doi:10.1016/j.ijrobp.2007.02.005 PHYSICS CONTRIBUTION AUTOMATIC SEGMENTATION OF PELVIC STRUCTURES FROM MAGNETIC RESONANCE IMAGES FOR PROSTATE CANCER RADIOTHERAPY DAVID PASQUIER, M.D., PH.D.,* † THOMAS LACORNERIE,PH.D.,* MAXIMILIEN VERMANDEL,PH.D., † JEAN ROUSSEAU,PH.D., † ERIC LARTIGAU, M.D., PH.D.,* AND NACIM BETROUNI,PH.D. † *Département Universitaire de Radiothérapie, Centre Oscar Lambret, Université Lille II, Lille, France; † Laboratoire de Biophysique EA 1049, Institut National de la Santé Et de la Recherche Médicale U703 Thiais, Université Lille II, Lille, France; and Institut de Technologie Médicale, Centre Hospitalier Universitaire de Lille, Lille, France Purpose: Target-volume and organ-at-risk delineation is a time-consuming task in radiotherapy planning. The development of automated segmentation tools remains problematic, because of pelvic organ shape variability. We evaluate a three-dimensional (3D), deformable-model approach and a seeded region-growing algorithm for automatic delineation of the prostate and organs-at-risk on magnetic resonance images. Methods and Materials: Manual and automatic delineation were compared in 24 patients using a sagittal T2-weighted (T2-w) turbo spin echo (TSE) sequence and an axial T1-weighted (T1-w) 3D fast-field echo (FFE) or TSE sequence. For automatic prostate delineation, an organ model-based method was used. Prostates without seminal vesicles were delineated as the clinical target volume (CTV). For automatic bladder and rectum delineation, a seeded region-growing method was used. Manual contouring was considered the reference method. The following parameters were measured: volume ratio (Vr) (automatic/manual), volume overlap (Vo) (ratio of the volume of intersection to the volume of union; optimal value  1), and correctly delineated volume (Vc) (percent ratio of the volume of intersection to the manually defined volume; optimal value  100). Results: For the CTV, the Vr, Vo, and Vc were 1.13 (0.1 SD), 0.78 (0.05 SD), and 94.75 (3.3 SD), respectively. For the rectum, the Vr, Vo, and Vc were 0.97 (0.1 SD), 0.78 (0.06 SD), and 86.52 (5 SD), respectively. For the bladder, the Vr, Vo, and Vc were 0.95 (0.03 SD), 0.88 (0.03 SD), and 91.29 (3.1 SD), respectively. Conclusions: Our results show that the organ-model method is robust, and results in reproducible prostate segmentation with minor interactive corrections. For automatic bladder and rectum delineation, magnetic resonance imaging soft-tissue contrast enables the use of region-growing methods. © 2007 Elsevier Inc. Image segmentation, Deformable organ models, Radiotherapy, Treatment planning, Prostate cancer, MRI. INTRODUCTION Prostate cancer is one of the most common cancers in men. A common curative treatment is external-beam radiother- apy, administered as surgery or brachytherapy. Target-volume and organ-at-risk delineation is a time-consuming task in radiotherapy planning. The growth of conformal techniques, intensity-modulated radiotherapy (RT) with inverse plan- ning, and adaptive four-dimensional RT has increased the difficulty of organ-delineation tasks. The development of automated segmentation tools is thus essential but remains problematic, because of variable pelvic-organ shape and poor soft-tissue contrast on computed tomography (CT) scans. Pekar et al. (1) proposed an automated method for model-based organ-at-risk (rectum, bladder, and femoral head) delineation on CT images; deformable image regis- tration remains an area of active research in the context of image-guided RT (2–6). However, most published auto- matic techniques for organ-at-risk (OAR) or gross tumor volume (GTV) segmentation concern brain magnetic reso- nance imaging (MRI) (7–11). Soft-tissue contrast is better on magnetic resonance (MR) images than on CT scan images, and thus ensures better delineation of the prostate in general, and the apex and Reprint requests to: David Pasquier, M.D., Ph.D., Départe- ment Universitaire de Radiothérapie, Centre Oscar Lambret, Université Lille II, 3 rue F. Combemale, 59020 Lille, France. Tel: (+33) 3-20-29-59-11; Fax: (+33) 3-20-29-59-72; E-mail: d-pasquier@o-lambret.fr Presented at the 25th Annual Meeting of the European Society for Therapeutic Radiology and Oncology (ESTRO), Leipzig, Ger- many, October 8 –12, 2006. This study was supported by grants from the Centre Oscar Lambret, the Institut National de la Santé Et de la Recherche Médicale, and the Ministère de l’Education Nationale, de la Re- cherche et de la Technologie. Conflict of interest: none. Acknowledgments—The authors thank Emilie Ferrand, the radiol- ogists of the Centre Oscar Lambret, Charles Fournier of the Bio- statistics Unit in the Centre Oscar Lambret, and Aquilab SAS for their assistance. This work was planned by the Equipe de Recher- che Technologique Radiothérapie Conformationnelle et Imagerie (ERT 23, Lille, France). Received July 3, 2006, and in revised form Feb 6, 2007. Ac- cepted for publication Feb 8, 2007. Int. J. Radiation Oncology Biol. Phys., Vol. 68, No. 2, pp. 592– 600, 2007 Copyright © 2007 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/07/$–see front matter 592