Dataset of Annotated Oral Cavity Images for Oral Cancer Detection

Citation

When using this resource, please cite the original publication:

N.S. Piyarathne, S.N. Liyanage, R.M.S.G.K. Rasnayaka, P.V.K.S. Hettiarachchi, G.A.I. Devindi, F.B.A.H. Francis, D.M.D.R. Dissanayake, R.A.N.S. Ranasinghe, M.B.D. Pavithya, I.B. Nawinne, R.G. Ragel, R.D. Jayasinghe, “A comprehensive dataset of annotated oral cavity images for diagnosis of oral cancer and oral potentially malignant disorders,” Oral Oncology, vol. 156, p. 106946, 2024.

Available at: https://doi.org/10.1016/j.oraloncology.2024.106946

Overview

The dataset consists of 3,000 high-quality images of oral cavities taken with mobile phone cameras from the Sri Lankan population. The images are categorized into healthy, benign, oral potentially malignant disorders (OPMD), and oral cancer (OCA) by domain experts. Each image contains annotations for oral cavity and lesion boundaries in COCO format. Additionally, patient metadata, such as age, sex, diagnosis, and risk factors like smoking, alcohol consumption, and betel quid chewing, is included in the meta-data files. The dataset contains the following files: 

Images.zip: Folder of oral cavity images.

Annotation.json: Annotations for the images provided in the JSON (JavaScript Object Notation) file.

Imagewise_data.csv: File including the image ID, category, clinical diagnosis, and the number of annotated regions for each image.

Patientwise_data.csv: File containing patient-level metadata, including age, sex, and total image count per patient. Additionally, it includes binary indicators for risk factors such as smoking, chewing betel quid, and alcohol consumption.

Request Access

You need to satisfy these terms and conditions in order for this request to be accepted:

Any request for access must be made through a Zenodo account associated with an official affiliation email address (e.g., institution, organization).

The following details are required as the request message:

A brief description of the intended use of the dataset, outlining the purpose of their study (maximum 250 words).

Principal Investigator’s name and affiliation.

Principal investigator’s page on the affiliation’s official website.

You may use this work only for non-commercial purposes. Any use intended for commercial gain or profit is strictly prohibited.

Proper credit must be given to the creator by citing the original article associated with this dataset. 

No modifications, derivatives, or adaptations of this work are allowed.

Requests that adhere to the terms and conditions will be processed within 2-3 working days.

Authors' Publications

A comprehensive dataset of annotated oral cavity images for diagnosis of oral cancer and oral potentially malignant disorders. DOI: 10.1016/j.oraloncology.2024.106946

Multimodal Deep Convolutional Neural Network Pipeline for AI-Assisted Early Detection of Oral Cancer. DOI: 10.1109/ACCESS.2024.3454338