JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Enhanced CT-Based Delta-Radiomics: Predicting Lymphovascular and Perineural Invasion in Rectal Cancer Preoperatively

  • 0Department of Radiology, Affiliated Dongyang Hospital of Wenzhou Medical University, Dongyang, 322100, China.
Journal of Imaging Informatics in Medicine +

|

June 26, 2025

|

June 26, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

This study developed a novel delta-radiomics model using multi-phase contrast-enhanced CT to predict lymphovascular invasion (LVI) and perineural invasion (PNI) in rectal cancer (RC) patients. The combined model achieved an AUC of 0.81, offering accurate, non-invasive risk assessment.

Area Of Science

  • Radiology
  • Oncology
  • Medical Imaging Analysis

Background

  • Rectal cancer (RC) poses significant challenges in preoperative risk stratification.
  • Accurate prediction of lymphovascular invasion (LVI) and perineural invasion (PNI) is crucial for tailoring treatment strategies in RC.
  • Current imaging methods have limitations in precisely identifying these high-risk features.

Purpose Of The Study

  • To develop and validate a multi-phase contrast-enhanced computed tomography (CECT) delta-radiomics signature for preoperative prediction of LVI and PNI in RC.
  • To assess the performance of different delta-radiomics models in identifying LVI and PNI.
  • To evaluate the clinical utility of integrating delta-radiomics with clinical predictors for risk stratification in RC.

Main Methods

  • Retrospective analysis of 519 RC patients' CECT scans (January 2017-December 2022).
  • Extraction of radiomic features from routine (A0), arterial (A1), and venous (A2) phases.
  • Construction of delta-radiomics signatures (Delta-1 to Delta-4) using image subtraction and feature extraction, with a combined model (C-Delta-12) developed.
  • Model performance evaluated using ROC, calibration curves, and decision curve analysis.

Main Results

  • Individual delta-radiomics models showed moderate predictive performance for LVI and PNI (AUCs ranging from 0.67 to 0.73).
  • The combined C-Delta-12 model demonstrated superior predictive performance with an AUC of 0.81, accuracy of 0.76, sensitivity of 0.86, and specificity of 0.65.
  • Calibration curves indicated a good fit, and decision curve analysis confirmed the model's clinical value.

Conclusions

  • The developed multi-phase CECT delta-radiomics signature provides an accurate and non-invasive method for preoperative risk assessment of LVI and PNI in RC patients.
  • Integration of delta-radiomics with clinical predictors enhances prediction accuracy, potentially guiding individualized treatment decisions.
  • This approach facilitates better patient stratification based on LVI and PNI status, optimizing therapeutic strategies.

Keywords: