The effect of base image window level selection on the dimensional measurement accuracy of resultant three-dimensional image displays A. P. Kurmis, BAppSc (Med Rad), BMR (Hons), PhD candidate* T. C. Hearn, PhD, MSc, BSc, Professor and Chair, Orthopaedic Researchyz and K. J. Reynolds, PhD, MA, MS, BA (Hons), Senior Lecturer *§ *School of Informatics and Engineering, Flinders University, GPO Box 2100, Adelaide, SA 5001, Australia; yRepatriation General Hospital, Flinders University, Daws Road, Daw Park, SA 5041, Australia Purpose: The aim of this study was to determine the effect of base image window level selection on direct linear measurement of knee structures displayed using new magnetic resonance (MR)-based three-dimensional reconstructed computer imaging techniques. Methods: A prospective comparative study was performed using a series of three- dimensional knee images, generated from conventional MR imaging (MRI) sections. Thirty distinct anatomical structural features were identified within the image series of which repeated measurements were compared at 10 different window grey scale levels. Results: Statistical analysis demonstrated an excellent raw correlation between measurements and suggested no significant difference between measurements made at each of the 10 window level settings (P [0:05). Conclusions: The findings of this study suggest that unlike conventional MR or CT applications, grey scale window level selection at the time of imaging does not signifi- cantly affect the visual quality of resultant three-dimensional reconstructed images and hence the accuracy of subsequent direct linear measurement. The diagnostic potential of clinical progression from routine two-dimensional to advanced three-dimensional re- constructed imaging techniques may therefore be less likely to be degraded by inappro- priate MR technician image windowing during the capturing of image series. Ó 2003 The College of Radiographers. Published by Elsevier Ltd. All rights reserved. INTRODUCTION The aim of the present study was to determine the effect of window level grey scale selection on resul- tant three-dimensional image generation. Previous work has demonstrated the accuracy of post-scan- ner reconstruction computer packages in generating dimensional measurements taken from three-dimen- sional images [1], while utilising software defined ÔoptimalÕ default settings. While the general presentation abilities of many commercially available reconstructive software pack- ages have been recognised, there is little published evidence to support the fundamental assumption that the images displayed by such systems under ÔdefaultÕ settings are actually representative of the optimal viewing capabilities of the software output functions. KEY WORDS: three-dimensional reconstructive imaging; MR imaging; grey scale. (Received 21 November 2002; revised 12 May 2003; accepted 19 June 2003) Correspondence should be addressed to: Andrew P. Kurmis, RMB 971 Ackland Hill Road, Coromandel East, SA 5157, Australia. Tel: 61-8-8275-1753; Fax: 61-8-8374-1998. E-mail address: andrew.kurmis@flinders.edu.au z Tel: 61-8-8275-1753; Fax: 61-8-8374-1998. E-mail address: trevor.hearn@rgh.sa.gov.au § Tel: 61-8-8201-5190; Fax: 61-8-8201-3618. E-mail address: karen.reynolds@flinders.edu.au 1078–8174/03/000211+08 $30.00/0 Ó 2003 The College of Radiographers. Published by Elsevier Ltd. All rights reserved. Radiography (2003) 9, 211–218 doi:10.1016/S1078-8174(03)00055-5