3DVIEWNIX-AVS: a software package for the separate visualization of arteries and veins in CE-MRA images.

Our earlier study developed a computerized method, based on fuzzy connected object delineation principles and algorithms, for artery and vein separation in contrast enhanced Magnetic Resonance Angiography (CE-MRA) images. This paper reports its current development-a software package-for routine clinical use. The software package, termed 3DVIEWNIX-AVS, consists of the following major operational parts: (1) converting data from DICOM3 to 3DVIEWNIX format, (2) previewing slices and creating VOI and MIP Shell, (3) segmenting vessel, (4) separating artery and vein, (5) shell rendering vascular structures and creating animations. This package has been applied to EPIX Medical Inc's CE-MRA data (AngioMark MS-325). One hundred and thirty-five original CE-MRA data sets (of 52 patients) from 6 hospitals have been processed. In all case studies, unified parameter settings produce correct artery-vein separation. The current package is running on a Pentium PC under Linux and the total computation time per study is about 3 min. The strengths of this software package are (1) minimal user interaction, (2) minimal anatomic knowledge requirements on human vascular system, (3) clinically required speed, (4) free entry to any operational stages, (5) reproducible, reliable, high quality of results, and (6) cost effective computer implementation. To date, it seems to be the only software package (using an image processing approach) available for artery and vein separation of the human vascular system for routine use in a clinical setting.

Performance evaluation of finite normal mixture model-based image segmentation techniques.

Finite Normal Mixture (FNM) model-based image segmentation techniques adopt the following detection-estimation-classification paradigm: 1) detect the number of image regions by using theoretical information criteria; 2) estimate model parameters by using expectation-maximization (EM)/classification-maximization (CM) algorithms; and 3) classify pixels into regions by using various classifiers. This paper presents a theoretical framework to evaluate the performance of this class of image segmentation techniques. For the detection performance, probabilities of over-detection and under-detection of the number of image regions are defined, and the associated formulae in terms of model parameters and image quality are derived. For the estimation performance, both EM and CM algorithms are showed to produce asymptotically unbiased ML estimates of model parameters in the case of no-overlap. Cramer-Rao bounds of variances of these estimates are derived. For the classification performance, misclassification probability for the Bayesian classifier is defined, and a simple formula based on parameter estimates and classified data is derived to evaluate segmentation errors. This evaluation method provides both theoretically approachable accuracy limits of the techniques and practically achievable performance of the given images. Theoretical and experimental results are in good agreement and indicate that, for images of moderate quality, the detection operation is robust, the parameter estimates are accurate, and the segmentation errors are small.