Department of Computer Engineering, Faculty of Engineering, Chulalongkorn University, Pathumwan, Bangkok, Thailand.
Division of Gastroenterology, Department of Medicine, Faculty of Medicine, Chulalongkorn University and King Chulalongkorn Memorial Hospital, Bangkok, Thailand.
Comput Biol Med. 2023 Mar;154:106582. doi: 10.1016/j.compbiomed.2023.106582. Epub 2023 Jan 24.
This work presents real-time segmentation viz. gastric intestinal metaplasia (GIM). Recently, GIM segmentation of endoscopic images has been carried out to differentiate GIM from a healthy stomach. However, real-time detection is difficult to achieve. Conditions are challenging, and include multiple color modes (white light endoscopy and narrow-band imaging), other abnormal lesions (erosion and ulcer), noisy labels etc. Herein, our model is based on BiSeNet and can overcome the many issues regarding GIM. Application of auxiliary head and additional loss are seen to improve performance as well as enhance multiple color modes accurately. Further, multiple pre-processing techniques are utilized for leveraging detection performance: namely, location-wise negative sampling, jigsaw augmentation, and label smoothing. Finally, the decision threshold can be adjusted separately for each color mode. Work undertaken at King Chulalongkorn Memorial Hospital examined 940 histologically proven GIM images and 1239 non-GIM images, obtained over 173 frames per second (FPS). In terms of accuracy, our model is seen to outperform all baselines. Our results demonstrate sensitivity, specificity, positive predictive, negative predictive, accuracy, and mean intersection over union (IoU), achieving GIM segmentation values of 91%, 96%, 91%, 91%, 96%, and 55%, respectively.
本研究提出了实时分割，即胃肠上皮化生（GIM）。最近，已经对内镜图像的 GIM 分割进行了研究，以将 GIM 与健康胃区分开来。然而，实时检测很难实现。存在多种挑战条件，包括多种颜色模式（白光内镜和窄带成像）、其他异常病变（糜烂和溃疡）、噪声标签等。在此，我们的模型基于 BiSeNet，可以克服 GIM 相关的许多问题。辅助头的应用和附加损失的引入，提高了性能，同时准确地增强了多种颜色模式。此外，还利用了多种预处理技术来提高检测性能：即位置负采样、拼图增强和标签平滑。最后，可以分别为每种颜色模式调整决策阈值。在朱拉隆功国王纪念医院进行的工作中，检查了 940 张经组织学证实的 GIM 图像和 1239 张非 GIM 图像，每秒可获得 173 帧。在准确性方面，我们的模型明显优于所有基线。我们的结果表明，敏感性、特异性、阳性预测值、阴性预测值、准确性和平均交并率（IoU）分别达到 91%、96%、91%、91%、96%和 55%，实现了 GIM 分割。
