Radiomics models based on enhanced computed tomography to distinguish clear cell from non-clear cell renal cell carcinomas

Abstract

This study was to assess the effect of the predictive model for distinguishing clear cell RCC (ccRCC) from non-clear cell RCC (non-ccRCC) by establishing predictive radiomic models based on enhanced-computed tomography (CT) images of renal cell carcinoma (RCC). A total of 190 cases with RCC confirmed by pathology were retrospectively analyzed, with the patients being randomly divided into two groups, including the training set and testing set according to the ratio of 7:3. A total of 396 radiomic features were computationally obtained and analyzed with the Correlation between features, Univariate Logistics and Multivariate Logistics. Finally, 4 features were selected, and three machine models (Random Forest (RF), Support Vector Machine (SVM) and Logistic Regression (LR)) were established to discriminate RCC subtypes. The radiomics performance was compared with that of radiologist diagnosis. In the testing set, the RF model had an area under the curve (AUC) value of 0.909, a sensitivity of 0.956, and a specificity of 0.538. The SVM model had an AUC value of 0.841, a sensitivity of 1.0, and a specificity of 0.231, in the testing set. The LR model had an AUC value of 0.906, a sensitivity of 0.956, and a specificity of 0.692, in the testing set. The sensitivity and specificity of radiologist diagnosis to differentiate ccRCC from non-ccRCC were 0.850 and 0.581, respectively, with the AUC value of the radiologist diagnosis as 0.69. In conclusion, radiomics models based on CT imaging data show promise for augmenting radiological diagnosis in renal cancer, especially for differentiating ccRCC from non-ccRCC.

Evaluating Rumor Debunking Effectiveness During the COVID-19 Pandemic Crisis: Utilizing User Stance in Comments on Sina Weibo

“…Third, for the 100,000-level dataset in this research, based on an empirical rule, therefore we had to use a greater percentage of data to develop and test models and used the old way of splitting data to divide the dataset scientifically, that is, the non-redundant 12,000 comments were randomly divided into the independent training set and testing set according to the ratio of 7:3 (58)(59)(60). The label distribution is presented in Table 3.…”

Integrative Analysis of Histopathological Images and Genomic Data in Colon Adenocarcinoma

“…The samples were randomly divided into 10 parts, including 7 parts of the training set (n = 140) and 3 parts of the test set (n = 59). The ratio of 7:3 is commonly used in machine learning algorithms (32)(33)(34). The 10-fold cross-validation was used during model constructing.…”

Radiomics analysis of contrast-enhanced CT scans can distinguish between clear cell and non-clear cell renal cell carcinoma in different imaging protocols

“…Previous studies showed that, relative contrast enhancement of kidney tumors to the renal cortex ( 18 ) and CT imaging traits such as heterogeneous contrast enhancement, enhancement degree in corticomedullary phase, the presence of necrosis, and the presence of calcification show association with RCC subtypes ( 19 ). However, the morphology-based, conventional radiological evaluation of CT scans is subjective, has low specificity in differentiating RCC subtypes ( 20 ), and is highly dependent on the expertise of the radiologists ( 21 ).…”

“…Previously published studies have focused mainly on distinguishing between benign and malignant renal lesions ( 28 – 30 ) or on identifying aggressive tumor features of ccRCCs ( 31 – 37 ), and only a minority of studies have sought to distinguish between subtypes of RCC ( 20 , 38 – 41 ). A few studies also showed that radiomics analysis combined with machine learning could facilitate the non-invasive diagnostics of kidney cancers including both classification of renal tumors, prediction of nuclear grade, identification of patients with poor prognosis, and prediction of treatment response ( 42 , 43 ).…”