QUANTIFYING ACTIVE CONTOUR MODEL (ACM) SEGMENTED REGION USING EDGE-LINKING AND REGION-FILLING ALGORITHM (ERA)

Abstract

Active contour model (ACM) algorithm is an effective and accurate method to segment out the regions of interest (ROIs) and has been widely used in many clinical imaging systems. In addition to segmenting out the lesion contour, how to accurately quantify the area/volume of the lesion is also important and challenging. Some quantification methods performed manually are time-consuming. To increase the accuracy and efficiency in quantifying the ACM segmented regions, a simple, straightforward, accurate and efficient automatic region quantification method with edge-linking and region-filling algorithm (ERA) for the ACM segmented regions was developed. Three types of computer simulated images, namely closed contour simulated images (CCSIs), discrete contour simulated images (DCSIs) and intact region simulated images (IRSIs) were created to evaluate the accuracy of the region-filling process, edge-linking process and ERA after ACM image segmentation process, respectively. Furthermore, the results of ERA were compared with those of the direct subtraction method (DSM). The ERA was also applied to quantify the area/volume of an irregularly-shaped lymphoma in a MR brain image. The results showed that the average percentage errors on the region-filling process, edge-linking process and ERA (0%, 0.03% and 7.52%, respectively) were significantly lower than those of DSM (15.85%, 15.08% and 20.36%, respectively) in quantifying the ROIs. The application of ERA to quantify the area/volume of an irregularly-shaped lymphoma in a MR brain image was successfully demonstrated. In conclusions, ERA outperforms the DSM in quantifying the area/volume of ACM segmented regions and might increase the efficiency and accuracy in clinical diagnosis and treatment.