Analysis of Computed Tomography Image Features of Clear Cell Renal Cell Carcinoma on The Basis of Machine Learning: Differentiation of High-Grade from Low-Grade Fuhrman Nuclear Grades

Abstract Background Previous researches have clarified potential clinical applications of radiomics-based prediction of tumor phenotype. The purpose of our research is to utilize radiomic features in computer-aided diagnosis (CAD) system of prediction for high Fuhrman nuclear grade and low Fuhrman nuclear grade in clear cell renal cell carcinoma (RCC) Methods We selected 110 images from 109 case number of axial contrast- enhanced computed tomography with a pathological diagnosis of clear cell RCC from The Cancer Imaging Achieve (TCIA) portal. After preprocessing and extraction of features of from radiological images, the Weka Experiment Environment was run to compare performance of different classifiers to predict high Fuhrman nuclear grade and low Fuhrman nuclear grade. Results Seventy features and 23 selected features were input into classifiers to predict high Fuhrman nuclear grade and low Fuhrman nuclear grade, respectively. The SVM based on poly-kernel classifier (Accuracy, sensitivity, specificity, area under the receiver operative characteristic curve (AUROC), precision, recall and F1 score were 82%, 82%, 82%, 82%, 82%, 82% and 82%, respectively) and K-Nearest neighbors (Accuracy, sensitivity, specificity, AUROC, precision, recall and F1 score were 90.91%, 91%, 89%, 90%, 92%, 91% and 91.5%. Respectively) showed significant better performance than other classifiers, such as Bagging (Bagging), Naive Bayes (NaiveBayes), Boosting (AdaBoostM1), Decision trees (J48), Bayesian Network (BayesNet), library Support vector machines (LibSVM), Artificial Neural Networks (Multilayer Perceptron), Logistic regression (Logistic), Random Forest Tree (RandomForest),respectively. Conclusions The developed CAD system generated by the optimal machine-learning methods with radiomic features is a non-invasive method of predicting and monitoring tumor-nuclear grading characteristics in image-defined response evaluation in clinical practice of targeted therapy of clear cell RCC. Graphic Abstract: Process of machine learning

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-16
DOI
https://doi.org/10.21203/rs.3.rs-11364186/v1
Primary Topic
Radiomics and Machine Learning in Medical Imaging
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Analysis of Computed Tomography Image Features of Clear Cell Renal Cell Carcinoma on The Basis of Machine Learning: Differentiation of High-Grade from Low-Grade Fuhrman Nuclear Grades

Chung‐Ming Lo, Jenn Yeu Wang
1 links
Research Square
Radiomics and Machine Learning in Medical Imaging
preprint

Analysis of Computed Tomography Image Features of Clear Cell Renal Cell Carcinoma on The Basis of Machine Learning: Differentiation of High-Grade from Low-Grade Fuhrman Nuclear Grades

Chung‐Ming Lo, Jenn Yeu Wang
preprint en

Abstract

Abstract Background Previous researches have clarified potential clinical applications of radiomics-based prediction of tumor phenotype. The purpose of our research is to utilize radiomic features in computer-aided diagnosis (CAD) system of prediction for high Fuhrman nuclear grade and low Fuhrman nuclear grade in clear cell renal cell carcinoma (RCC) Methods We selected 110 images from 109 case number of axial contrast- enhanced computed tomography with a pathological diagnosis of clear cell RCC from The Cancer Imaging Achieve (TCIA) portal. After preprocessing and extraction of features of from radiological images, the Weka Experiment Environment was run to compare performance of different classifiers to predict high Fuhrman nuclear grade and low Fuhrman nuclear grade. Results Seventy features and 23 selected features were input into classifiers to predict high Fuhrman nuclear grade and low Fuhrman nuclear grade, respectively. The SVM based on poly-kernel classifier (Accuracy, sensitivity, specificity, area under the receiver operative characteristic curve (AUROC), precision, recall and F1 score were 82%, 82%, 82%, 82%, 82%, 82% and 82%, respectively) and K-Nearest neighbors (Accuracy, sensitivity, specificity, AUROC, precision, recall and F1 score were 90.91%, 91%, 89%, 90%, 92%, 91% and 91.5%. Respectively) showed significant better performance than other classifiers, such as Bagging (Bagging), Naive Bayes (NaiveBayes), Boosting (AdaBoostM1), Decision trees (J48), Bayesian Network (BayesNet), library Support vector machines (LibSVM), Artificial Neural Networks (Multilayer Perceptron), Logistic regression (Logistic), Random Forest Tree (RandomForest),respectively. Conclusions The developed CAD system generated by the optimal machine-learning methods with radiomic features is a non-invasive method of predicting and monitoring tumor-nuclear grading characteristics in image-defined response evaluation in clinical practice of targeted therapy of clear cell RCC. Graphic Abstract: Process of machine learning

Research Square
Taipei Municipal YangMing Hospital (TW), Taipei City Hospital (TW), National Chengchi University (TW)
Good health and well-being
Radiomics and Machine Learning in Medical Imaging
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon