Purpose:Public datasets for training artificial intelligence (AI) models in breast cancer screening are limited in size and quality, making it difficult to develop reliable systems. We introduce OMAMA-DB, an extensive publicly available collection of two-dimensional (2D) mammograms and three-dimensional (3D) tomosynthesis volumes. Approach:Starting from 967,991 images, we created a curated set of 231,080 images using a multi-stage filtering process that removes missing labels, uncommon dimensions, rare scanner types, duplicate studies, and invalid DICOM files. All 2D images then undergo additional outlier detection using histogram filtering and a variational autoencoder to remove low-quality outliers. OMAMA-DB includes pathology-based cancer labels and automated lesion annotations generated using DeepSight. We also provide a web-based annotation tool for expert validation. To demonstrate usability, we fine-tuned MedGemma on a balanced subset of OMAMA-DB. We conducted a preliminary user study comparing human and automated classification of real and synthetic mammograms. Results:OMAMA-DB contains 231,080 images, including 7351 2D and 374 3D cancer cases. Fine-tuned MedGemma achieved 0.989 accuracy, 0.997 sensitivity, and an F 1 score of 0.989 on a balanced validation set of 2942 images. In real-versus-synthetic classification, humans achieved 0.485 accuracy, and logistic regression and convolutional neural network achieved 0.972 and 0.997, respectively. Conclusions:OMAMA-DB provides a large mammography dataset with pathology-based labels and automated lesion annotations to support medical imaging research. Fine-tuned foundation models demonstrate strong cancer classification performance, and the gap between human and automated detection of synthetic images highlights the importance of real clinical data. All data, models, and parameters are openly available for research use.
Background. Advancements in digital technologies have revolutionized surgical education. One such innovation is the remote mentor technology, which enables experienced surgeons to assist less experienced colleagues in real time using mixed reality and telecommunication tools. Aim. To develop a remote mentoring system and assess its usability during laparoscopic partial nephrectomy from both the trainee and the mentor perspectives. Materials and methods. Ten laparoscopic partial nephrectomies were performed using augmented reality and the Remote Mentor system. After each surgery, both mentors and trainees completed structured questionnaires assessing the quality of audiovisual exchange, interface intuitiveness, and communication completeness. The system used Hololens II and cloud-based software. Data were analyzed using non-parametric tests and correlation analysis. Results. The median age of trainees was 36.5 years; mentors – 60 years. All participants rated the system highly, with no negative responses recorded. Significant positive correlations were found between operator age/experience and satisfaction scores. The system demonstrated strong technical performance and high subjective efficacy. Conclusion. The remote mentoring system presents a promising tool for expanding access to surgical training. It showed high user satisfaction and technical feasibility. However, further validation with larger randomized studies is necessary.
Morbid obesity is a significant current medico-social problem, and bariatric surgery is a highly effective method for losing weight in individuals with severe obesity. Laparoscopic sleeve gastrectomy is the most commonly performed bariatric procedure. The most formidable complication of this operation is gastric leak. Our report demonstrates the diagnosis and management of early staple line leakage after laparoscopic sleeve gastrectomy. A 34-year-old female patient (BMI 40 kg/m2) underwent laparoscopic sleeve gastrectomy using a calibration bougie 36 F. The failure was suspected on the 2nd day after the operation, but the X-ray examination of the stomach failed to reveal a water-soluble contrast leak outside the gastric wall. The gastric leak was detected on the 3rd day after the procedure on abdominal CT-scan. The abscess was drained on re-laparoscopy. No closure of the insolvency zone and endoluminal stenting of the stomach were performed. The patient maintained fluid intake. On the 7th day after the re-laparoscopy, she was discharged from the hospital in a satisfactory condition with drainage installed in the abscess. On the follow-up examination in 2 weeks, the general condition was satisfactory, the patients got food following the dietary recommendations; fistulography showed a slight leakage of contrast material into the gastric remnant. After another 2 weeks, no contrast material in the gastric lumen was detected on fistulography. In 1 month, no defect of staple line was revealed on esophagogastroduodenoscopy, including insufflation. The used approach allowed us to eliminate the early staple line leakage after laparoscopic sleeve gastrectomy in a relatively short period.
Introduction: Data indicates that rural residents are at 30% higher risk of stroke. 2 Thirty percent of Wisconsin’s (WI) residents live in a rural area and are more often served by smaller, critical access hospitals (CAH). WI Coverdell Stroke Program (Coverdell) data entry into Get With The Guidelines-Stroke® (GWTG) for 2022 shows 22.3% of stroke patient’s transfer from a spoke, often CAH, to a larger tertiary or hub hospital. To assist Coverdell participating (CAH) in timely care of acute stroke patients, Coverdell, American Heart Association® (AHA), and 32 CAH’s partnered to form a Speed and Efficiency Taskforce (TF). Methods: An agreed upon charter, framework and data goals were established in May 2022. Metrics of EMS pre-notification, arrival to CT, and door to needle (DTN) times were regularly reviewed. Evaluation of patients arriving within 4.5 hours of last known well (LKW), median DTN, and percent DTN < 60 minutes were analyzed in two arrival windows. Hypothesized differences when comparing arrival windows of 0700-1900 and 1901-0659 were that the 1901-0659 window would have extended treatment times due to lack of 24/7 CT coverage and LKW not validated from others. Results: Data analysis for 2022. N=179 • Arrival to CT: 10 minutes faster in 1901-0659 window • Median DTN: 3 minutes faster in 0700-1900 window (71 vs 74 minutes respectively) • % DTN within 60 minutes: 12% higher compliance in 0700-1900 window Conclusions: Coverdell CAH’s are initiating the CT in a timely manner. Times are lower for arrivals between 1901-0659, ruling out delays due to CT tech non-availability. Delays in DTN treatment times within 60 minutes of arrival are more significant in the 1901-0659 arrival timeframe. The potential causes are a delay in telestroke consult with decision to treat, internal hospital delays in mixing and delivering thrombolytics, CT availability, and family/significant others present. Further research regarding arrival time affecting treatment need to be considered.
