Clinical MRI scans frequently exhibit anisotropic resolution, where in-plane resolution is high but the through-plane (slice thickness) is significantly lower. Achieving truly isotropic, high-resolution 3D imaging typically requires long acquisition times, which limit scanner throughput and reduce patient comfort.
This AI-driven reconstruction technology converts routine MRI scans into high-quality, isotropic 3D images, without increasing scan duration.
The approach integrates Generative AI with Self-Supervised Learning, enabling accurate reconstruction of missing spatial information while maintaining clinical reliability.
The Problem
Standard MRI protocols acquire data as a sequential stack of 2D slices. This creates:
- Thick slices with poor resolution in the Z-axis.
- Blurring and anatomical inconsistencies in 3D reconstructions.
- Reduced diagnostic performance, especially for small or subtle findings.
- Operational inefficiencies, since high-resolution protocols prolong scan time.
- Business Impact: Imaging centers must compromise between speed and quality. Longer scans reduce daily patient volume and limit return on investment for MRI equipment.
The Technology
The technology introduces an AI-based reconstruction pipeline combining:
- Generative AI: Learns anatomical structures and synthesizes missing high-resolution details in the low-quality dimension. For non-engineers: It is similar to restoring a blurred image based on learned patterns of how real anatomical structures are expected to appear.
- Self-Supervised Learning: Trains directly on raw MRI data without requiring manual annotations. Why it matters: MRI datasets are abundant but difficult and costly to label. Self-supervision enables scalable training and strong generalization across scanners, anatomies, and imaging conditions.
Advantages
- High-Quality 3D at Standard Speed: Delivers isotropic, high-resolution volumes from standard 2D scans without prolonging acquisition time.
- Enhanced Clinical Fidelity: Improves visualization of fine details and anatomical boundaries for better diagnosis and planning, using physics-guided constraints to ensure reliability.
- Operational Efficiency: Increases patient throughput and scanner ROI by eliminating the need for time-consuming high-resolution protocols.
- Seamless Integration: A scalable, vendor-agnostic software solution that requires no hardware modifications.
Applications
- Medical imaging manufacturers integrating advanced reconstruction tools.
- Hospitals and imaging centers seeking improved quality without longer scan times.
- Radiology AI software platforms enhancing diagnostic and 3D modeling workflows.
- Surgical planning systems requiring high-fidelity isotropic datasets.
- Research institutions performing large-scale MRI studies or multi-site collaborations.
Business Development Contacts
