CT-based multi-phase Radiomic models for differentiating clear cell renal cell carcinoma

Abstract

Background The aim of the study is to compare the diagnostic value of models that based on a set of CT texture and non-texture features for differentiating clear cell renal cell carcinomas(ccRCCs) from non-clear cell renal cell carcinomas(non-ccRCCs). Methods A total of 197 pathologically proven renal tumors were divided into ccRCC(n = 143) and non-ccRCC (n = 54) groups. The 43 non-texture features and 296 texture features that extracted from the 3D volume tumor tissue were assessed for each tumor at both Non-contrast Phase, NCP; Corticomedullary Phase, CMP; Nephrographic Phase, NP and Excretory Phase, EP. Texture-score were calculated by the Least Absolute Shrinkage and Selection Operator (LASSO) to screen the most valuable texture features. Model 1 contains the three most distinctive non-texture features with p < 0.001, Model 2 contains texture scores, and Model 3 contains the above two types of features. Results The three models shown good discrimination of the ccRCC from non-ccRCC in NCP, CMP, NP, and EP. The area under receiver operating characteristic curve (AUC)values of the Model 1, Model 2, and Model 3 in differentiating the two groups were 0.748–0.823, 0.776–0.887 and 0.864–0.900, respectively. The difference in AUC between every two of the three Models was statistically significant (p < 0.001). Conclusions The predictive efficacy of ccRCC was significantly improved by combining non-texture features and texture features to construct a combined diagnostic model, which could provide a reliable basis for clinical treatment options.

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

“…The results of this study showed that adding non-texture features can improve the prediction performance of the texture feature-based model and that the CM phase and the NG phase radiomics 10.3389/fmed.2022.974485 features have similar diagnostic ability in differentiating between ccRCCs and non-ccRCCs. However, these models were not validated on an independent test set in this manuscript (38). The results of this study are comparable with the results of our SVC model trained on the CM phase radiomics features, especially with the results we reported on the training set (AUC = 0.951), however, we also validated the performance of our model on both independent internal (AUC = 0.873) and external test cases (AUC = 0.834).…”

“…In addition, we evaluated the accuracy of our SVC against that of an expert radiologist, which showed that the performance of the machine learning model is comparable (accuracy of 0.79 vs. 0.78 on the external dataset) which further supports the current literature and demonstrates the potential of CT texture analysis in this application. The majority of the previously published studies focused on differentiating between benign and malignant kidney lesions (28)(29)(30) or identifying aggressive tumor features of ccRCCs (31-37), and only a handful of studies aimed to distinguish between the RCC subtypes (20,(38)(39)(40)(41). It is important to highlight that previous studies used different softwares for radiomics feature extraction including both in-house developed algorithms (40,41), and open-source tools such as the MaZda software (39) and the pyRadiomics package (38) which complicates the direct comparison of the previously published results.…”

“…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 ).…”

“…Yu et al demonstrated the ability of first-order statistics and texture features to predict RCC subtypes ( 41 ). By analyzing triphasic CT scans of 143 ccRCCs and 54 non-ccRCCs, Chen et al illustrated that the radiomics features extracted from the corticomedullary (CM) phase have similar diagnostic ability compared to those extracted from the NG phase in differentiating between ccRCCs and non-ccRCCs ( 38 ). In their recent study, Wang et al analyzed 147 ccRCCs and 43 non-ccRCCs and built a RFC, an SVC, and a logistic regression algorithm-based machine learning model from four selected radiomics features.…”

Solitary fibrous tumor of the renal pelvis: A case report

“…Renal pelvis SFTs need to be differentiated from the more common renal pelvis tumors such as RCC, RAMLs, and UTUCs. Contrast-enhanced CT is the main method for the diagnosis of RCCs[ 10 ]. RCC is characterized by abundant blood supply, and tumor blood vessels and tumor staining can be observed on renal angiography.…”