Int J CARS DOI 10.1007/s11548-010-0497-5 ORIGINAL ARTICLE Liver tumors segmentation from CTA images using voxels classification and affinity constraint propagation Moti Freiman · Ofir Cooper · Dani Lischinski · Leo Joskowicz Received: 10 January 2010 / Accepted: 28 May 2010 © CARS 2010 Abstract Objective We present a method and a validation study for the nearly automatic segmentation of liver tumors in CTA scans. Materials and methods Our method inputs a liver CTA scan and a small number of user-defined seeds. It first classifies the liver voxels into tumor and healthy tissue classes with an SVM classification engine from which a new set of high- quality seeds is generated. Next, an energy function describ- ing the propagation of these seeds is defined over the 3D image. The functional consists of a set of linear equations that are optimized with the conjugate gradients method. The result is a continuous segmentation map that is thresholded to obtain a binary segmentation. Results A retrospective study on a validated clinical dataset consisting of 20 tumors from nine patients’ CTA scans from the MICCAI’08 3D Liver Tumors Segmentation Challenge Workshop yielded an average aggregate score of 67, an aver- age symmetric surface distance of 1.76 mm (SD = 0.61 mm) which is better than the 2.0 mm of other methods on the same database, and a comparable volumetric overlap error (33.8 vs. 32.6%). The advantage of our method is that it requires less user interaction compared to other methods. Moti Freiman and Ofir Cooper are equally contributed. M. Freiman (B ) · O. Cooper · D. Lischinski · L. Joskowicz School of Engineering and Computer Science, The Hebrew University of Jerusalem, Jerusalem, Israel e-mail: freiman@cs.huji.ac.il O. Cooper e-mail: coopeo@cs.huji.ac.il D. Lischinski e-mail: danix@cs.huji.ac.il L. Joskowicz e-mail: josko@cs.huji.ac.il Conclusion Our results indicate that our method is accurate, efficient, and robust to wide variety of tumor types and is comparable or superior to other semi-automatic segmenta- tion methods, with much less user interaction. Keywords Liver tumors segmentation · Computed Tomography · Constraint optimization Introduction Accurate detection and monitoring of liver tumors is a key task in many clinical applications. These include hepatomeg- aly and liver cirrhosis assessment, hepatic volumetry, hepatic transplantation planning, liver regeneration after hepatec- tomy, evaluation and planning for resection liver surgery, and monitoring of liver metastases, among many others. Repeat- able and reliable detection and quantification of tumor bur- den and tumor volume is necessary for accurate and timely decision-making regarding therapy options. Currently, most radiologists use simple guidelines to esti- mate tumor volume and response from clinical images. For 2D X-ray images, the 1979 World Health Organization [1] and the newer RECIST [2] guidelines define the tumor burden as the product of its maximum diameter—the largest distance between in-tumor points—and the maximum perpendicular diameter. This measure provides only a rough approximation from a single 2D projection image. When 3D CTA images are available, radiologists estimate the tumor volume with the three-parameter ellipsoid formula: max-length × max-depth × max-width × 0.5233 [3]. This yields reasonable volume estimates for tumors with nearly spherical or ellipsoid shapes, which occur in specific types of benign and malignant tumors. However, it is less accurate for most other tumors, as they usually have irregular borders 123