Summary of the technology
methods to provide automated segmentation of lesions and tumors, with minimal user interaction; the computerized algorithms will enable a reproducible measure, which is one of the most critical criteria for measurement; in addition, automated tracking over time will be developed. Given a new case, automated tools will enable the extraction of a previous case of the patient, based on which the current case can be analyzed more in context, and the abnormalities can be more accurately detected, segmented and measured.
Project ID : 8-2012-310
Description of the technology
The invention is a diagnostic aid for physicians to detect pathology in medical images. Image features are represented as "visual words" and can then be classified and retrieved according to relevance for quick reference by the physician. The tool uses a stored dictionary of features to differentiate healthy from pathological tissues.
The rapid growth of computerized medical imagery using picture archiving and communication systems (PACS) in hospitals throughout the world has generated a critical need for efficient and powerful search engines to classify and search the visual data. In addition, the growing workload on radiologists in recent years increases the need for computerized systems which could help the radiologist in prioritization and in the diagnosis of findings. The system was tuned to achieve high accuracy in diagnosis, with an average of over 90% correct classification on a publicly available database of 12,000 medical radiographs. In the ImageClef 2008 medical annotation challenge it ranked second. It is a highly efficient, with less than 200 milliseconds training and classification time per image.
In addition to being a diagnosis aid, the system is highly efficient as an image retrieval system based on desired criteria. In the ImageClef 2008 image retrieval challenge it ranked first among all algorithms tested. Although the system has been applied to date on x-ray images, it can easily be adapted to other types of images such as ultrasound, MRI and CT.
US Patent pending
Director, Business Development for high-tech & IT
T.A.U Tel Aviv University, Engineering