Cancer Classification with Radiomics

Radiomics involves the extraction of high-dimensional quantitative features from medical images to aid clinical decision-making. While radiomics has shown promise in predicting disease characteristics, concerns regarding confounders, reproducibility, and interpretability limit its widespread adoption. In this study, we assessed the ability of radiomic features extracted from contoured CT images to classify two distinct tumour models, CT26 colorectal cancer (CRC) and 4T1 breast cancer (BC), in a highly controlled murine setting. In such a highly controlled environment, we hoped to provide compelling evidence for the true merit of radiomics as a cancer biomarker. We benchmarked radiomics-based classification against previously established blood-based biomarkers, including leukocyte populations and plasma proteins. Feature filtering reduced the original 1409 radiomic features to 18 non-redundant, high-importance predictors, primarily texture-based transformations. Unsupervised clustering via UMAP revealed that radiomics-based features did not segregate tumour types as effectively as blood biomarkers, suggesting potential confounding factors. Supervised machine learning using Random Forest showed that radiomic features achieved a classification accuracy of 0.87, lower than the 0.96 and 0.99 accuracies obtained with cell and plasma biomarkers, respectively. Furthermore, integrating radiomics with blood biomarkers did not enhance classification performance, and feature importance analysis using SHAP identified blood-based markers as the dominant predictors. These findings suggest that while radiomics retains some predictive value, it is less effective than blood biomarkers in this classification task and does not significantly contribute to multimodal tumour classification models. Our study underscores the need for further standardization and validation of radiomics before its broader clinical implementation.
Type
collection
Title
Cancer Classification with Radiomics
Collection Type
Dataset
Access Privileges
The John Curtin School of Medical Research
DOI - Digital Object Identifier
10.25911/zkcm-ab43
Metadata Language
English
Data Language
English
Full Description
Radiomics involves the extraction of high-dimensional quantitative features from medical images to aid clinical decision-making. While radiomics has shown promise in predicting disease characteristics, concerns regarding confounders, reproducibility, and interpretability limit its widespread adoption. In this study, we assessed the ability of radiomic features extracted from contoured CT images to classify two distinct tumour models, CT26 colorectal cancer (CRC) and 4T1 breast cancer (BC), in a highly controlled murine setting. In such a highly controlled environment, we hoped to provide compelling evidence for the true merit of radiomics as a cancer biomarker. We benchmarked radiomics-based classification against previously established blood-based biomarkers, including leukocyte populations and plasma proteins. Feature filtering reduced the original 1409 radiomic features to 18 non-redundant, high-importance predictors, primarily texture-based transformations. Unsupervised clustering via UMAP revealed that radiomics-based features did not segregate tumour types as effectively as blood biomarkers, suggesting potential confounding factors. Supervised machine learning using Random Forest showed that radiomic features achieved a classification accuracy of 0.87, lower than the 0.96 and 0.99 accuracies obtained with cell and plasma biomarkers, respectively. Furthermore, integrating radiomics with blood biomarkers did not enhance classification performance, and feature importance analysis using SHAP identified blood-based markers as the dominant predictors. These findings suggest that while radiomics retains some predictive value, it is less effective than blood biomarkers in this classification task and does not significantly contribute to multimodal tumour classification models. Our study underscores the need for further standardization and validation of radiomics before its broader clinical implementation.
Contact Email
ben.quah@anu.edu.au
Principal Investigator
Benjamin Quah
Fields of Research
320206 - Diagnostic radiography
Socio-Economic Objective
200101 - Diagnosis of human diseases and conditions
Keywords
Radiomics, cancer, classification, machine learning, virtual biopsy
Type of Research Activity
Experimental development
Date Coverage
2024
2020
Date of data creation
2025
Year of data publication
2025
Creator(s) for Citation
Quah
Benjamin
Publisher for Citation
The Australian National University Data Commons
Access Rights
Open Access allowed
Access Rights Type
Open
Licence Type
CC-BY-NC - Attribution-NonCommercial (Version 4.0)
Data Size
1.3GB
Data Management Plan
No

Request data files

Number of files: 378

Size: 1.39 GB

Identifier: anudc:6380

Status: Published
Published to:
  • Australian National University
  • Australian National Data Service
Related items