BoneMRI v1.6

BoneMRI v1.6 is an image processing software designed to enhance bone visualization in MRI scans of the pelvic and lumbar spine regions using a convolutional neural network to generate images with CT-like bone contrast. The methodology involves analyzing 3D gradient echo MRIs with a dedicated protocol to reconstruct images that quantitatively assess bone morphology and radiodensity, offering an alternative but not a replacement for CT.

BoneMRI v1.6 is a standalone image processing software application developed by MRIguidance B.V. to enhance the visualization of bone structures within MRI scans. It processes 3D gradient echo MRI acquisitions, obtained using a specific protocol, to generate 3D tomographic images with radiodensity contrast, referred to as BoneMRI images. These images aim to provide improved differentiation between bone and surrounding soft tissues, assisting imaging experts such as radiologists and orthopedic surgeons. The system functions as a server application, compatible with DICOM standards for image input from and output to PACS, and allows viewing of the processed images on standard medical imaging software. The software's algorithm leverages a convolutional neural network trained on diverse datasets to detect bone features from MRIs, assigning Hounsfield Unit-like values based on signal intensity and contextual information. The software is indicated for use in the pelvic region (including sacrum, hip bones, femoral heads) and the lumbar spine region (L3 to S1 vertebrae) but is not intended for diagnosing or monitoring tumors, nor is it meant to replace CT imaging. Performance validation involved quantitative, voxel-by-voxel comparison of BoneMRI images against co-registered CT scans from 101 pelvic and 103 lumbar spine patient datasets. The results demonstrated clinically acceptable accuracy in reconstructing 3D bone morphology (mean absolute cortical delineation error below 1.0 mm), tissue radiodensity (mean deviation below 25 HU generally, and below 55 HU specifically for bone), and tissue radiodensity contrast (mean HU correlation coefficient above 0.75 for bone). This indicates BoneMRI's utility for qualitative and quantitative assessment of bone in the specified anatomical areas.

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 BoneMRI v1.6 compare to traditional MRI sequences specifically designed for bone visualization in terms of diagnostic accuracy for non-tumor related pathologies in the pelvic and lumbar regions?; What is the impact of varying MRI acquisition parameters, particularly those deviating from the 'dedicated MRI scan protocol', on the accuracy and reliability of BoneMRI-generated images?; Are there specific patient populations or subtle bone pathologies (e.g., early-stage avascular necrosis, stress fractures) where BoneMRI offers a more significant diagnostic advantage over standard MRI, or where its performance might be compromised?; Given that BoneMRI assigns Hounsfield Unit (HU) like values from MRI data, how do these values quantitatively correlate with actual bone mineral density measurements obtained through established methods like DXA or QCT for the indicated anatomical regions?; What are the specific computational requirements (e.g., processing time per scan, recommended GPU specifications) for the on-premise deployment of BoneMRI v1.6, and how might these influence workflow efficiency in diverse clinical settings?