Dental Odontogenic Lesion CBCT and Histopathology Integrated Dataset for Benchmarking Deep Learning Algorithms

The Dental Odontogenic Lesion CBCT and Histopathology Integrated Dataset (DOLCHID) is the first publicly available paired resource that combines radiological and pathological data for odontogenic lesion research. It comprises 262 patient cases, each including a pre-operative cone-beam computed tomography (CBCT) scan with expert-verified segmentation masks and a corresponding post-operative haematoxylin and eosin (H&E)-stained histopathology slide with annotated regions of interest. The dataset covers four major subtypes of odontogenic lesions, including 44 dentigerous cysts, 54 radicular cysts, 92 odontogenic keratocysts, and 72 ameloblastomas, thereby reflecting clinically relevant diversity across lesion types. This resource is designed to support the development and benchmarking of artificial intelligence methods in dental imaging, with the ultimate goal of enabling integrative diagnostic modelling that leverages complementary radiological and histological information.