ChecKuko: Non-Invasive Early Detection of Iron Deficiency Nail Symptoms through Image Processing Using Faster R-CNN

Nails are biomarkers and can be a source of reference for one's nutritional status. Depending on anomalies in the nails' condition, such as their color, shape, and hardness, they may point out different nutritional deficiencies and can be correlated with certain diseases. The researchers focused on creating a standalone device that can detect iron deficiency symptoms through nail image analysis. The device uses a webcam to detect a fingernail and its condition. The result of the device was determined through image processing using Faster R-CNN object detection, which was trained to detect four (4) possible nail symptoms of iron deficiency. To make the device portable, the program along with the model was transferred to a Raspberry Pi, designed to work as a standalone even without external power. The device reported an overall predictive performance of 91.3% (precision of 87.5% and sensitivity of 95.45%), 85.71% accuracy, and 100% consistency in detecting nail aberrations that are symptoms of iron deficiency (
$\mathrm{n}=2$
participants). The results of the study determined that the device could be used as an aid to health centers, especially in remote and rural areas for prior testing of iron deficiency.