نوع مقاله : مقاله پژوهشی
نویسندگان
پژوهشکده کاربرد پرتوها، پژوهشگاه علوم و فنون هستهای، سازمان انرژی اتمی، تهران، ایران
چکیده
کلیدواژهها
عنوان مقاله [English]
نویسندگان [English]
This paper presents a novel artificial neural network-based method for the classification of lung CT images to enable early diagnosis of lung cancer. For this purpose, the entire lung is segmented from the CT images, and statistical parameters such as mean, standard deviation, skewness, kurtosis, fifth-order central moment, and sixth-order central moment are computed from the segmented images. A feedforward backpropagation neural network is employed for the classification process. The results show that among the existing training functions for backpropagation neural networks, the best classification accuracy of 91.1% is achieved using the Traingdx training function. Additionally, two novel training functions are introduced in this paper, one of which achieved an accuracy of 93.3%, 100% detection rate, 91.4% sensitivity, and a mean squared error of 0.998, while the other achieved an accuracy of 93.3% and a mean squared error of 0.0942. Overall, the early diagnosis of lung cancer using artificial neural networks is a promising approach to increase the survival rate of patients.