SIS Software version 3.3.0

SIS Software version 3.3.0 is an advanced medical imaging application designed for enhanced visualization of the subthalamic nuclei (STN) and intraoperative functional planning, utilizing machine learning algorithms to process and fuse pre-operative MRI and post-operative CT images. The software leverages a database of high-resolution 7T MRI to improve STN identification on clinical MRIs and provides tools for co-registration and segmentation, aiming to improve accuracy in stereotactic neurosurgical procedures.

SIS Software version 3.3.0 is a medical imaging application intended for viewing, presentation, and documentation of medical images, incorporating modules for image processing, image fusion, and intraoperative functional planning, with a specific capability to aid in the visualization of the subthalamic nuclei (STN). The software employs machine learning and image processing techniques to enhance standard clinical MRI scans (1.5T or 3T) by referencing a database of high-resolution 7T MRI brain images; this approach leverages the superior clarity of 7T MRI to identify structures like the STN on a patient's clinical scans. Version 3.3.0 introduces the ability to co-register post-operative CT scans with the patient's pre-operative MRI (already processed for STN visualization) and to segment relevant features in the CT image, thereby facilitating comprehensive visualization for neurosurgical procedures intended for use by medical professionals such as surgeons, neurologists, and radiologists. Performance validation demonstrated high accuracy for STN visualization, with 98.3% of center of mass distances and 100% of surface distances below 2.0mm, and a Dice coefficient of 0.69, significantly outperforming standard care. The co-registration feature achieved high precision, with 95% confidence that 90% of MR to CT registrations will have reference point errors below 0.454mm. Electrode segmentation in post-operative CT scans also showed strong results, with 95% confidence that 90% of segmentations will have center of mass distances below 0.491mm and orientation differences below 2.486 degrees. The updated anomaly detection model improved sensitivity to 50% while maintaining high specificity (89.39%), enhancing overall system success to 98.33%. The STN smoothing functionality was also validated as performing acceptably without negatively impacting device output.

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 accuracy of STN visualization by SIS Software compare to manual segmentation by experienced neuroradiologists in a real-world clinical setting, beyond the validation dataset?; What is the learning curve for clinicians (surgeons, neurologists, radiologists) to effectively utilize all functionalities of SIS Software 3.3.0, particularly the co-registration and CT segmentation features?; Are there specific patient populations or MRI/CT scanner characteristics for which the performance of SIS Software (STN visualization, co-registration, segmentation) is known to be suboptimal?; How does the improved anomaly detection in version 3.3.0, specifically the increased sensitivity, translate to tangible benefits in clinical decision-making or workflow efficiency?; Could the machine learning model be further trained or adapted to visualize other deep brain structures relevant to neurosurgical procedures beyond the STN?