# MRI radiomic study on prediction of nonenlarged lymph node metastasis of rectal cancer: reduced field-of-view versus conventional DWI

## TL;DR

This study compares MRI techniques to predict small lymph node metastasis in rectal cancer, finding that a newer method called rADC performs best.

## Contribution

The study introduces reduced field-of-view diffusion-weighted imaging (rDWI) radiomics as a more effective tool for predicting nonenlarged lymph node metastasis in rectal cancer.

## Key findings

- Radiomic features based on rADC outperformed conventional methods like T2WI and cDWI in predicting nonenlarged lymph node metastasis.
- rADC showed significantly higher diagnostic accuracy than cADC, cDWIb800, and T2WI in test cohorts.
- ADC maps from rDWI were more accurate than DWI for region-of-interest delineation in assessing metastasis.

## Abstract

Nonenlarged lymph node metastasis (NELNM) of rectal cancer is easily overlooked because these apparently normal lymph nodes are sometimes too small to measure directly using imaging techniques. Radiomic-based multiparametric imaging sequences could predict NELNM based on the primary lesion of rectal cancer. We aimed to study the performance of magnetic resonance imaging (MRI) radiomics derived from reduced field-of-view diffusion-weighted imaging (rDWI) and conventional DWI (cDWI) for the prediction of NELNM.

A total of 86 rectal cancer patients (60 and 26 patients in training and test cohorts, respectively), underwent multiparametric MRI. Radiomic features were extracted from the whole primary lesion of rectal cancer segmented on T2-weighted imaging (T2WI), rDWI, and cDWI, both with b-value of 800 s/mm2 and apparent diffusion coefficient (ADC) maps from both DWI sequences (rADC and cADC). The radiomic models based on the above imaging methods were built for the assessment of NELNM status. Their diagnostic performances were evaluated in comparison with subjective evaluation by radiologists.

rADC demonstrated a significant advantage over subjective assessment in predicting NELNM in both training and test cohorts (p ≤ 0.002). In the test cohort, rADC exhibited a significantly higher area under the receiver operating characteristics curve than cADC, cDWIb800, and T2WI (p ≤ 0.020) in assessing NELNM for region-of-interest (ROI) delineation while excelling over rDWIb800 for prediction of NELNM (p = 0.0498).

Radiomic features based on rADC outperformed those derived from T2WI and fDWI in predicting the NELNM status of rectal cancer, rADC was more advantageous than rDWIb800 in assessing NELNM.

Advanced rDWI excelled over cDWI in radiomic assessment of NELNM of rectal cancer, with the best performance observed for rADC, in contrast to rDWIb800, cADC, cDWIb800, and T2WI.

rDWI, cDWI, and T2WI radiomics could help assess NELNM of rectal cancer.Radiomic features based on rADC outperformed those based on rDWIb800, cADC, cDWIb800, and T2WI in predicting NELNM.For rDWI radiomics, the ADC map was more accurate and reliable than DWI to assess NELNM for region of interest delineation.

rDWI, cDWI, and T2WI radiomics could help assess NELNM of rectal cancer.

Radiomic features based on rADC outperformed those based on rDWIb800, cADC, cDWIb800, and T2WI in predicting NELNM.

For rDWI radiomics, the ADC map was more accurate and reliable than DWI to assess NELNM for region of interest delineation.