JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

PET/computed tomography radiomics combined with clinical features in predicting sarcopenia and prognosis of diffuse large B-cell lymphoma

  • 0PET Center, Department of Nuclear Medicine, Guangdong Provincial People's Hospital (Guangdong Academy of Medical Sciences).
Nuclear Medicine Communications +

|

November 5, 2024

|

November 5, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

Machine learning models combining clinical data and 18F-FDG PET/CT radiomics effectively predict sarcopenia and prognosis in diffuse large B-cell lymphoma (DLBCL) patients. These models improve patient risk stratification and survival outcome prediction.

Area Of Science

  • Oncology
  • Radiology
  • Medical Imaging

Background

  • Diffuse large B-cell lymphoma (DLBCL) is an aggressive non-Hodgkin lymphoma.
  • Sarcopenia, or muscle loss, is a common complication in cancer patients and impacts prognosis.
  • Predictive biomarkers for sarcopenia and prognosis in DLBCL are crucial for personalized treatment.

Purpose Of The Study

  • To evaluate the efficacy of machine learning (ML) models integrating 18F-fluorodeoxyglucose (FDG) PET/computed tomography (CT) radiomics and clinical features.
  • To predict sarcopenia and assess prognosis in patients with DLBCL.

Main Methods

  • Retrospective analysis of 178 DLBCL patients who underwent pretreatment 18F-FDG PET/CT.
  • Feature selection using univariate logistic regression and correlation analysis.
  • Development and evaluation of ML models for sarcopenia prediction and survival prognosis (progression-free survival and overall survival).

Main Results

  • Combined clinical and PET/CT radiomics features significantly improved sarcopenia prediction models, with a support vector machine achieving an AUC of 0.862.
  • Prognostic models demonstrated consistency indices of 0.753 for PFS and 0.807 for OS.
  • Patients were stratified into high-risk and low-risk groups with significant differences in PFS and OS.

Conclusions

  • Machine learning models incorporating clinical data and PET/CT radiomics are effective tools for predicting sarcopenia in DLBCL patients.
  • These models can accurately assess prognosis, aiding in risk stratification and treatment planning for DLBCL.