Texture Analysis of Magnetic Resonance Enterography Contrast Enhancement Can Detect Fibrosis in Crohn Disease Strictures.
J Pediatr Gastroenterol Nutr 2019;
69:533-538. [PMID:
31365485 DOI:
10.1097/mpg.0000000000002454]
[Citation(s) in RCA: 23] [Impact Index Per Article: 4.6] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 12/19/2022]
Abstract
OBJECTIVE
The aim of the study was to investigate if texture analysis of contrast-enhanced magnetic resonance enterography (MRE) images can determine Crohn disease (CD) stricture histologic type.
MATERIALS AND METHODS
A radiology report database query identified 25 pediatric patients with established CD who underwent MRE followed by bowel resection within 30 days. MRE images were reviewed to identify strictures on enteric phase T1-weighted fat-suppressed images, that were matched with sites of histologic sectioning. Regions of interest were drawn over the bowel wall and texture analysis was performed using TexRAD software (Cambridge, UK), with skewness, mean, entropy and standard deviation parameters assessed. A pathologist reviewed all stricture histology specimens to assess for active mucosal inflammation and mural fibrosis. Multivariate logistic regression and analysis of variance were performed to identify texture features associated with stricture fibrosis.
RESULTS
Sixty-four bowel segments from 25 patients (mean age 16 ± 2 years) with imaging-histologic correlation were included. Of note, all strictures included had undergone surgical resection with MRE imaging available within 30 days. The histologic distribution of these bowel segments included 9 segments that showed active inflammation without fibrosis, 23 segments that showed only fibrosis, and 32 mixed segments with concomitant active inflammation and fibrosis. Bivariate regression analysis demonstrated that skewness, standard deviation, entropy, and mean texture analysis features are independently associated with stricture fibrosis. Stepwise logistic regression showed that the combination of mean, skewness, and entropy texture predicted stricture fibrosis with a goodness-of-fit value of 0.995. A combination of threshold values for these 3 texture analysis parameters was able to correctly classify 100% of the strictures in the study cohort for presence (55/55) and absence (9/9) of fibrosis.
CONCLUSIONS
MRE texture analysis (MRE-TA) texture features can differentiate CD stricture types and accurately detect fibrosis.
Collapse