Computer-assisted classification of the squamocolumnar junction

Abstract

Background and Aims

An irregular Z-line is characterized by a squamocolumnar junction (SCJ) that extends proximally above the gastroesophageal junction by <1 cm, whereas Barrett’s esophagus is defined as a columnar-lined esophagus (CLE) that extends proximally by ≥1 cm with the presence of specialized intestinal metaplasia on biopsy sampling. Measurement of the CLE is most accurate for lengths ≥1 cm, and, as such, guidelines do not recommend biopsy sampling of an irregular Z-line when seen on endoscopy. However, a CLE is often estimated by visual inspection rather than direct measurement, making this characterization imprecise. In this study, we present methodology to standardize the characterization of the SCJ, hypothesizing that the shape of the Z-line can be used as a surrogate classifier. We present a computer-generated algorithm capable of automated segmentation and shape complexity quantification of the Z-line.

Methods

We selected and manually segmented 849 images of the Z-line. We used the nnUNet framework (Nature Methods, Heidelberg, Germany) to train a model to segment the Z-line. An additional dataset of 58 videos containing the Z-line were obtained from the Mayo Clinic Endoscopy video library. A high-quality image containing the Z-line was selected from each video. Ten gastroenterologists (5 esophageal experts) rated each of the 58 video–image pairs containing the Z-line as “regular” or “irregular,” including their degree of confidence. Fleiss κ statistics were used to determine interobserver variability. The “ground truth” classification was determined by the esophageal expert majority vote. A wavelet decomposition model was then used to determine the threshold of irregularity based on the ground truth. Heat maps were generated for each Z-line to determine localized areas of complexity.

Results

Fair agreement, with a Fleiss κ of .39, was observed among the 10 endoscopists when rating the Z-line as regular versus irregular using this dataset. Moderate agreement was observed among the 5 esophageal experts with a Fleiss κ statistic of .42, and fair agreement was observed among the 5 nonesophageal experts with a Fleiss κ statistic of .31. The wavelet energy coefficient optimal threshold to classify an SCJ as irregular was determined to be 1.53 × 10⁷ with an accuracy of 78%.

Conclusions

Our computer-generated model was capable of automatic segmentation and classification of the Z-line. We established a threshold of complexity using the wavelet energy coefficient to standardize the classification of the SCJ.