Clinical Study to Assess the Safety and Performance of Kenko Breast AI+ Medical Device as an Aid to Breast Cancer Diagnosis in Adult Women

Overview

The goal of this observational study is to learn whether a new medical software tool called Kenko Breast AI+ can help healthcare professionals read breast magnetic resonance imaging (MRI) scans better to help diagnose breast cancer. The study includes adult women who may have breast cancer or who are being followed for the disease. The main questions it aims to answer is: * Does Kenko Breast AI+ help healthcare professionals find possible signs of breast cancer more accurately and efficiently when they read MRI scans? * Does using the software cause any safety problems or software-related issues during the study? Researchers will compare MRI readings done with the software to readings done without it. Participants will attend their regularly scheduled breast MRI appointment. All diagnoses are based on standard medical care, not on the study software.

This clinical investigation is a study designed to assess the safety, diagnostic performance, and usability of Kenko Breast AI+, a medical device software intended to support breast cancer diagnosis using breast MRI. The study is includes adult women who undergo breast MRI as part of routine clinical care for suspected breast cancer or oncological surveillance. The study does not alter standard clinical management and does not introduce additional diagnostic or therapeutic procedures. Independent radiologist readers assess MRI examinations either with or without assistance from Kenko Breast AI+. Each MRI case is evaluated once per reader during the first assessment phase and reassessed after a washout period using the alternate method. The sequence of assessments and reader allocation is randomized. Reader assessments are compared with a reference diagnosis established according to routine clinical practice. MRI datasets acquired during routine care are reviewed for technical quality, pseudonymized at the study site, and uploaded to both the investigational software platform and the site's picture archiving and communication system (PACS). Readers access the assigned cases through the appropriate platform according to randomization. After study completion, readers complete usability evaluations of the investigational device. The study is conducted in accordance with ISO 14155:2020, Good Clinical Practice, and applicable regulatory requirements (EU Medical Device Regulation 2017/745). Quality assurance measures include site qualification and initiation, staff training, predefined workflows, and documented delegation of responsibilities. Monitoring activities are performed by an independent sponsor-appointed monitor and include on-site and remote visits, verification of protocol compliance, review of safety reporting, and assessment of data accuracy. Study close-out procedures confirm completeness and proper archiving of study documentation. Audits and regulatory inspections may be conducted as required. Study data are recorded in validated electronic data capture systems using predefined case report forms. Automated and manual data checks are applied to ensure completeness, consistency, and plausibility. Discrepancies or missing values are resolved through a formal query process, and all data modifications are documented in audit trails. Source data verification is performed during monitoring visits by comparing electronic case report form entries with original source documents. A total of 200 participants are included to provide adequate precision for estimating diagnostic performance parameters. The sample size is based on assumptions of sensitivity and specificity greater than 85% and a predefined acceptable margin of error. The planned enrollment exceeds that of many published diagnostic imaging studies and is considered sufficient for the study objectives. MRI cases or lesions without an established reference diagnosis are excluded from primary diagnostic performance analyses. Reasons for missing data are documented and summarized. Descriptive comparisons are performed to assess potential bias. Invalid or technically inadequate MRI datasets are excluded according to predefined criteria. Analyses are conducted according to a predefined statistical analysis plan. Analysis populations include full analysis, per-protocol, and safety sets. Diagnostic performance measures are calculated and compared between assessments with and without the investigational device. Interpretation time and inter-reader agreement are evaluated using appropriate statistical methods. Usability outcomes are summarized descriptively. Data processing complies with applicable data protection regulations, including the EU General Data Protection Regulation 2016/679. The study is conducted in accordance with the Declaration of Helsinki and applicable regulatory and ethical requirements. Ethics committee approval is obtained prior to study initiation. As the study is observational and does not impact clinical decision-making, it is considered to pose minimal risk to participants.