Lijun Lu1, Wenbing Lv2, Jun Jiang2, Jianhua Ma3, Qianjin Feng4, Arman Rahmim5,6, Wufan Chen2. 1. School of Biomedical Engineering and Guangdong Provincial Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou, Guangdong, 510515, China. ljlubme@gmail.com. 2. School of Biomedical Engineering and Guangdong Provincial Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou, Guangdong, 510515, China. 3. School of Biomedical Engineering and Guangdong Provincial Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou, Guangdong, 510515, China. jhma@smu.edu.cn. 4. School of Biomedical Engineering and Guangdong Provincial Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou, Guangdong, 510515, China. qianjinfeng08@gmail.com. 5. Department of Radiology, Johns Hopkins University, Baltimore, MD, 21287, USA. 6. Department of Electrical and Computer Engineering, Johns Hopkins University, Baltimore, MD, 21218, USA.
Abstract
PURPOSE: Radiomic features are increasingly utilized to evaluate tumor heterogeneity in PET imaging and to enable enhanced prediction of therapy response and outcome. An important ingredient to success in translation of radiomic features to clinical reality is to quantify and ascertain their robustness. In the present work, we studied the impact of segmentation and discretization on 88 radiomic features in 2-deoxy-2-[18F]fluoro-D-glucose ([18F]FDG) and [11C]methyl-choline ([11C]choline) positron emission tomography/X-ray computed tomography (PET/CT) imaging of nasopharyngeal carcinoma. PROCEDURES: Forty patients underwent [18F]FDG PET/CT scans. Of these, nine patients were imaged on a different day utilizing [11C]choline PET/CT. Tumors were delineated using reference manual segmentation by the consensus of three expert physicians, using 41, 50, and 70 % maximum standardized uptake value (SUVmax) threshold with background correction, Nestle's method, and watershed and region growing methods, and then discretized with fixed bin size (0.05, 0.1, 0.2, 0.5, and 1) in units of SUV. A total of 88 features, including 21 first-order intensity features, 10 shape features, and 57 second- and higher-order textural features, were extracted from the tumors. The robustness of the features was evaluated via the intraclass correlation coefficient (ICC) for seven kinds of segmentation methods (involving all 88 features) and five kinds of discretization bin size (involving the 57 second- and higher-order features). RESULTS: Forty-four (50 %) and 55 (63 %) features depicted ICC ≥0.8 with respect to segmentation as obtained from [18F]FDG and [11C]choline, respectively. Thirteen (23 %) and 12 (21 %) features showed ICC ≥0.8 with respect to discretization as obtained from [18F]FDG and [11C]choline, respectively. Six features were obtained from both [18F]FDG and [11C]choline having ICC ≥0.8 for both segmentation and discretization, five of which were gray-level co-occurrence matrix (GLCM) features (SumEntropy, Entropy, DifEntropy, Homogeneity1, and Homogeneity2) and one of which was an neighborhood gray-tone different matrix (NGTDM) feature (Coarseness). CONCLUSIONS: Discretization generated larger effects on features than segmentation in both tracers. Features extracted from [11C]choline were more robust than [18F]FDG for segmentation. Discretization had very similar effects on features extracted from both tracers.
PURPOSE: Radiomic features are increasingly utilized to evaluate tumor heterogeneity in PET imaging and to enable enhanced prediction of therapy response and outcome. An important ingredient to success in translation of radiomic features to clinical reality is to quantify and ascertain their robustness. In the present work, we studied the impact of segmentation and discretization on 88 radiomic features in 2-deoxy-2-[18F]fluoro-D-glucose ([18F]FDG) and [11C]methyl-choline ([11C]choline) positron emission tomography/X-ray computed tomography (PET/CT) imaging of nasopharyngeal carcinoma. PROCEDURES: Forty patients underwent [18F]FDG PET/CT scans. Of these, nine patients were imaged on a different day utilizing [11C]choline PET/CT. Tumors were delineated using reference manual segmentation by the consensus of three expert physicians, using 41, 50, and 70 % maximum standardized uptake value (SUVmax) threshold with background correction, Nestle's method, and watershed and region growing methods, and then discretized with fixed bin size (0.05, 0.1, 0.2, 0.5, and 1) in units of SUV. A total of 88 features, including 21 first-order intensity features, 10 shape features, and 57 second- and higher-order textural features, were extracted from the tumors. The robustness of the features was evaluated via the intraclass correlation coefficient (ICC) for seven kinds of segmentation methods (involving all 88 features) and five kinds of discretization bin size (involving the 57 second- and higher-order features). RESULTS: Forty-four (50 %) and 55 (63 %) features depicted ICC ≥0.8 with respect to segmentation as obtained from [18F]FDG and [11C]choline, respectively. Thirteen (23 %) and 12 (21 %) features showed ICC ≥0.8 with respect to discretization as obtained from [18F]FDG and [11C]choline, respectively. Six features were obtained from both [18F]FDG and [11C]choline having ICC ≥0.8 for both segmentation and discretization, five of which were gray-level co-occurrence matrix (GLCM) features (SumEntropy, Entropy, DifEntropy, Homogeneity1, and Homogeneity2) and one of which was an neighborhood gray-tone different matrix (NGTDM) feature (Coarseness). CONCLUSIONS: Discretization generated larger effects on features than segmentation in both tracers. Features extracted from [11C]choline were more robust than [18F]FDG for segmentation. Discretization had very similar effects on features extracted from both tracers.
Authors: Virginie Frings; Floris H P van Velden; Linda M Velasquez; Wendy Hayes; Peter M van de Ven; Otto S Hoekstra; Ronald Boellaard Journal: Radiology Date: 2014-05-26 Impact factor: 11.105
Authors: Ralph T H Leijenaar; Georgi Nalbantov; Sara Carvalho; Wouter J C van Elmpt; Esther G C Troost; Ronald Boellaard; Hugo J W L Aerts; Robert J Gillies; Philippe Lambin Journal: Sci Rep Date: 2015-08-05 Impact factor: 4.379
Authors: Pierre Lovinfosse; Marc Polus; Daniel Van Daele; Philippe Martinive; Frédéric Daenen; Mathieu Hatt; Dimitris Visvikis; Benjamin Koopmansch; Frédéric Lambert; Carla Coimbra; Laurence Seidel; Adelin Albert; Philippe Delvenne; Roland Hustinx Journal: Eur J Nucl Med Mol Imaging Date: 2017-10-18 Impact factor: 9.236
Authors: Muhammad Shafiq-Ul-Hassan; Geoffrey G Zhang; Kujtim Latifi; Ghanim Ullah; Dylan C Hunt; Yoganand Balagurunathan; Mahmoud Abrahem Abdalah; Matthew B Schabath; Dmitry G Goldgof; Dennis Mackin; Laurence Edward Court; Robert James Gillies; Eduardo Gerardo Moros Journal: Med Phys Date: 2017-03 Impact factor: 4.071