Turn abandoned patents
into your own filing
into your own filing
This invention describes a way to automatically divide a digital medical image into distinct regions. It first groups similar pixels using a standard K-means clustering technique. Then, it refines these groupings by considering the relationships between neighboring pixels, which helps to create smoother and more consistent segmented areas, especially in images that are not very clear. The claims specify using K-means for the initial pixel labeling and then an energy function to refine the segmentation.
Why it matters: Filed before the widespread adoption of powerful GPUs, the computational demands of iterative MRF models are now easily met. Its stated advantage for 'poor or limited data' remains relevant, offering an alternative or complementary approach to data-hungry deep learning methods that dominate medical image segmentation today.
AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.
Reinvent this with AI
