Researchers from China built a combined model to predict the likelihood of liver metastasis in high-risk gastrointestinal stromal tumors (GISTs).
“Texture features from [volumes of interest] on portal venous phase [computed tomography (CT)] images have the value of effectively predicting liver metastasis of high-risk GISTs noninvasively, and further incremental discriminative value was verified when it incorporated with significant clinical variables,” they wrote in the journal Abdominal Radiology.
The classification performance of the model was evaluated in a validation cohort (n=62), showing an area under
the curve of 0.919, sensitivity of 83.9%, specificity of 89.7%, and accuracy of 84.9%.
To build the method, Zheng et al selected the 9 most robust predictors out of 110 texture features extracted from contrast-enhanced CT scans. Then, they added clinical predictors to the texture-based model, ie, independent clinical risk factors identified through univariate logistic analysis.
The clinical parameters included were body mass index (BMI), tumor localization, gastrointestinal bleeding, and the level of cancer antigen 125 (CA 125).
The use of quantitative CT texture features has the advantage of not involving invasive techniques, in contrast to most conventional biomarkers. Moreover, noninvasive imaging biomarkers are thought to be relatively objective and highly reproducible.
Read more about GIST testing
“Texture analysis refers to the investigation of the spatial distribution of the intensities or colors in images,” Zheng et al explained. “Some relative studies have proved that texture-specific patterns generate tumor phenotype that may be imperceptible by a human reader leads to accuracy improvement of the tumor diagnosis and prognosis assessment, as well as provides significant information about tumor heterogeneity and individualized clinical decision making for clinicians.”
Zheng J, Xia Y, Xu A, et al. Combined model based on enhanced CT texture features in liver metastasis prediction of high-risk gastrointestinal stromal tumors. Abdom Radiol. Published online October 27, 2021. doi:10.1007/s00261-021-03321-3