Measurement of Tumor Blood Flow Using Dynamic Contrast-enhanced Magnetic Resonance Imaging and Deconvolution Analysis: A Preliminary Study in Musculoskeletal Tumors Yoshifumi Sugawara, MD,* Kenya Murase, PhD,Þ Keiichi Kikuchi, MD,* Kenshi Sakayama, MD,þ Tatsuhiko Miyazaki, MD,§ Makoto Kajihara, MD,* Hitoshi Miki, MD,* and Teruhito Mochizuki, MD* Objective: To measure tumor blood flow (TBF) using dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). Methods: A DCE-MRI was performed using inversion recovery- preparation fast-field echo sequences. Dynamic data were obtained every 3.2 seconds for 2 minutes, immediately after gadolinium injection. In 14 patients with malignant musculoskeletal tumors, TBF maps were generated pixel-by-pixel by deconvolution analysis. For preclinical studies, muscle blood flow in 5 volunteers and signal intensities of different gadolinium concentrations were measured. Results: There was a good linear relationship between signal intensities and gadolinium concentrations (r = 0.989, P G 0.001, at gadolinium concentrations e2 mmol/L). The average value of mus- cle blood flow in volunteers was 11.1 T 2.7 mLI100 mL j1 Imin j1 . In 14 patients with musculoskeletal tumors, TBF showed wide variances: the lowest of 9.6 mLI100 mL j1 Imin j1 in liposarcoma and the highest of 182.0 mLI100 mL j1 Imin j1 in osteosarcoma. After chemotherapy, the TBF values (7.9, 11.0, and 11.7 mLI 100 mL j1 Imin j1 ) in the good responders were lower than those (26.8, 31.0, and 62.4 mLI100 mL j1 Imin j1 ) in the poor responders. Conclusions: A functional map of TBF generated by DCE-MRI and deconvolution analysis would be a promising tool for evaluating tumor blood flow in vivo. Key Words: musculoskeletal tumor, MRI, tumor blood flow, treatment response (J Comput Assist Tomogr 2006;30:983Y990) I t has been reported that angiogenesis correlates with tumor growth, invasion, and metastasis. 1Y4 In vivo tumor blood flow (TBF) increases as a result of angiogenesis, and it plays a key role in tumor growth and formation of metastasis. 5 Measurement of TBF in vivo may be valuable in the assessment of malignant tumors and their treatment responses. Because several antiangiogenic/antivascular agents are currently in clinical trials, 5,6 it is warranted to measure TBF noninvasively in vivo. However, in clinical practice, how to noninvasively measure TBF in vivo has not been established. Tumor blood flow has been measured with Doppler ultrasonography or positron emission tomography (PET), although several limitations in clinical use have been reported. 7,8 Doppler ultrasonography has limited sensitivity for recognizing blood flow in deeply located tumors and limited reproducibility because of its dependence on the examiner’s experience. 7 Quantitative measurement of TBF could be done by PET using diffusible tracers such as oxygen- 15 water; 8,9 however, it requires a cyclotron nearby, and such accessibility is limited. With recent advances in magnetic resonance (MR) systems and ultrafast magnetic resonance imaging (MRI) sequences, dynamic data with high spatial resolution can be obtained. 10Y12 These advances have enabled the monitoring of dynamic changes in signal intensities in vivo after a bolus injection of contrast materials. However, reports concerning the measurement of TBF using dynamic contrast-enhanced MRI (DCE-MRI) are limited. 13 According to the indicator dilution theory 14 and the nonparametric deconvolution tech- nique based on the singular value decomposition proposed by Ostergaard et al, 12 Pahernik et al 13 recently reported that, in experimental animal models, imaging of TBF could be obtained in vivo using DCE-MRI and deconvolution analysis. However, to the best of our knowledge, the feasibility of this technique has not been reported for measurement of TBF in humans. In this study, we developed methods for noninvasive measurement of TBF in human tumors using DCE-MRI and deconvolution analysis. MATERIALS AND METHODS Human Study Subjects Dynamic contrast-enhanced MRI was performed on 5 normal healthy volunteers (4 men and 1 woman; age range, ORIGINAL ARTICLE J Comput Assist Tomogr & Volume 30, Number 6, November/December 2006 983 From the *Department of Radiology, Ehime University School of Medicine, Ehime; †Department of Medical Engineering, Division of Allied Health Sciences, Osaka University Medical School, Osaka; and Departments of ‡Orthopaedic Surgery and §Pathology, Ehime University School of Medicine, Ehime, Japan. Received for publication April 11, 2006; accepted June 1, 2006. Reprints: Yoshifumi Sugawara, MD, Department of Radiology, Ehime University School of Medicine, Shitsukawa, Toon, Ehime 791-0295, Japan (e-mail: sugawara@m.ehime-u.ac.jp). Supported in part by a grant-in-aid for Scientific Research (C) (2) nos. 12670877 and 14570864 from the Japan Society for the Promotion of Science (JSPS) and Magnetic Health Science Foundation. Copyright * 2006 by Lippincott Williams & Wilkins Copyr ight © Lippincott Williams & Wilkins. Unauthor iz ed reproduction of this article is prohibited.