MODID: Multispectral oral disease image dataset with segmentaion

In oral imaging spectroscopy, tissue data is often collected from resected
samples, which may not accurately reflect true tissue signatures due to
preservation chemicals. To overcome this limitation, we aim to capture
in-vivo spectral signatures for oral diseases. Our dataset comprises
spectral images acquired in 16 bands, ranging from 460 nm to 600 nm, with
a resolution of 270x510 pixels. We collected these images from 91
volunteers, including 15 healthy individuals and 78 patients with various
oral health conditions. Detailed patient history records are also included
for each case. This publicly available multispectral dataset holds great
potential for advancing spectroscopy diagnosis. By leveraging artificial
intelligence in conjunction with this robust spectral image repository,
more accurate and insightful outcomes can be achieved in the analysis and
diagnosis of oral diseases.  Access to this comprehensive oral
health dataset contributes to enhancing our understanding of tissue
spectroscopy variability and facilitates further research in the field of
disease analysis.

搜集汇总

背景与挑战

背景概述

MODID是一个多光谱口腔疾病图像数据集，包含来自91名志愿者的244个样本，每个样本包含16个波段（460 nm至600 nm）的多光谱图像。数据集提供了未处理和已处理的数据文件，以及RGB图像和分割掩码，适用于口腔疾病的AI辅助诊断研究。