VisAble.IO

VisAble.IO is an advanced image processing software package designed to enhance liver ablation procedures by utilizing CT and MR imaging for improved planning and confirmation. Its methodology integrates AI-driven segmentation, multi-modal image registration, and quantitative analysis tools to assist physicians in accurately targeting lesions, placing virtual needles, and evaluating ablation zone coverage, with performance validated through rigorous testing.

VisAble.IO version 1.4, developed by TechsoMed Medical Technologies, is a stand-alone medical image processing software designed to support physicians during liver ablation procedures. Its primary purpose is to assist in both the planning phase, by facilitating the identification of ablation targets and enabling virtual ablation needle placement, and in the post-procedure confirmation phase, by helping to confirm the extent of ablation zones. The software processes Computed Tomography (CT) and Magnetic Resonance (MR) images imported from scanners or PACS systems. It operates through a user-controlled interface and follows a five-step workflow: data import, segmentation of anatomical structures (liver, vessels, target), instrument placement planning, ablation zone segmentation, and treatment confirmation, which includes registration of pre- and post-interventional images and quantitative analysis. The underlying science of VisAble.IO involves advanced image processing algorithms for segmentation, image registration, and quantification. Notably, it employs AI-based algorithms for liver segmentation from both CT and MR images, and for liver vessel segmentation from CT images. For treatment planning, it supports specific ablation needle models, while its confirmation tools are compatible with all ablation devices. Performance data from bench testing demonstrated its effectiveness, with segmentation accuracy (measured by DICE coefficient) exceeding predefined goals for liver, ablation target, and ablation zone segmentation on both CT and MR modalities. Image registration accuracy (measured by Mean Corresponding Distance) also surpassed performance targets for various pre- and post-ablation image combinations, including CT-to-CT and MR-to-CT. The software computes achieved margins and missed volumes to aid in assessing treatment coverage, though it's emphasized that final clinical accuracy of segmentations and registrations relies on user assessment and potential manual adjustments.

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-driven segmentation algorithms (e.g., for liver and ablation targets) vary across different MR sequences or contrast phases not explicitly detailed in the training datasets?; Given that user assessment is crucial for final segmentation and registration accuracy, what specific tools or visual aids does VisAble.IO offer to help users identify and correct potential algorithmic inaccuracies effectively?; What is the typical processing time required for each of the five distinct workflow steps, particularly for complex cases or large image datasets, and how might this impact intraoperative use?; The document mentions compatibility with specific needle models for planning. How readily can the system be updated or customized to incorporate new or different ablation needle models as they become available?; Beyond DICE scores and Mean Corresponding Distance, what other metrics or methods were used during validation to assess the clinical relevance and utility of the computed achieved margins and missed volumes in guiding treatment decisions?