JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Machine learning-assisted radiogenomic analysis for miR-15a expression prediction in renal cell carcinoma

  • 0Voxel Medical Diagnostic Centers, Katowice, Poland. mytsyk.yulian@gmail.com.
Bmc Cancer +

|

August 21, 2025

|

August 21, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

Machine learning models predict microRNA-15a (miR-15a) expression in renal cell carcinoma (RCC) using radiogenomics. This non-invasive approach aids in stratifying tumors for personalized RCC management.

Area Of Science

  • Oncology
  • Radiology
  • Genomics
  • Machine Learning

Background

  • Renal cell carcinoma (RCC) outcomes vary significantly.
  • MicroRNA-15a (miR-15a) is a prognostic biomarker in RCC, influencing angiogenesis, apoptosis, and proliferation.
  • Radiogenomics integrates imaging features with molecular data for non-invasive biomarker prediction.

Purpose Of The Study

  • To develop a machine learning-assisted radiogenomic model for predicting miR-15a expression in RCC.
  • To correlate radiological features with miR-15a expression levels.
  • To stratify RCC tumors based on miR-15a expression and aggressiveness.

Main Methods

  • Retrospective analysis of 64 RCC patients with preoperative CT/MRI.
  • Evaluation of radiological features (tumor size, necrosis, enhancement).
  • Quantification of miR-15a expression via qPCR; prediction using Polynomial regression and Random Forest models; phenotypic stratification using hierarchical clustering and K-means analysis.

Main Results

  • Tumor size was the strongest predictor of miR-15a expression (adjusted R²=0.8281, p<0.001).
  • High miR-15a correlated with necrosis and nodular enhancement; low levels associated with cystic components and fat.
  • Random Forest model explained 65.8% of miR-15a expression variance (R²=0.658); classifier achieved AUC 1.0, precision 1.0, recall 0.9, F1-score 0.95.

Conclusions

  • Machine learning-based radiogenomics offers a robust, non-invasive method for predicting miR-15a expression in RCC.
  • This approach enhances tumor stratification and supports personalized RCC management.
  • Integrating radiological and molecular data is crucial for advancing precision medicine in oncology.

Keywords: