Background: Accurate labeling of obstruction site on upright abdominal radiograph is a challenging task. The lack of ground truth leads to poor performance on supervised learning models. To address this issue, self-supervised learning (SSL) is proposed to classify normal, small bowel obstruction (SBO), and large bowel obstruction (LBO) radiographs using a few confirmed samples.

Methods: A few number of confirmed and a large number of unlabeled radiographs were categorized based on the ground truth. The SSL model was firstly trained on the unlabeled radiographs, and then fine-tuned on the confirmed radiographs. ResNet50 and VGG16 were used for the embedded base encoders, whose weights and parameters were adjusted during training process. Furthermore, it was tested on an independent dataset, compared with supervised learning models and human interpreters. Finally, the t-SNE and Grad-CAM were used to visualize the model's interpretation.

Inclusion Criteria:

1. The hospital imaging system looked for plain abdominal standing films diagnosed as intestinal obstruction or normal between 2022 and 2024
2. Aged 18 to 80 years
3. The main complaint was gastrointestinal symptoms

Exclusion Criteria:

1. Image interference, fuzzy performance, difficult to distinguish
2. Non-gastrointestinal symptoms were the main complaint
3. Supine, prone, or lateral decubitus radiography
4. Paralytic obstruction, closed loop obstruction, et al

participants received upright abdominal radiographs were included in this study. They can be divided with participants with normal abdominal radiographs, participants with small bowel obstruction, and participants with large bowel obstruction. The study strictly follow the inclusion and exclusion criteria.