[en] The purpose of this study was to investigate the predictive performance of 2D and 3D image features across multi-organ cancers using multi-modality images in radiomics studies.

In this retrospective study, we included 619 patients with three different cancer types (intrahepatic cholangiocarcinoma (ICC), high-grade osteosarcoma (HOS), pancreatic neuroendocrine tumors (pNETs)) and four clinical end points (early recurrence (ER), lymph node metastasis (LNM), 5-year survival and histologic grade). The image features included fifty-eight 2D image features and fifty-eight 3D image features. The 3D image features were extracted based on the 3D tumor volumes. The 2D image features were extracted based on 2D tumor region, which was the layer with the maximum tumor diameter within the 3D tumor volume. The predictive performance of individual 2D and 3D image feature was measured using the area under the receiver operating characteristic curve (AUC) with univariate analysis. Radiomics signatures were further developed using multivariable analysis with 4-fold cross-validation method.

Using univariate analysis, we found that more 3D image features showed the statistically predictive capabilities than 2D image features across all the included cancer types. By comparing the predictive performance of radiomics signatures developed by 2D and 3D image features, we observed better prediction performance in radiomics signatures based on 3D image features than those based on 2D image features for patients with ICC and HGO. Meanwhile, the signatures based on 2D and 3D image features performed closely in the pNETs dataset with the clinical end point of the histologic grade. The reason for this inconsistent result might be that the gross tumor volumes of pNETs were generally small. The tumor heterogeneity was mostly presented in the middle several layers within the tumor volume.

Both 2D and 3D image features have certain predictive capacities. By contrast, the 3D image features show better or close predictive performance than 2D image features using both univariate analysis and multivariate analysis. In brief, 3D image features are recommended in radiomics studies. (paper)

[en] Objectives. To test the effect of traditional up-sampling slice thickness (ST) methods on the reproducibility of CT radiomics features of liver tumors and investigate the improvement using a deep neural network (DNN) scheme. Methods. CT images with ≤ 1 mm ST in the public dataset were converted to low-resolution (3 mm, 5 mm) CT images. A DNN model was trained for the conversion from 3 mm ST and 5 mm ST to 1 mm ST and compared with conventional interpolation-based methods (cubic, linear, nearest) using structural similarity (SSIM) and peak-signal-to-noise-ratio (PSNR). Radiomics features were extracted from the tumor and tumor ring regions. The reproducibility of features from images converted using DNN and interpolation schemes were assessed using the concordance correlation coefficients (CCC) with the cutoff of 0.85. The paired t-test and Mann–Whitney U test were used to compare the evaluation metrics, where appropriate. Results. CT images of 108 patients were used for training (n = 63), validation (n = 11) and testing (n = 34). The DNN method showed significantly higher PSNR and SSIM values (p < 0.05) than interpolation-based methods. The DNN method also showed a significantly higher CCC value than interpolation-based methods. For features in the tumor region, compared with the cubic interpolation approach, the reproducible features increased from 393 (82%) to 422(88%) for the conversion of 3–1 mm, and from 305(64%) to 353(74%) for the conversion of 5–1 mm. For features in the tumor ring region, the improvement was from 395 (82%) to 431 (90%) and from 290 (60%) to 335 (70%), respectively. Conclusions. The DNN based ST up-sampling approach can improve the reproducibility of CT radiomics features in liver tumors, promoting the standardization of CT radiomics studies in liver cancer. (paper)

[en] Previous studies have suggested that the intratumoral texture features may reflect the tumor recurrence risk in intrahepatic cholangiocarcinoma (ICC). The peritumoral features may be associated with the distribution of microsatellites. Therefore, integrating the imaging features based on intratumoral and peritumoral areas may provide more accurate predictions in tumor recurrence (both early and late recurrences) than the predictions conducted based on the intratumoral area only. This retrospective study included 209 ICC patients. We divided the patient population into two sub-groups according to the order of diagnosis time: a training cohort (159 patients) and an independent validation cohort (50 patients). The MR imaging features were quantified based on the intratumoral and peritumoral (3 and 5 mm) areas. The radiomics signatures, clinical factor-based models and combined radiomics-clinical models were developed to predict the tumor recurrence. The prediction performance was measured based on the validation cohort using the area under receiver operating characteristic curve (AUC) index. For the prediction of early recurrence, the combined radiomics-clinical model of intratumoral area with 5 mm peritumoral area showed the highest performance (0.852(95% confidence interval (CI), 0.724–0.937)). The AUC for the clinical factor-based model was 0.805(95%CI, 0.668–0.903). For the prediction of late recurrence, the radiomics signature of intratumoral area with 5 mm peritumoral area had the optimal performance with an AUC of 0.735(95%CI, 0.591–0.850). The clinical factor-based showed inferior performance (0.598(95%CI, 0.450–0.735)). For both early and late recurrences prediction, the optimal models were all constructed using imaging features extracted based on intratumoral and peritumoral areas together. These suggested the importance of involving the intratumoral and peritumoral areas in the radiomics studies. (paper)