© FUJIFILM Corporation

Novel Computer-Aided Diagnosis Software for the Prevention of Retained Surgical Items

2026.04.24Publications

Retained surgical gauze or sponges are a serious medical error, accounting for approximately 70% of retained foreign bodies. They can cause tumor-like masses, fistulas, reoperation, and even death. Although the incidence is estimated at about one in 10,000 surgeries, the additional medical cost can exceed USD 200,000 per case. Current preventive measures mainly rely on manual sponge counting by nurses and visual inspection of postoperative X-ray images by physicians; however, retained gauze has still been reported even when sponge counts were correct. Other technologies, such as barcode-based systems, have also been proposed, but their adoption is limited by additional equipment costs and increased workload.

This study developed computer-aided detection software that automatically identifies retained surgical gauze on postoperative X-ray images using an improved Faster R-CNN object detection algorithm. Because actual clinical cases of retained gauze are extremely rare, the authors generated approximately 5,000 synthetic X-ray images by superimposing gauze images onto normal postoperative radiographs; 4,554 images were used for training and 470 for validation. The software was also evaluated using 12 chest phantom images, 369 cadaver X-ray images acquired after gauze insertion, and 1,776 normal postoperative X-ray images. The system achieved high detection performance, with 100% sensitivity and specificity for phantom images, 97.9% sensitivity and 83.8% specificity for synthetic images, and 97.7% sensitivity and 90.4% specificity for cadaver images. It also successfully detected gauze overlapping with bones or drainage tubes, and the processing time was approximately 10 seconds, suggesting feasibility for use in clinical workflow.

Figure 1. Postoperative X-ray image with synthetic gauze overlay [1]

Figure 2. Cadaver X-ray image with inserted gauze [1]
[1] Yamaguchi S, Soyama A, Ono S, Hamauzu S, Yamada M, Fukuda T, Hidaka M, Tsurumoto T, Uetani M, Eguchi S. Novel computer-aided diagnosis software for the prevention of retained surgical items. J Am Coll Surg. 2021;233(6):686–696.

DOI: https://doi.org/10.1016/j.jamcollsurg.2021.08.689


CAUTION:This is Fujifilm Global Website. Fujifilm makes no representation that products on this website are commercially available in all countries. Approved uses of products vary by country and region. Specifications and appearance of products are subject to change without notice.

Index