syngo.via MI Workflows; Scenium; syngo MBF

The syngo.via MI Workflows is an advanced medical image processing software that integrates multi-modality data from PET, SPECT, CT, and MR scanners to provide quantitative analysis tools for oncology, cardiology, and neurology. It leverages deep learning for automated organ segmentation and offers specialized workflows, including tau pathology quantification, to enhance diagnostic capabilities and support clinicians in disease management and treatment planning.

The syngo.via MI Workflows, encompassing Scenium and syngo MBF applications, is a software-only medical device designed for multi-modality post-processing of medical images. It assists clinicians in managing diseases related to oncology, cardiology, neurology, and organ function by facilitating the viewing, manipulation, quantification, analysis, and comparison of functional (PET, NM) and anatomical (CT, MR) datasets across one or more time-points. Operating on a client/server architecture, the system integrates with standard IT hardware and supports DICOM data transfer, offering functionalities for both automated and manual data import. It provides tailored workflows such as syngo.MM Oncology and syngo.MI General, alongside applications like Scenium for neurology and syngo MBF for cardiology, aiming to optimize the workflow for radiologists and technologists. This iteration introduces significant enhancements, particularly in automated organ segmentation and tau pathology assessment. The MI General workflow now features deep-learning based organ segmentation to evaluate organ-specific tumor burden, validated using DICE coefficient and Average Symmetric Surface Distance (ASSD) metrics; segmentation accuracy can be influenced by image acquisition parameters, patient anatomy, and user expertise. The Scenium application incorporates a new Tau workflow to quantify tau protein distribution in the brain via PET imaging, employing methods analogous to existing amyloid PET quantification. This includes a dedicated tau brain template and novel regions for SUVR calculations based on the AAL atlas for Braak staging, which has shown good correlation with MR-based methods using over 700 flortaucipir images. Additional updates include support for CT large matrix data, iodine thresholding, and interactive spectral imaging within oncology workflows.

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 range of DICE coefficients and Average Symmetric Surface Distance (ASSD) values observed for the deep-learning based organ segmentation across different organs and imaging conditions, and how does this performance compare to other established segmentation algorithms?; How does the 'harmonization of SUV (PET) across different PET systems or different PET reconstruction methods' feature technically operate, and what is its validated impact on longitudinal study consistency and multi-center trial data?; For the Tau workflow in Scenium, what are the specific Braak regions defined using the AAL atlas, and how sensitive is the SUVR quantification to variations in PET acquisition protocols or patient-specific brain atrophy?; Could the 'Interactive Spectral Imaging' feature in the MM Onco workflow be detailed further, particularly its application in differentiating tissue types or assessing treatment response in oncological cases?; Given the listed factors affecting organ segmentation accuracy, what are the recommended best practices or specific image acquisition protocols to optimize the performance of the deep-learning segmentation algorithm in diverse clinical settings?