A Machine Learning Model Based on MRI Radiomics to Predict Response to Chemoradiation Among Patients with Rectal Cancer

2024

Predicting Rectal Cancer Treatment Response with MRI and Machine Learning

Sample size: 102 publication 10 minutes Evidence: moderate

Author Information

Author(s): Filippo Crimì, Carlo D’Alessandro, Chiara Zanon, Francesco Celotto, Christian Salvatore, Matteo Interlenghi, Isabella Castiglioni, Emilio Quaia, Salvatore Pucciarelli, Gaya Spolverato, Andrew Edet Ekpenyong

Primary Institution: University of Padova

Hypothesis

Can a machine learning model based on MRI radiomics predict the response to chemoradiation in rectal cancer patients?

Conclusion

The study demonstrates that radiomics and machine learning can effectively predict treatment response in rectal cancer patients.

Supporting Evidence

The best-performing model achieved a ROC-AUC of 73% and an accuracy of 70%.
In the validation cohort, the model showed a sensitivity of 81% and specificity of 75%.
High sensitivity and positive predictive value (PPV) were observed, indicating good accuracy in detecting non-responders.

Takeaway

Doctors can use special computer programs to look at MRI pictures of rectal cancer to see how well treatment is working.

Methodology

The study used MRI data from 102 patients, dividing them into training and validation cohorts, and applied machine learning models to predict treatment response.