Multi-center Study Collecting Ultrasound Images of Patients Suspected of Deep Vein Thrombosis for AI Model Training

What this Trial Is About

Researchers are collecting and creating a labeled ultrasound image data set from patients undergoing routine ultrasound scans of their lower limbs due to suspected deep vein thrombosis DVT. This observational study aims to gather imaging data, patient information, and ultrasound metadata to train artificial intelligence AI models that can automatically detect DVT from conventional ultrasound scans. The study is part of the ThrombUS project, which focuses on developing tools for early and automated DVT diagnosis to improve patient safety and outcomes. The study involves selecting patients suspected of having DVT who are referred for ultrasound scans. These patients, both inpatients and outpatients, will be consecutively included to ensure diversity of demographics, medical conditions, and ultrasound operators. The collected data will include ultrasound image series and video clips, along with anonymized patient demographics, referral notes, medical conditions, scan diagnoses, operator IDs, and ultrasound equipment details. The anonymized data set will be prepared according to regulations for research use and shared publicly through European science portals. Participants will undergo routine ultrasound scans for suspected DVT, and their scan data and related information will be collected and anonymized. The primary outcome is the ultrasound data collected on the day of the scan. The data set will support AI model training and research, with the anonymized information made available for further study. Participation involves providing informed consent and allowing the inclusion of their scan data and metadata. The study will monitor data quality and maintain privacy throughout, with no interventional treatments involved.

A Multi-center Cohort Study for Conventional Ultrasound Image Set Collection to Create a Data Set for Research Purposes.