Fractal analysis of contrast-enhanced CT images for preoperative prediction of malignant potential of gastrointestinal stromal tumor

Purpose The purpose of this study is to assess the heterogeneity of tumor enhancement using fractal analysis on contrast-enhanced computed tomography (CE-CT) for predicting malignant potential of gastrointestinal stromal tumor (GIST). Methods We retrospectively identified 64 patients (36 M/28 W; median age: 65) with GISTs who received CE-CT and 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) followed by curative surgery. Fractal analysis was applied to CE-CT image, and fractal dimension (FD) was measured. Diagnostic value of FD for malignant potential of GIST was compared with that of FDG-PET using the risk classification and Ki67 index. Results 14 patients were categorized as the high risk, and 50 patients were as the very low, low or intermediate risk. FD of high-risk group was significantly higher than that of the other-risk group ( p < 0.05). The areas under the ROC curves of FD and SUV max for prediction of high-risk group were 0.82 and 0.93 (accuracy: 84.4% and 98.5%). FD showed a significant positive correlation with Ki67 index ( p = 0.01). Conclusion Diagnostic value of CT fractal analysis for prediction of high-risk GIST is comparable with FDG-PET. In terms of cost and availability, fractal analysis has a potential to be an optimal preoperative biomarker.