Comparison of Techniques Based on Computer Vision and Machine Learning for the Early Detection of Anemia From Nail Analysis
DOI:
https://doi.org/10.26439/ciis2020.5478Keywords:
computer vision, machine learning, anemia, detection, nailsAbstract
In Peru, anemia is a disease that affects more than 40% of the population, being common in both children and teenagers, and prevailing in pregnant women and children under 2 years of age, which seriously compromises their development. To diagnose anemia, it is necessary to perform blood tests to determine hemoglobin levels. However, most hospitals do not have the proper equipment to conduct the tests, which causes delays in the delivery of diagnoses. The objective of this research is to compare techniques based on computer vision and machine learning for the early detection of anemia from nail analysis, so that doctors can use such analysis as support in the early detection of this disease. With a timely diagnosis, it will be possible to prevent patients from suffering the disease in its different stages, especially for those who are at a chronic stage where the consequences are serious, since anemia can be a sign of an underlying disease. Several experiments were carried out and the best results were the following: accuracy 0.989, precision 0.98, recall 0.98 and F1-score 0.98. VGG19 architec ture was used as a feature extractor in combination with the support vector machine (SVM) classifier. The research has shown that anemia can be detected without a blood test with quic ker and reliable results.
Downloads
References
Banerji, S., Sinha, A., y Liu, C. (2013). HaarHOG: Improving the HOG Descriptor for Image Classification [Sesión de conferencia]. IEEE International Conference on Systems, Man, and Cybernetics (pp. 4276-4281). 10.1109/SMC.2013.729
Bay H., Tuytelaars T., Van Gool L. (2006). SURF: Speeded Up Robust Features. Computer Science, 3951. https://doi.org/10.1007/11744023_32
De Benoist, B., Cogswell, M., Egli, I., y McLean, E. (2008). Worldwide Prevalence of Anaemia 1993-2005. WHO Global Database of Anaemia.
Frossard, D. (2016). VGG en TensorFlow. [Figura]. https://www.cs.toronto.edu/~frossard/post/vgg16/
Haykin, S. (2008). Neuronal Networks and Learning Machines. Pearson Prentice Hall. https://cours.etsmtl.ca/sys843/REFS/Books/ebook_Haykin09.pdf
Instituto Nacional de Estadística e Informática. (2018). Resultados de la Encuesta Demográfica y de Salud Familiar-ENDES 2018. https://www.ipe.org.pe/portal/wp-content/uploads/2018/06/INEI-Encuesta-Demográfica-y-de-Salud-Familiar-ENDES.pdf
Instituto Nacional de Estadística e Informática. (2018b). Pobreza y gasto social. https://www.inei.gob.pe/estadisticas/indice-tematico/sociales/
Lowe, D. (2004). Distinctive Image Features from Scale-Invariant Keypoints. International Journal of Computer Vision, 60(2), 91-110. 10.1023/B:VISI.0000029664.99615.94
Mannino, R. G., Myers, D. R., Tyburski, E. A., Caruso, C., Boudreaux, J., Leong, T., Clifford, G. D., y Lam, W. A. (2018). Smartphone app for non-invasive detection of anemia using only patient-sourced photos. Nature Communications, 9(1), 4924. https://doi.org/10.1038/s41467-018-07262-2
MedlinePlus. (6 de agosto del 2020). Anemia. National Library of Medicine. https://medlineplus.gov/anemia.html
Ministerio de Salud. (2015). Necesidad de médicos especialistas en establecimientos de salud del Sector Salud, Perú 2010. http://bvs.minsa.gob.pe/local/MINSA/1613.pdf
Sharma, V., y Shrivastava, A. (2015). System for Disease Detection by Analyzing Finger Nails Color and Texture. International Journal of Advanced Engineering Research and Science, 5.
Thaver, I. H., y Baig, L. (1994). Anaemia in Children: Part I. Can Simple Observations by Primary Care Provider Help in Diagnosis? Journal of Pakistan Medical Association 44, 282-284.
Torres, J. (2018). Deep Learning. Introducción práctica con Keras (1.ª ed.). https://torres.ai/deep-learning-inteligencia-artificial-keras
