2 research outputs found
Improvement of Preprocessing for Spiral and Wave Handwriting Image Classification Using DenseNet-169
No full textParkinson\u27s disease (PD) is the second most common neurodegenerative disorder, impacting over 10 million people. Key symptoms include slowed limb movements, difficulty writing, and involuntary tremors. Tremor is the first motor symptom of Parkinson\u27s disease, seen in about 75% of patients. Neurologists assess tremors through various non-invasive tests. This may involve assessing handwriting and spiral drawing. The analysis is still performed manually by neurologists, which can introduce subjectivity. Applications using computer vision techniques should be developed to classify handwriting as healthy or tremor-affected, aiding neurologists in making more objective decisions. DenseNet-169 can classify spiral and wave images in tremor and non-tremor classes with the addition of preprocessing obtained a training accuracy of 100% while the system test accuracy is 93% while without preprocessing, the system accuracy is 81%
Improvement of Preprocessing for Spiral and Wave Handwriting Image Classification Using DenseNet-169
No full textParkinson\u27s disease (PD) is the second most common neurodegenerative disorder, impacting over 10 million people. Key symptoms include slowed limb movements, difficulty writing, and involuntary tremors. Tremor is the first motor symptom of Parkinson\u27s disease, seen in about 75% of patients. Neurologists assess tremors through various non-invasive tests. This may involve assessing handwriting and spiral drawing. The analysis is still performed manually by neurologists, which can introduce subjectivity. Applications using computer vision techniques should be developed to classify handwriting as healthy or tremor-affected, aiding neurologists in making more objective decisions. DenseNet-169 can classify spiral and wave images in tremor and non-tremor classes with the addition of preprocessing obtained a training accuracy of 100% while the system test accuracy is 93% while without preprocessing, the system accuracy is 81%
