Detection of Immune Reactions in Renal Transplants Based on Ultrasonic Image Analysis

Abstract

Transplantation is the preferred therapy for terminal renal insufficiency. In case a transplant is subjected to immunological rejection reactions, timely detection is essential in order to initiate appropriate therapeutic measures. Sonographic investigation, which usually consists of visual examination of the ultrasound scan of a transplant, offers a non-invasive method of detection. To enhance the reliability of the diagnosis, it is expedient to develop methods to quantify the image properties for classifying imaged tissues and tissue changes as a basis for detection of immune reactions. In this paper, a classification method is described in which a vector of parameters characterizing image features of the ultrasound scan of the transplant is processed to obtain a diagnosis. The parametrized image features are chosen first on the basis of the criteria employed by the expert physician for diagnosis and secondly by correlating various parameters with the result of diagnosis by other (invasive) means. The method could be used to classify 83% of the scans correctly.