EuroRV3: EuroVis Workshop on Reproducibility, Verification, and Validation in Visualization (2016) K. Lawonn, M. Hlawitschka, and P. Rosenthal (Editors) On the Evaluation of a Semi-Automatic Vortex Flow Classification in 4D PC-MRI Data of the Aorta M.Meuschke 1 , B. Köhler 1 , B. Preim 1 , K. Lawonn 2 1 Department of Simulation and Graphics, University of Magdeburg, Germany 2 Institute of Computational Visualistics, University of Koblenz - Landau, Germany Abstract In this paper, we report on our experiences that we made during our contributions in the field of the visualization of flow characteristics. Mainly, we focused on the vortex flow classification in 4D PC-MRI as current medical studies assume a strong correlation between cardiovascular diseases and blood flow patterns such as vortices. For further analysis, medical experts are asked to manually extract and classify such vortices according to specific properties. We presented and evaluated techniques that enable a fast and robust vortex classification [MLK * 16, MKP * 16] that supports medical experts. The main focus in this paper is a report that describes our conversations with the domain experts. The dialog was the fundament that gave us the direction of what the experts need. We derived several requirements that should be fulfilled by our tool. From this, we developed a prototype that supports the experts. Finally, we describe the evaluation of our framework and discuss currently limitations. Categories and Subject Descriptors (according to ACM CCS): I.4.9 [Computer Graphics]: Image Processing and Computer Vision—Applications 1. Introduction Cardiovascular diseases (CVDs) represent the world’s leading cause of death [MPN11]. Medical researchers are interested in bet- ter understanding the causes of their initiation and evolution. They report on the importance to consider many different variables to judge a patient’s situation. Besides the vessel morphology and wall motion, the initiation and evolution of CVDs depends strongly on the blood flow characteristics. The needed information about the patient-specific hemodynam- ics can be non-invasively acquired by four-dimensional phase- contrast magnetic resonance imaging (4D PC-MRI) [DBB * 15]. 4D PC-MRI data describe time-resolved, 3D blood flow information that represent one full heartbeat consisting of systole and diastole. A qualitative data analysis enables the visualization of vortices that are considered as an indicator of pathologies [HHM * 10, HWS * 12, MBS * 15]. To investigate the influence of vortices on CVDs, med- ical studies with homogeneous patient collectives are performed. The vortex occurrences are counted and classified according to spe- cific characteristics [GMH * 12, HMW * 07]. Healthy persons exhibit only a slight systolic helix in the aortic arch, which is considered as physiological [KYM * 93]. CVDs lead to altered vessel geome- tries that increase the probability of emerging vortex flow patterns. For example, patients with a bicuspid aortic valve (AV), where two of the three leaflets are fused, show strong correlations to helical flow in the ascending aorta during systole [HHM * 10, MBS * 15]. In addition, vortex flow close to the vessel wall is associated with higher shear forces [GBvO * 15, vOPP * 16], that increase the risk of aneurysm development [BSK * 14]. Further understanding this mutual influence of hemodynamics and vessel morphology can support treatment decision-making and the corresponding risk as- sessment. However, the classification of vortices is a challenging and unstandardized process. Therefore, we present an Aortic Vor- tex Classification (AVOCLA) that allows to classify vortices in the human aorta semi-automatically. Our approach was developed in collaboration with two domain experts: a radiologist specialized in cardiac imaging with four years of work experience and an expert specialized in the visualization of 4D PC-MRI data with three years of work experience. 2. From Manuel to Semi-Automatic Vortex Classification The following description is based on [MKP * 16]. Manuel Classification. For developing an improved vortex clas- sification, we had to analyze the previous classification approach. Our medical expert stated that the vortex analysis is manually per- formed until now. Therefore, blood flow-representing path lines are depicted within the vessel surface and common flow visualization techniques such as illumination and particle animations are pro- vided. Moreover, we investigated and discussed widely used clas- sification characteristics and prepared the following list: • The shape refers to the extent of a vortex. Elongated vortices are c  2016 The Author(s) Eurographics Proceedings c  2016 The Eurographics Association. DOI: 10.2312/eurorv3.20161114