SIS System

The SIS System is an advanced software platform employing machine learning and image processing, including deep learning neural networks trained on high-resolution 7T MRI data, to generate patient-specific 3D models of deep brain structures like the STN and GPi/GPe from standard clinical MR images. Its purpose is to enhance visualization and localization for neurological procedures such as Deep Brain Stimulation, integrating pre-operative planning with post-operative assessment by identifying implanted leads on CT scans relative to the modeled anatomy.

The SIS System by Surgical Information Sciences is a software-based medical image management and processing system designed for viewing, presenting, and documenting medical imaging. Its core functionality involves advanced image processing, image fusion, and intraoperative functional planning, providing 3D outputs compatible with stereotactic image-guided surgery systems. The system utilizes machine learning, specifically pre-trained deep learning neural networks, to enhance standard clinical MR images and generate patient-specific 3D anatomical models of critical brain structures such as the subthalamic nuclei (STN) and globus pallidus (GPe/GPi). This aids clinicians, including surgeons, neurologists, and radiologists, in visualizing and locating these structures for neurological procedures, particularly for planning Deep Brain Stimulation (DBS) lead placement, by supplementing standard clinical methods. This version of the SIS System builds upon a prior iteration (K223032) with key modifications: the addition of two compatible leads, minor adjustments to the image registration algorithm, and a new feature allowing users to view post-operative 3D models in different coordinate systems. Comprehensive software verification and validation were conducted, including repeated tests for visualization accuracy of STN and GPi/GPe, MRI to CT registration, CT image processing for lead segmentation, electrode orientation for new and existing leads, and extended validation of the head pose for standardized views. The performance data confirmed that the updated SIS System is fully verified and validated, functions as intended, and maintains its safety and effectiveness for aiding in the visualization of brain structures and planning stereotactic surgical procedures.

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 system's performance in visualizing STN and GPi/GPe, particularly the segmentation accuracy, vary with different MRI sequence parameters or across images from MRI scanners of different field strengths (e.g., 1.5T vs 3T clinical scanners)?; What are the specific metrics used to validate the 'minor modification' to the image registration algorithm, and what was the quantitative improvement, if any, in MRI-CT registration accuracy?; Given the deep learning models are pre-trained using 7T MRI data for ground truth, what is the process or feasibility for site-specific adaptation or validation if discrepancies are noted with local clinical imaging protocols or patient populations?; For the two newly added compatible leads, what were the key challenges in validating their segmentation and electrode orientation, and how does their performance compare to previously supported leads?; How does the feature allowing users to view the post-operative 3D model in a different coordinate system enhance clinical utility or integrate with existing neuronavigation workflows beyond the standardized head position view?