Measurement of Heterogeneity From 18F-FDG PET For Classification of Metastatic And Benign Bone Lesions: A Study In Patients With Cervical Cancer

Abstract Purpose Heterogeneity assessment has been applied in medical imaging analysis. We aim to evaluate first-order and texture analysis (TA) metrics in 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) imaging for classification of metastatic and benign bone lesions in patients with cervical cancer.Materials and Methods: 18F-FDG PET studies for cervical cancer patients performed on a specific PET/CT system from 2016 to 2018 were retrieved. Bone lesions extracted from studies in 2016-2017 were used as the training dataset and those in 2018 as the validation dataset. Metastatic bone lesions were identified in each dataset, with an equal number of benign bone lesions selected. Cuboid volume of interest (VOI) consisting of 3 by 3 by 5 reconstructed voxels was defined for first-order metrics and cubic VOI consisting of smaller voxels with trilinear interpolation of standardized uptake value (SUV) was adopted for TA metrics. First-order metrics included maximum SUV (SUVmax) of lesion and mean, standard deviation (SUVsd), skewness and kurtosis of voxel SUV in VOI. A total of 4464 TA metrics based on 62 texture features were evaluated. Logistic regression was used for classification with area under the receiver operating characteristic curve (AUC) as the performance measure.Results: From the training and validation datasets, 98 and 42 metastatic bone lesions were identified respectively. SUVsd achieves a better performance than SUVmax in both training (AUC .798 vs .732, P < .001) and validation (AUC .786 vs .684, P < .001) datasets. Top-performing TA metrics achieved significantly better performance in the training dataset but failed to retain this advantage in the validation dataset.Conclusion: This study identified a simple first-order measure of heterogeneity, SUVsd, to be superior to SUVmax in the classification of metastatic and benign bone lesions. The effect of multiple hypothesis testing can result in false-positive findings in TA with multiple features and parameters and careful validation is needed.