TumorSight Viz

TumorSight Viz is a cloud-based image processing system that visualizes and analyzes breast MRI studies in patients with biopsy-proven early-stage or locally advanced breast cancer. It reads DICOM MRI images, performs processing functions including image registration, subtractions, measurements, 3D renderings, and reformats, and displays results through a web application accessed via internet-connected devices. Patient management decisions should not be based solely on TumorSight Viz results.

Both the subject and predicate devices use the same technological principle of visualizing dynamic MRI images in DICOM format and perform identical core features: standard image viewing tools, parametric maps, kinetic curves, automatic volume segmentation, automatic linear measurements, and automated DICOM import. The key difference is that TumorSight Viz (v1.3) features an updated segmentation model compared to the predicate (v1.2), with the Communication & Storage feature reclassified and moved to the TumorSight Platform.

ISO 14971 for risk management was applied. The device was validated using mean absolute error analysis comparing automated measurements to radiologist ground truth, volumetric and surface Dice coefficients for segmentation assessment, and inter-radiologist variability comparisons.

TumorSight Viz demonstrates substantial equivalence through: (1) identical intended use and indications as the predicate for breast DCE-MRI visualization and analysis; (2) equivalent technological features and principle of operation based on dynamic MRI processing; (3) performance testing showing measurement errors (tumor volume 5.2±12.5 cc, longest dimension 1.32±1.65 cm, landmark distances 0.60-1.17 cm) comparable to inter-radiologist variability, indicating clinically acceptable accuracy; and (4) direct comparison to the predicate device across 192 cases showing equivalent performance on all measurements including tumor dimensions and anatomical distances.