Pneumonia Classification using Gabor-Convolutional Neural Networks and Image Enhancement

Autor: Muhammad Rifal Alfarizy, Agus Eko Minarno, Syaifuddin Syaifuddin, Agus Hendryawan, Yuda Munarko
Rok vydání: 2021
Předmět:
Zdroj: 2021 9th International Conference on Information and Communication Technology (ICoICT).
DOI: 10.1109/icoict52021.2021.9527427
Popis: Pneumonia is acknowledged as a respiratory disease caused by bacterial and, viral or fungal infections and has a high mortality rate. Identification of pneumonia is typically performed with Chest X-Ray image, but hampered by other lung problems that have been experienced by the patient. Therefore, this study proposes a Convolutional Neural Networks method by adding a Gabor filter and an Image Enhancement Preprocessing technique. The application of the Gabor filter obtains the best accuracy with a value of 94.4% and a loss of 44%, while Image Enhancement obtains an accuracy of 87.8% and the best loss of 35.8%. Combining the Gabor filter and Image Enhancement obtains better accuracy and loss of 93.9% and 40% than utilizing these methods separately.
Databáze: OpenAIRE