# Preoperative prediction of lymph node metastasis in pancreatic ductal adenocarcinoma using MRI-derived whole-tumor ADC histogram analysis

## TL;DR

This study shows that MRI-based ADC histogram analysis can help predict lymph node metastasis in pancreatic cancer before surgery.

## Contribution

The study introduces a multiparametric model using whole-tumor ADC histogram parameters for improved preoperative prediction of lymph node metastasis in PDAC.

## Key findings

- ADC histogram parameters (except coefficient of variation and kurtosis) significantly differ between lymph node metastasis and non-metastasis groups.
- The multi-parameter model achieved an AUC of 0.865 with 86.2% sensitivity and 75.0% specificity for predicting lymph node metastasis.
- Baseline clinical and conventional MRI features did not show significant differences between the groups.

## Abstract

To evaluate the predictive value of whole-tumor apparent diffusion coefficient (ADC) histogram parameters derived from MRI for assessing lymph node metastasis (LNM) in pancreatic ductal adenocarcinoma (PDAC).

Preoperative MRI and clinical data from 53 patients with pathologically confirmed PDAC were retrospectively analyzed. Patients were divided into two groups: LNM (n = 29) and non-LNM (NLNM, n = 24). ADC maps were generated from diffusion-weighted images acquired on a 3.0 T MRI scanner. Whole-tumor regions of interest were delineated in FireVoxel software to extract the full-volume ADC histogram parameters. A predictive model was developed and assessed using ROC analysis.

All ADC histogram parameters except the coefficient of variation and kurtosis showed significant differences between LNM and NLNM groups (p < 0.05); the first-order ADC values of LNM were significantly lower than those of NLNM. Baseline clinical characteristics (age, sex, clinical symptoms, CA19-9 levels) and conventional MRI features (size and volume) did not differ significantly. The multi-parameter model, based on select ADC-derived metrics, achieved an AUC of 0.865, with 86.2% sensitivity and 75.0% specificity.

Whole-tumor ADC histogram analysis provides a non-invasive and quantitative tool for preoperative prediction of lymph node metastasis in PDAC. The integrated multiparametric model demonstrates superior diagnostic performance compared with single-parameter analysis.