Multiparametric MRI-based multi-channel deep learning model for accurate preoperative prediction of perineural invasion in lymph node-negative rectal cancer.
An MRI-based AI model predicts perineural invasion in rectal cancer, helping doctors decide which patients need stronger treatment.
A retrospective multicenter study of 266 rectal cancer patients developed a 12-channel deep learning model using multiparametric MRI sequences (T2WI, DWI, CE-T1WI) preprocessed into four ROI views to predict perineural invasion preoperatively in lymph node-negative cases. The multi-channel fusion DL model demonstrated strong diagnostic performance (AUC assessed by bootstrap resampling) and outperformed individual radiomics approaches, supporting its potential to guide treatment intensification decisions.
What the study was
- Study design
- retrospective multicenter diagnostic accuracy
- Population
- lymph node-negative rectal cancer patients (preoperative MRI cohort)
- Sample size
- 266
- Category
- Diagnostics
- Maturity
- Validated
- Journal
- Abdominal Radiology
Why it surfaced
Addresses clinically important PNI prediction gap in rectal cancer where PNI influences neoadjuvant treatment decisions; multicenter design supports generalizability, though retrospective nature and single-center enrollment limit immediate clinical readiness.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.