Dataset
Download the dataset
About the dataset
Summary: We provide a dataset of skin Niacin flushing dataset consisting of pictures for people from four groups of HC (Health Control), DP (Depression), BP (Bipolar disorder), and SZ (Schizophrenia).
All the data were originally part of a larger, closed dataset of Wang et al. (2020), where all the patients were from hospitals in China. All the patients have a diagnosis of SZ, BP or DP according to ICD10 (International Classification of Diseases, 10th Revision) and are inpatient. The HCs do not have a lifetime history of psychiatric illness. Besides, all the participants should follow these principles: (1) no severe neurological diseases, or severe brain trauma; (2) no substance dependence; (3) not use of nonsteroidal or steroidal anti-inflammatory drugs during the week before niacin skin test; (4) no skin diseases or immune system diseases; (5) no pregnancy. All participants signed informed consent forms before the test. Data were collected following the guideline of the local ethics committee.
To develop an AI algorithm for screening, we (i.e., FWI) randomly selected a small set of pictures from the original set, then cut out the area of interest and marked the Niacin flushing area by 5 students, where a pixel is marked as a flushing pixel if at least two students marked so.
We are also developing algorithms for the segmentation of this dataset. While we cannot provide the source file due to IP issues, the results are shown in the benchmark page. If you are interested in this project, you are welcome to contact us. If you have a publication that used this dataset, please cite the following papers.
- Wang Dandan, Hu Xiaowen, Jiang Jie, Sun Liya, Qing Ying, Yang Xuhan, Gao Yan, Cui Gaoping, Li Minghui, Wang Pengkun, Zhang Juan, Zhuang Yan, Li Zezhi, Li Jing, Guan Lili, Zhang Tianhong, Wang Jijun, Ji Feng, Wan Chunling. Attenuated and Delayed Niacin Skin Flushing in Schizophrenia and Affective Disorders: A Potential Clinical Auxiliary Diagnostic Marker. Submitted for publication, 2021.
- Xuening Lyu, Liang Zhao, Dandan Wang, Chunling Wan, An Open Dataset and an Algorithm for Niacin Skin-flushing Segmentation with Application in Psychiatric Disorders Classification, under preparation, 2021.
Detail of the dataset
The dataset consists of the following information.
- number of pictures: 600 (1.png, 2.png, ... 600.png)
- format of pictures: PNG
- content of the original picture: a common label for reference of size, four flushing areas
- marked pictures: human-decided skin-flushing area (1-marked.png, ..., 600-marked.png)
- reference score (info.csv): human-decided relative size of the flushing area compared to the label (originally the hole in the reference Wang et al. (2020))