Lead Inventor:
Andrew Laine
Medical Imaging Analysis Slowed by Manual Segmentation:
High resolution, real-time volumetric (3-D) medical imaging is an indispensible diagnostic tool. Image acquisition time is much shorter than analysis time, leading to a bottleneck in delivering a diagnostic or pre-surgical opinion. Precise delineation the image object boundaries is a prerequisite for analysis. For example, in cardiac imaging, the chambers of the heart and associated valves are outlined on the image; this facilitates interpretation by cardiologists. Manual tracing of hundreds of scans is tedious and impractical. Several automated or semi-automated segmentation methods have been proposed to address this problem, but trade flexibility for performance. As imaging technology evolves and the demand for real-time feedback grows, there will be a need for an analysis framework to perform real-time segmentation on a variety of data in a far more efficient manner.
3-D Medical Imaging Segmented in Real-Time for Faster Diagnoses:
The proposed analysis framework for real-time segmentation is termed Surface Function Actives (SFA). The technique uses a surface function to reduce the dimensionality of the image data (i.e. using a 2-D function to represent a 3-D surface in space). Motion estimation is performed by a specific ""optical flow"" algorithm, by looking at displacement of a specific pixel intensity pattern. This reduction in dimensionality enables rapid, accurate image segmentation for medical imaging applications. Studies have focused on ventricular border and echocardiology applications, but may be extended to any volumetric medical image.
Applications:
• Real-time, automated segmentation for image data, with particular use for volumetric images (3D) and motion (4D) data
• Analysis and processing of a variety of medical images
• Ventricular border segmentation from cardiac MRI images
• Real-time 3D (RT3D) echocardiography analysis for deformation studies
• Analysis of cardiac dynamics (e.g. component displacement and strain)
Advantages:
• Offers real-time segmentation capability, even in massive 4D datasets
• Increased efficiency by reduction of dimensionality
• Provides capability of novel applications involving real-time feedback
• Performance demonstrated on both synthetic and real patient data
Opportunities:
• Available for licensing and sponsored research support
Patent Status: Patent Pending