SPIE Medical Imaging 2010 SPIE vol. 7624 Filter learning and evaluation of the Computer Aided Visualization and Analysis (CAVA) paradigm for pulmonary nodules using the LIDC-IDRI database Rafael Wiemker ∗ a , Ekta Dharaiya b , Amnon Steinberg b , Thomas Bülow a , Axel Saalbach a , Torbjörn Vik a a Philips Research Europe – Hamburg & Aachen, Germany b Philips Healthcare CT – Cleveland, USA & Haifa, Israel ABSTRACT We present a simple rendering scheme for thoracic CT datasets which yields a color coding based on local differential geometry features rather than Hounsfield densities. The local curvatures are computed on several resolution scales and mapped onto different colors, thereby enhancing nodular and tubular structures. The rendering can be used as a navigation device to quickly access points of possible chest anomalies, in particular lung nodules and lymph nodes. The underlying principle is to use the nodule enhancing overview as a possible alternative to classical CAD approaches by avoiding explicit graphical markers. For performance evaluation we have used the LIDC-IDRI lung nodule data base. Our results indicate that the nodule-enhancing overview correlates well with the projection images produced from the IDRI expert annotations, and that we can use this measure to optimize the combination of differential geometry filters. Keywords: pulmonary nodules, computer aided detection, computer aided visualization, Hesse rendering, LIDC-IDRI database 1. COMPUTER AIDED VISUALIZATION AND ANALYSIS (CAVA) AS AN ALTERNATIVE PARADIGM Computer Aided Detection (CADe) [1][2] has sometimes been received with a certain guardedness in radiological practice because explicit graphical markers such as arrows or circles may be felt to clutter the original image and distract the reader. Therefore we would like to present an alternative concept for 3D imagery such as CT or MR: Computer Aided Visualization and Analysis (CAVA) [3][4]. As an example of this concept, a volume-rendering-like overview is offered to the radiological user in addition to the original image volume, without explicit graphical markers. Rather, the volume rendering enhances anomalous structures and color-codes them according to their conspicuity. Instead of providing a list of objects as in classical CADe approaches, the volume rendering can be used as a navigator such that a mouse-click on any point in the anomaly-enhancing rendering will mark the associated position in the original image volume for further inspection. In this approach (CAVA) no internal object lists or object segmentations are used, but all voxels are treated on equal footing. As a navigator, we propose a freely rotatable maximum intensity projection of various multiscale differential geometry features (eigenvalues of structure and Hesse matrix, shape index, etc.) which enhance pulmonary nodules in thoracic CT datasets (Fig.1 and 10). In the projection, nodular and tubular structures (tumors and vessels) can be intuitively differentiated by their color so that nodules and lymph nodes stand out from other anatomical structures (section 2). ∗ Rafael.Wiemker@philips.com