Machine learning-based quantitative texture analysis of CT images of small renal masses: Differentiation of angiomyolipoma without visible fat from renal cell carcinoma

Abstract

• Although conventional CT is useful for diagnosis of SRMs, it has limitations. • Machine-learning based CT texture analysis facilitate differentiation of small AMLwvf from RCC. • The highest accuracy of SVM-RFE+SMOTE classifier reached 93.9 %. • Texture analysis combined with machine-learning methods might spare unnecessary surgery for AMLwvf.

Differentiation of atypical pancreatic neuroendocrine tumors from pancreatic ductal adenocarcinomas: Using whole‐tumor CT texture analysis as quantitative biomarkers

“…Kurtosis and entropy could, respectively, represent the shape and irregularity of the voxel distribution, which could represent tumor heterogeneity from different perspectives . Liu SL and Feng Z reported that kurtosis and entropy based on CTTA could help differentiate benign and malignant tumors. However, we did not find significant difference in kurtosis and entropy between PDAC and atypical pNET.…”

Quantitative Identification of Nonmuscle‐Invasive and Muscle‐Invasive Bladder Carcinomas: A Multiparametric MRI Radiomics Analysis

“…The results show that the first 19 features achieved the highest AUC (0.9819) with an accuracy of 92.59%, whereas the first nine features achieved the highest accuracy (92.59%) with an AUC of 0.9556 in classification. Owing to the imbalance of sample size between two groups, the AUC could better evaluate the comprehensive performance of the classifier for the differentiation task . Therefore, the subset consisting of 19 top‐ranked features was determined as the optimal subset …”

“…After 10 subsets are successively validated, one round classification is finished and the performance can be obtained. Owing to the random allocation of the 10‐fold subsets, only one round classification may not well reflect the overall performance of the samples . Instead, the procedures above are usually repeated for 100 rounds, and the final average performance can be achieved.…”

“…Considering that NMIBC samples were 20% less than MIBC samples, which may cause an adverse impact on the performance of an optimal subset with the SVM classifier, the synthetic minority oversampling technique (SMOTE) was performed with joint weighting of features in the optimal subset to generate samples from the minority group to balance the size of the majority group . The advantage of this method is to obtain the synthetic samples that have similar attribution values to the existing samples, “not merely replications,” thus enhancing the representation of the minority group while retaining the original structure of the samples . Using the balanced datasets, the performance of the optimal subset was revalidated.…”

Editorial Comment to Intensity ratio curve analysis of small renal masses on T2‐weighted magnetic resonance imaging: Differentiation of fat‐poor angiomyolipoma from renal cell carcinoma

“…However, the visual characteristics of CT imaging are insufficiently specific or overlap among various SRMs; for instance, between fat‐poor angiomyolipoma and renal cell carcinoma subtypes. Recently, quantitative methods have been developed to detect subtle variations beyond visual assessment on CT images…”