Methods and Apparatuses for Detection of Abnormalities in Low-Contrast Images

US Patent Application (Pub. No. US 2018/0053297 A1) assigned to Ohio University. The invention provides a computer-implemented method for detecting abnormalities such as microcalcifications in low-contrast medical images (e.g., mammograms, ultrasound). The method: (i) determines regions of interest via image segmentation, and (ii) detects abnormalities using Gibbs random field clique pattern analysis with adaptive thresholding. The system supports network-based delivery — a server receives the image, runs inference, and returns the evaluation to a client device — enabling cloud-assisted clinical decision support. Inventors: Mehmet Celenk and Akshay S. Bharadwaj. Filed: August 11, 2017. Published: February 22, 2018.

Recommended citation: Celenk, M. & Bharadwaj, A. S. (2018). "Methods and Apparatuses for Detection of Abnormalities in Low-Contrast Images." US Patent Application Publication No. US 2018/0053297 A1. Filed Aug. 11, 2017.
Download Paper