RT-Mind-AI

RT-Mind-AI is a server-based software that employs deep learning algorithms to automatically segment organs-at-risk in non-contrast CT images for adult radiation oncology patients. Its purpose is to provide accurate anatomical delineations to support radiation treatment planning, evaluation, and adaptation, with its segmentation performance demonstrated to be non-inferior to established benchmarks.

RT-Mind-AI is a standalone medical imaging software designed for radiation oncology departments. Its primary function is to segment non-contrast CT images using deep learning algorithms to automatically contour organs-at-risk (OAR) in the Head and Neck, thorax, and abdominal/pelvic regions. This information is crucial for treatment planning, evaluation, and adaptation. The software supports both desktop and web-based automatic segmentation, with the latter allowing results to be sent to specified network nodes within a local area network. Additionally, RT-Mind-AI provides manual segmentation tools for adjusting the automated contours and includes features like preset ROIs, templates, DICOM data transmission, patient management, image review, and ROI management. Non-clinical testing demonstrated RT-Mind-AI's capability to contour a broader range of OARs than its predicate device, including specific structures such as the external auditory meatus, middle ear, spleen, intestinal tube, peritoneal cavity, femoral head/neck, humeral heads, and breasts. For organs segmented by both devices, RT-Mind-AI showed non-inferior segmentation accuracy, as measured by DICE similarity coefficients (DSC) using CT images from various US healthcare institutions and scanner models (GE, Siemens, Philips), with ground truth established by at least three licensed physicians. Similar non-inferiority was found for the additionally segmented OARs when their DSC values were compared to the device's performance on its other segmented organs. The software operates on Windows, is server-based, and is intended for use with adult patients (over 21 years of age) previously selected for radiation therapy.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. What is the specific list of all organs-at-risk (OARs) that RT-Mind-AI can automatically segment for each anatomical region (Head and Neck, Thorax, Abdominal & Pelvic), beyond the examples provided as 'additional'?; How does the deep learning model in RT-Mind-AI handle variations in CT image quality, noise levels, or artifacts from different scanner manufacturers (GE, Siemens, Philips) beyond the general non-inferiority finding?; What were the mean and standard deviation of the DICE Similarity Coefficients (DSC) for each of the additionally segmented organs, and how do these compare to established clinically acceptable DSC values for those specific structures?; Given its 'Adults Only (>21 years of age)' limitation, what specific anatomical, physiological, or dataset-related factors prevent its application in pediatric populations, and are there plans to validate it for younger patients?; What are the typical processing times for automatic segmentation on desktop versus web-based platforms for standard datasets of varying complexity (e.g., Head and Neck vs. Pelvic), and how does this impact clinical workflow efficiency?