Anant Madabhushi, PhD
Department of Biomedical Engineering
Rutgers Technology #: 09-078
Invention Summary:
Rutgers Investigators have developed a novel algorithm and methodology that enables rapid, quantitative analysis of biomarkers in tissue samples. The algorithm utilizes an iterative mean shift approach to capture snapshots at various levels of color resolution in the tissue sample as it approaches convergence (defined here to mean all points have reached their associated mode based off of the bandwidth parameter). The layers are then Normalized Cut, guided by a small swatch of user specified domain knowledge, and then mapped to a final segmented result. The procedure can be performed in less than one minute to obtain accur-ately segmented images of the stained regions allowing quantification and analysis to easily take place. By selecting representative points(pixels) from the class of interest in the tissue sample, the system can rapidly extract all similar values across many samples. The overall approach provides an objective segmentation that is user-independent and can analyze very large numbers of samples, reducing time, cost of diagnosis and user bias.
Market Applications: Diagnostics, Medical Imaging, Bioimaging, Drug Efficacy.
Advantages: Rapid, accurate detection and analysis of tissue samples; objective segmentation; easily scaled to large numbers of samples; automated (lower cost); demonstrated utility in ground truth samples.
Intellectual Property & Development Status: Patent pending.
PDF/
Posts
