CP-AnemiC (A Conjunctival Pallor) Dataset from Ghana
Description
Licensed biomedical scientists received proficiency training at the Centre for Research in Applied Biology, University of Energy and Natural Resources (UENR), Ghana, on the standardized procedure for capturing conjunctival images from children aged 6–59 months presenting at designated healthcare facilities during the study period. Data collection was conducted using an electronic data-collection system comprising KoboCollect v2021.2.4 (Massachusetts, USA), deployed on Samsung Galaxy Tab A7 tablets (Samsung Electronics, Vietnam). The electronic data-collection form was designed to capture relevant patient biodata and laboratory information, including age, gender, haemoglobin (Hb) concentration, and an assessment remark based on the Hb value obtained from the laboratory examination. Following the laboratory assessment, a photograph of the child's conjunctiva was captured and uploaded directly to the study database. This approach facilitated systematic data collection, storage, accessibility, and retrieval. For conjunctival image acquisition, the lower eyelid was gently retracted using the thumb while the index finger was used to stabilize the eyelid, thereby exposing the palpebral conjunctiva for imaging. All images were captured by trained laboratory personnel under ambient natural-light conditions using the standard 12-megapixel (12 MP) camera integrated into the electronic tablet. The camera flash/spotlight was disabled during image acquisition to minimize glare, reflections, and artificial illumination that could introduce variations in image quality and potentially affect the performance of subsequent image-processing and classification models. Following image acquisition, the region of interest (ROI) corresponding to the conjunctival area was extracted through a preprocessing procedure involving the Triangle Thresholding Algorithm and an entropy-based grayscale image-processing technique. This procedure was used to isolate the relevant conjunctival region while minimizing irrelevant image background and improving the consistency of the images used for subsequent analysis. Disclaimer This dataset is provided strictly for academic and research purposes and shall not be used for commercial purposes without appropriate authorization. Although reasonable measures were undertaken to ensure the quality, integrity, and anonymization of the dataset, users should be aware that minor anomalies, inconsistencies, or apparent image duplications may have occurred during image preparation, preprocessing, anonymization, file conversion, or dataset uploading. Researchers using the dataset are therefore encouraged to conduct appropriate quality-control checks before undertaking analysis or drawing conclusions from the data.
Files
Steps to reproduce
This dataset can be reused by other author(s) for academic purpose only and should be cited as such.