Laryngeal Dataset for Comparative Study on CNN Based Semantic Segmentation

本研究介绍了一个名为‘Laryngeal Dataset for Comparative Study on CNN Based Semantic Segmentation’的数据集，由汉诺威应用科学大学创建。该数据集包含536张手动分割的人类喉部内窥镜图像，用于深度卷积神经网络（CNN）的语义分割研究。数据集涵盖了两个患者在激光切口手术期间的图像，每张图像分辨率为512×512像素，分为7个不同的类别。创建过程涉及手动标注和分类，旨在通过计算机视觉技术辅助医生进行早期病理检测和手术干预。该数据集的应用领域包括自动分割喉部软组织，用于监测形态变化和自主检测病理。

This study presents a dataset named 'Laryngeal Dataset for Comparative Study on CNN-Based Semantic Segmentation', developed by Hannover University of Applied Sciences. The dataset comprises 536 manually segmented human laryngeal endoscopic images intended for semantic segmentation research on deep convolutional neural networks (CNNs). It covers images from two patients during laser incision surgery, with each image having a resolution of 512×512 pixels and divided into 7 distinct categories. The dataset was constructed via manual annotation and classification, aiming to assist clinicians in early pathological detection and surgical intervention through computer vision technologies. Its potential application fields include automatic segmentation of laryngeal soft tissues, for monitoring morphological changes and autonomously detecting pathological conditions.