JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Predicting early recurrence in locally advanced gastric cancer after gastrectomy using CT-based deep learning model: a multicenter study

  • 0Zhejiang Key Laboratory of Imaging and Interventional Medicine, Zhejiang Engineering Research Center of Interventional Medicine Engineering and Biotechnology, Lishui Hospital, School of Medicine, Zhejiang University, Lishui, China.
International Journal of Surgery (london, England) +

|

December 23, 2024

|

December 23, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

A deep learning model (DLER MLP) using CT scans accurately predicts early recurrence in locally advanced gastric cancer (LAGC) patients. This tool aids in optimizing treatment strategies and monitoring for improved patient outcomes.

Area Of Science

  • Oncology
  • Radiology
  • Artificial Intelligence in Medicine

Background

  • Early recurrence in locally advanced gastric cancer (LAGC) is linked to poor prognosis.
  • Accurate prediction of early recurrence is crucial for tailoring treatment strategies.
  • Identifying predictive markers can improve patient management and outcomes.

Purpose Of The Study

  • To develop a deep learning model (DLER) for predicting early recurrence in LAGC patients using preoperative CT images.
  • To integrate clinical factors with the DLER model (DLER MLP) for enhanced prediction accuracy.
  • To explore the biological basis associated with early recurrence predicted by the DLER MLP model.

Main Methods

  • A retrospective study involving 620 LAGC patients from three medical centers and TCIA.
  • Development of a DLER model using DenseNet169 and multiphase 2.5D CT images.
  • Integration of clinical factors into a multilayer perceptron (MLP) classifier (DLER MLP) for prediction.
  • Performance evaluation using AUC, accuracy, sensitivity, and specificity; survival analysis using log-rank test; genetic analysis via RNA-sequencing.

Main Results

  • The DLER MLP model demonstrated superior performance in predicting early recurrence compared to DLER and clinical models across internal and external validation sets (AUCs ranging from 0.814 to 0.891).
  • DLER MLP effectively stratified patients based on early recurrence-free survival, disease-free survival, and overall survival (P < 0.001).
  • High DLER MLP scores correlated with upregulated tumor proliferation pathways (WNT, MYC, KRAS) and increased immune cell infiltration.

Conclusions

  • The DLER MLP model, utilizing CT images, is a valuable tool for predicting early recurrence in LAGC patients.
  • This AI-driven approach can assist in optimizing treatment strategies and patient monitoring.
  • The model's findings provide insights into the biological underpinnings of early recurrence in gastric cancer.