http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
( Rifaldy Fajar ),( Nana Indri Kurniastuti ),( Dewi Mustika Sari ),( Prihantini Jupri ) 대한결핵 및 호흡기학회 2020 대한결핵 및 호흡기학회 추계학술대회 초록집 Vol.128 No.0
This study aims to explain the procedure, application, and accuracy of Recurrent Neural Network modeling and Recurrent Neuro-Fuzzy modeling for lung cancer nodule classification from lung photo images. Recurrent Neural Network and Recurrent Neuro-Fuzzy modeling steps are defining input and target variables, dividing data into training data and testing data, data normalization, designing the best model, and data denormalization. The input variable used is the feature of the lung photo image extraction, while the target tissue is the description of the condition from the lung photo image, namely normal lung, benign lung tumor, or malignant lung tumor. The image extraction step begins with an image transformation, namely from the original lung image (gray image) to a binary image, followed by extracting the transformed image using the Gray Level Co-occurrence Matrix Method. The design steps for the best Recurrent Neuro-Fuzzy model begin with the design steps for the best Recurrent Neural Network model followed by data clustering steps using the Fuzzy C-Means Method, learning Recurrent Neural Networks related to antecedents to fuzzy inference rules and consequent fuzzy inference rules, and Simplifying the consequent part by eliminating input and finding the consequent coefficient value of each cluster using the Least Square Estimator (LSE) Method. The Results obtained indicate that the classification of lung cancer nodules using the Recurrent Neural Network model gives better Results than the Recurrent Neuro-Fuzzy model. The sensitivity, specificity, and accuracy values of the Recurrent Neural Network model were 94%, 56%, and 81.33% for training data and 80%, 40%, and 64% for testing data, respectively. The Conclusion in this study is that the Recurrent Neural Network model is able to classify quite well compared to the Recurrent Neuro-Fuzzy model.