ART-Plan

ART-Plan v1.10.1 is a software designed to assist in radiation therapy planning by providing AI-driven tools for multi-modal medical image visualization, segmentation, and pseudo-CT generation. Its methodology relies on deep learning algorithms trained on diverse datasets to automatically delineate anatomical structures and synthesize CT images from MRI, with performance validated through rigorous non-clinical testing against established benchmarks.

ART-Plan v1.10.1 is an advanced medical image management and processing software designed for radiation treatment planning in cancer patients. It enables trained medical professionals to display, visualize, and process 3D multi-modal medical image data, including CT, PET-CT, CBCT, 4D-CT, and MR images. The system supports DICOM 3.0 compliant datasets, allowing users to import, define, transform, store, and export images, contours, and regions of interest. Key functionalities include AI-driven automatic contouring for organs at risk and healthy lymph nodes across various anatomical regions (head & neck, thorax/breast, abdomen, pelvis, brain) on both CT and MR images. It also offers semi-automatic and manual segmentation tools, alongside capabilities for rigid and deformable image registration of anatomical and functional images to facilitate multi-modal information assessment. The underlying science of ART-Plan v1.10.1 leverages deep learning algorithms for its automatic segmentation and for the generation of pseudo-CT images from MR scans (specifically for male pelvis and brain). This version, a modification of v1.10.0, introduces 48 new anatomical structures to its AI-based Annotate module, expanding its utility for protocols like SBRT for lung, and incorporates 8 bug fixes. The system's performance was validated through non-clinical testing against established criteria, including Dice Similarity Coefficient for segmentation accuracy, qualitative clinical evaluations, and gamma passing rates for synthetic CT fidelity. These tests confirmed that the modifications maintain the device's technical specifications and safety profile, demonstrating its ability to perform according to its design and meet clinical expectations for radiotherapy planning.

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. How does the performance of the AI-based contouring for the 48 newly added structures compare across different imaging modalities (CT vs. MR) and varying image qualities?; What specific deep learning architectures are employed for the auto-segmentation and pseudo-CT generation, and how were they optimized to ensure generalizability across diverse patient anatomies and imaging equipment?; Could the pseudo-CT generation feature be expanded to other anatomical regions beyond the male pelvis and brain, and what are the primary challenges in achieving this?; Are there ongoing or planned studies to evaluate the clinical impact of ART-Plan v1.10.1 on radiotherapy workflow efficiency and treatment plan quality compared to manual or other semi-automated methods?; Given the known limitations (e.g., pediatric patients, specific MRI sequences), what is the development roadmap for addressing these limitations and broadening the applicability of the automated features?