BoneMRI v1.4

BoneMRI v1.4 is an image processing software that employs a convolutional neural network to analyze dedicated 3D gradient echo MRI scans, generating images with enhanced bone contrast for the pelvic and lumbar spine regions. The system reconstructs 3D tomographic images with Hounsfield Unit-like values, demonstrating quantitative accuracy in bone morphology, radiodensity, and contrast when compared to CT scans.

BoneMRI v1.4 is a standalone image processing software application designed to enhance the visualization of bone structures in MRI images by improving contrast with respect to surrounding soft tissue. It analyzes 3D gradient echo MRI scans acquired using a dedicated protocol, employing a convolutional neural network-based algorithm to construct 3D tomographic radiodensity contrast images, referred to as BoneMRI images. The algorithm assigns Hounsfield Unit (HU) like values based on intensity and contextual information from the MRI. The software is intended for use by imaging experts, such as radiologists or orthopedic surgeons, for the pelvic region (sacrum, hip bones, femoral heads) and the lumbar spine region (L3 to S1 vertebrae). It is important to note that BoneMRI images are not intended for diagnosing or monitoring tumors, nor are they meant to replace CT images. Performance validation for the lumbar spine region involved a quantitative voxel-by-voxel analysis using imaging data from 73 patients, comparing BoneMRI images to co-registered CT scans as the reference standard. The study demonstrated clinically acceptable accuracy, with a mean absolute cortical delineation error below 1.0 mm for 3D bone morphology. Tissue radiodensity reconstruction showed a mean deviation below 25 HU on average and below 55 HU specifically for bone. Furthermore, the tissue radiodensity contrast achieved a mean HU correlation coefficient above 0.75 for bone. These results indicate that BoneMRI v1.4 accurately reconstructs bone morphology, radiodensity, and radiodensity contrast, making it a useful tool for qualitative and quantitative assessment of the specified anatomical regions.

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 BoneMRI v1.4 in depicting complex fracture patterns or subtle bone lesions compare to standard CT scans in the specified anatomical regions?; What are the specific input parameters from the 3D gradient echo MRI scans that most significantly impact the quality and accuracy of the generated BoneMRI images?; How does the algorithm account for variations in patient anatomy, bone density (e.g., osteoporotic bone), or the presence of metallic implants within the pelvic and lumbar regions?; What is the processing time required to generate BoneMRI images from a standard MRI dataset, and how does this integrate into typical clinical workflows?; Are there plans to expand the validated anatomical regions for BoneMRI beyond the pelvis and lumbar spine, or to adapt its use for pediatric populations?