Prediction of clinically relevant postoperative pancreatic fistula using radiomic features and preoperative data

Abstract

Clinically relevant postoperative pancreatic fistula (CR-POPF) can significantly affect the treatment course and outcome in pancreatic cancer patients. Preoperative prediction of CR-POPF can aid the surgical decision-making process and lead to better perioperative management of patients. In this retrospective study of 108 pancreatic head resection patients, we present risk models for the prediction of CR-POPF that use combinations of preoperative computed tomography (CT)-based radiomic features, mesh-based volumes of annotated intra- and peripancreatic structures and preoperative clinical data. The risk signatures were evaluated and analysed in detail by visualising feature expression maps and by comparing significant features to the established CR-POPF risk measures. Out of the risk models that were developed in this study, the combined radiomic and clinical signature performed best with an average area under receiver operating characteristic curve (AUC) of 0.86 and a balanced accuracy score of 0.76 on validation data. The following pre-operative features showed significant correlation with outcome in this signature ([Formula: see text]) - texture and morphology of the healthy pancreatic segment, intensity volume histogram-based feature of the pancreatic duct segment, morphology of the combined segment, and BMI. The predictions of this pre-operative signature showed strong correlation (Spearman correlation co-efficient, [Formula: see text]) with the intraoperative updated alternative fistula risk score (ua-FRS), which is the clinical gold standard for intraoperative CR-POPF risk stratification. These results indicate that the proposed combined radiomic and clinical signature developed solely based on preoperatively available clinical and routine imaging data can perform on par with the current state-of-the-art intraoperative models for CR-POPF risk stratification.

Figure 1

Examples of delineated structures: healthy pancreas, pancreatic pathology, portal vein, arteries, pancreatic duct, bile duct.

Figure 2

Feature expression maps for the combined radiomic and clinical signature. The values indicated in the heatmaps are z-scores of standardised features, probability of developing CR-POPF as predicted by the model and the true outcome for all the patients included in the study. Abbreviations: CR-POPF pred prob - predicted probability of CR-POPF, $F_M$ - morphological feature, $F_T$ - texture feature, , $F_{\mathit{IVH}}$ - intensity volume histogram based feature.

Figure 3

Comparison of radiomic features with standard CR-POPF risk factors. Comparison of estimated pancreatic remnant volume with - (a) the morphological feature $F{1}_M$ of the healthy pancreatic segment; (b) the morphological feature $F{4}_M$ of the combined segment. The marginal distributions for both axes are illustrated according to the CR-POPF outcome (0 - no CR-POPF, 1 - CR-POPF). The value $\rho$ indicates the Spearman correlation co-efficient between the two variables. For comparability, the radiomic features and the estimated pancreatic remnant volume were standardised.

Figure 4

Comparison of radiomic features with standard CR-POPF risk factors. (a) Comparison of intraoperatively determined pancreatic duct width with the intensity volume histogram based feature $F{2}_{\mathit{IVH}}$ of the pancreatic duct segment. The marginal distributions for both axes are illustrated according to the CR-POPF outcome (0 - no CR-POPF, 1 - CR-POPF). The value $\rho$ indicates the Spearman correlation co-efficient between the two variables; (b) Comparison of intraoperatively determined pancreatic texture with the texture feature ${F3}_T$ of the healthy pancreatic segment. For comparability, the radiomic features and the pancreatic duct width were standardised.

Figure 5

Comparison of CR-POPF risk predicted by the combined clinical and radiomic signature with the CR-POPF risk predicted by the intraoperatively determined ua-FRS (clinical gold standard). The marginal distributions for both axes are illustrated according to the CR-POPF outcome (0 - no CR-POPF, 1 - CR-POPF). The value $\rho$ indicates the Spearman correlation co-efficient between the two variables.