| Resumen |
Parkinson’s disease is a neurodegenerative condition for which the early detection is a very challenging activity for the medical community. Although traditional methods for Parkinson’s disease diagnosis involve the use of EEG (electroencephalographic) activity, previous works have proposed to analyze sketches of guided spirals and waves drawn by a patient versus those drawn by healthy people. In this work, we made use of the same dataset, employing data augmentation techniques for enriching the diversity of the images. Besides, architectures such as ResNet50 and VGG19 demonstrated promising results using transfer learning. Results reported in this manuscript are comparable with those of the stateof-the-art, but also have the potential to improve the accuracy in the near future. |
