Srinivas Raman1, Lee Chin1, Darby Erler1, Eshetu G Atenafu2, Patrick Cheung1, William Chu1, Hans Chung1, Andrew Loblaw1, Ian Poon1, Joel Rubenstein3, Hany Soliman1, Arjun Sahgal1, Chia-Lin Tseng4. 1. Department of Radiation Oncology, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, Ontario, Canada. 2. Department of Biostatistics, University Health Network, University of Toronto, Toronto, Ontario, Canada. 3. Department of Diagnostic Radiology, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, Ontario, Canada. 4. Department of Radiation Oncology, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, Ontario, Canada. Electronic address: chia-lin.tseng@sunnybrook.ca.
Abstract
PURPOSE: This study investigates the inter-observer variability of contouring non-spine bone metastases using the planning CT alone vs. the addition of MRI T1 and T2 imaging sequences. METHODS AND MATERIALS: 10 cases of non-spine bone metastases treated with SBRT at our institution were selected. The gross tumor volume (GTV) for each case was delineated by six SBRT radiation oncologists (RO) and one diagnostic radiologist (DR) on the treatment planning CT. After a minimum of three months, each case was re-contoured on the CT fused with a MRI T1 sequence followed by a MRI T2 sequence. STAPLE consensus contours were created from the RO volumes and inter-observer variability was measured using both κ agreement and the Dice coefficient (DSC). RESULTS: In total, 180 RO contours were analyzed within three datasets (CT, CT + MRI T1 and CT + MRI T1 + MRI T2). The mean GTV was 16.95 cm3 (range, 0.12-269.6 cm3). The RO κ agreement was 0.6129 based on CT alone, and significantly increased to 0.7045 in the CT + MRI T1 (P = .042) dataset and 0.7017 in the CT + MRI T1 + MRI T2 dataset (P = .048). The mean DSC in the CT alone dataset was 0.7047, and significantly increased to 0.7628 in the CT + MRI T1 dataset (P < .001) and 0.7544 in the CT + MRI T1 + MRI T2 dataset (P = .001). There were no statistical differences in RO κ agreement (P = .948) or mean DSC (P = .573) when comparing the CT + MRI T1 and CT + MRI T1 + MRI T2 datasets. The DSC agreement between DR and RO volumes was lowest (0.6887) in the CT alone dataset and significantly increased to 0.7398 in the CT + MRI T1 dataset (P = .003) and 0.7342 in the CT + MRI T1 + MRI T2 dataset (P = .008). CONCLUSIONS: The fusion of MRI T1 images to CT significantly reduced inter-observer variability amongst RO's in delineating non-spine bone metastases, and improved agreement between GTVs delineated by the RO to the DR.
PURPOSE: This study investigates the inter-observer variability of contouring non-spine bone metastases using the planning CT alone vs. the addition of MRI T1 and T2 imaging sequences. METHODS AND MATERIALS: 10 cases of non-spine bone metastases treated with SBRT at our institution were selected. The gross tumor volume (GTV) for each case was delineated by six SBRT radiation oncologists (RO) and one diagnostic radiologist (DR) on the treatment planning CT. After a minimum of three months, each case was re-contoured on the CT fused with a MRI T1 sequence followed by a MRI T2 sequence. STAPLE consensus contours were created from the RO volumes and inter-observer variability was measured using both κ agreement and the Dice coefficient (DSC). RESULTS: In total, 180 RO contours were analyzed within three datasets (CT, CT + MRI T1 and CT + MRI T1 + MRI T2). The mean GTV was 16.95 cm3 (range, 0.12-269.6 cm3). The RO κ agreement was 0.6129 based on CT alone, and significantly increased to 0.7045 in the CT + MRI T1 (P = .042) dataset and 0.7017 in the CT + MRI T1 + MRI T2 dataset (P = .048). The mean DSC in the CT alone dataset was 0.7047, and significantly increased to 0.7628 in the CT + MRI T1 dataset (P < .001) and 0.7544 in the CT + MRI T1 + MRI T2 dataset (P = .001). There were no statistical differences in RO κ agreement (P = .948) or mean DSC (P = .573) when comparing the CT + MRI T1 and CT + MRI T1 + MRI T2 datasets. The DSC agreement between DR and RO volumes was lowest (0.6887) in the CT alone dataset and significantly increased to 0.7398 in the CT + MRI T1 dataset (P = .003) and 0.7342 in the CT + MRI T1 + MRI T2 dataset (P = .008). CONCLUSIONS: The fusion of MRI T1 images to CT significantly reduced inter-observer variability amongst RO's in delineating non-spine bone metastases, and improved agreement between GTVs delineated by the RO to the DR.
Authors: F Lopez-Campos; J Cacicedo; F Couñago; R García; O Leaman-Alcibar; A Navarro-Martin; H Pérez-Montero; A Conde-Moreno Journal: Clin Transl Oncol Date: 2021-10-11 Impact factor: 3.405
Authors: Samuel Finkelstein; Srinivas Raman; Joanne Van Der Velden; Liying Zhang; Carolyn Tan; Amanpreet Dhillon; Frances Tonolete; Nicholas Chiu; Linda Probyn; Rachel McDonald; Arjun Sahgal; Edward Chow; Lee Chin Journal: Technol Cancer Res Treat Date: 2019-01-01
Authors: Fang Li; Jeonghoon Park; Ron Lalonde; Si Young Jang; Maria Stefania diMayorca; John C Flickinger; Andrew Keller; Mohammed Saiful Huq Journal: J Appl Clin Med Phys Date: 2021-11-29 Impact factor: 2.102
Authors: Ali Ataei; Florieke Eggermont; Milan Baars; Yvette van der Linden; Jacky de Rooy; Nico Verdonschot; Esther Tanck Journal: Int J Comput Assist Radiol Surg Date: 2021-07-15 Impact factor: 2.924