Intestinal Wall Texture Analysis: Finding Fibrosis in Pediatric Patients with Crohn Disease.

n this issue’s article, ‘‘Texture analysis of magnetic resonance enterography contrast enhancement can detect fibrosis in Crohn I disease strictures,’’ Tabari and co-workers demonstrated that a computer had greater diagnostic accuracy than a human radiologist (1). Using histology as the gold standard for fibrosis, the computer distinguished no fibrosis and mild fibrosis from moderate-to-severe fibrosis with an impressive area under the curve (AUC) of 0.995. To appreciate the significance of this article and that AUC, it will be helpful to review how far imaging has advanced in the 87 years since Dr Burrill Crohn’s seminal paper. In patients diagnosed with Crohn disease, almost half will develop a fibrotic stricture within 10 years of first receiving their diagnosis, often requiring surgical resection (2). The gold standard for diagnosing fibrosis is the deposition of collagen, detected histologically from full-thickness bowel resection. Due to the submucosal and subserosal location of the collagen deposition, fibrosis may be difficult to appreciate by endoscopic visualization or partial-thickness endoscopic biopsy. By imaging, a fibrotic stricture classically demonstrates 3 anatomic features: luminal narrowing, wall thickening, and prestenotic dilation (3). However, these 3 anatomic features alone are late features and are nonspecific, as active inflammation without fibrosis can have a similar appearance. In Crohn’s original 1932 article, he acknowledged the promise and limitations of anatomic imaging: ‘‘The barium meal, however, when carefully interpreted, gives definite positive findings. ... though only in the late or stenotic stages is the delay striking. The midler degrees of stasis and puddling in the ileal loops may easily be overlooked by any but a careful roentgenologist’’(4). It would be another 50 years before cross-sectional imaging (including computed tomography, magnetic resonance imaging ‘‘MRI,’’ and ultrasound) demonstrated diagnostic superiority compared with the traditional fluoroscopic ‘‘barium meal,’’ or what we would now call a ‘‘small bowel follow through’’(5). However, improvements in anatomic detail would only take diagnostic accuracy so far. What