ALAT PENDETEKSI DINI ANEMIA SECARA NON INVASIVE

Anemia is a disease where the hemoglobin level in the blood is lower than the normal limit, which is mostly caused by consuming foods that do not contain iron. Based on World Health Organization (WHO) anemia prevalence data collected from 1993 to 2005, it is predicted that around 1.6 billion people (a quarter of the world’s population) suffer from anemia. Measuring Hb (Hemoglobin) levels to detect anemia is usually done invasively by taking blood samples. However, this method is considered less efficient. This is because when the blood sampling process is carried out by injuring one of the patient’s fingers, it can cause pain to the patient. Detecting anemia non-invasively through the eye image segmentation method using the palpebral conjunctiva is a method that does not require painful procedures for the patient. The method used in this research consists of several stages, including hardware design and software design. This research uses an image processing method in the form of a CNN with RESNET-50 and YOLO architecture. The steps include image preprocessing, training process, testing process and displaying the results in web form. This study also utilized the use of the ESP32 Cam in the process of taking images of the eye conjunctiva. The accuracy value for the CNN method with Resnet architecture on test data using a confusion matrix is 0.76, while the accuracy value for the YOLO method on test data is 0.98. The scope of system testing is divided into two, namely the scope of functional testing and the scope of component performance testing. Testing related to accuracy is also carried out by carrying out blood tests (invasive) to determine hemoglobin in individuals.