One randomized trial and its secondary analyses supported the 52 mg LNG-IUD for emergency contraception. Therapeutic evidence was strongest for heavy menstrual bleeding and endometrial hyperplasia, whereas intrauterine ball and precious metal designs showed no consistent benefit. Progress refined rather than replaced the T-shaped IUD.

Background: Deep-learning artificial intelligence (AI) for mammographic screening moved from proof of concept to randomised evaluation in a single decade. We reviewed and meta-analysed its diagnostic accuracy and its effect on screening programmes, covering the literature published between 2015 and 2025. Methods: PubMed and Europe PMC were searched from 1 January 2015 to 31 December 2025, supplemented by ClinicalTrials.gov and by forward and backward citation searching (PRISMA 2020, PRISMA-DTA, PRISMA-S). Eligible studies evaluated a deep-learning system for cancer detection or triage in a screening population against a histopathological reference standard. Four syntheses were performed: standalone accuracy pooled on the logit-AUC scale (A), a bivariate sensitivity–specificity model (A2), cancer detection rate ratio for AI-integrated versus standard reading (B), and recall rate ratio (C). Random-effects models used restricted maximum likelihood with Knapp–Hartung intervals. Risk of bias was assessed with QUADAS-2 and QUADAS-C, and certainty with GRADE. Results: Twenty-six studies (27 reports, 2019–2025) were included. Pooled standalone AUC across 14 studies and 1,214,885 examinations was 0.890 (95% CI 0.858–0.915), with I² = 97.4% and a 95% prediction interval of 0.731–0.960. Neither publication year (p = 0.79) nor enriched versus consecutive sampling (p = 0.94) explained this dispersion in meta-regression. The bivariate model (k = 9) gave a summary sensitivity of 73.3% (64.5–80.6) at a specificity of 92.4% (86.8–95.8). Across randomised and paired prospective trials (k = 3), the pooled detection rate ratio was 1.13 (0.83–1.55), with the MASAI randomised trial alone reporting 1.29 (1.09–1.51) and a 44% reduction in screen reading. Non-randomised implementation studies (k = 5), which include a 463,094-women German programme evaluation, pooled to 1.22 (1.08–1.37) with little dispersion (I² = 19.0%). Recall changed little overall (0.95, 0.81–1.12). Certainty was very low for accuracy outcomes and moderate for the single randomised trial. Conclusions: A decade of evidence supports AI as a second reader and triage tool in organised screening, not as an autonomous replacement for the radiologist. Pooled accuracy is high on average but so dispersed that it cannot be transferred to a new programme; local validation before deployment remains necessary. The larger and more precise detection gains come from non-randomised designs, which is the pattern confounding would produce, so the randomised evidence remains the anchor. Interval-cancer and mortality endpoints are still awaited.