Computer aided 3d modeling of malign tumors in lungs for the performance estimation of radiotherapy

Uncontrolled and rapidly growing malignant tumors in the lungs cause lung cancer disease. Lung cancer is the most common cancer disease worldwide. It is the deadliest of all cancer diseases. The growth and spread of malignant tumors causes respiratory function to become difficult and ultimately dysfunctional. Malignant tumors have the potential to spread to other tissues or organs and render these organs and systems dysfunctional. The diagnosis of lung cancer is made with the help of images obtained by imaging technologies such as Computed Tomography (CT), MRI and biopsy. In addition to surgical treatment in lung cancer disease, radiotherapy or chemotherapy treatment is given when necessary. The aim of radiotherapy treatment is to stop the growth of the uncontrolled malignant tumor and to destroy or shrink the mass. Monitoring the response of tumors causing lung cancer to radiotherapy treatment gives information about the success of the treatment. Computer Aided Diagnostic Systems(CADs) based on Machine learning are used to monitor and interpret the change and response of detected tumors during radiotherapy treatment. In this way, treatment results can be examined accurately and in detail. In this study, the tumor mass was segmented from the lung and modeled in three dimensions. Tissue, shape and statistical features of the segmented tumor image were obtained. In addition, the demographic data of the patients and the pathological results of the tumor were added to the data set. The success of patients in radiotherapy treatment was estimated by applying classification methods based on machine learning and deep learning on the obtained data set. Among Machine Learning Methods, the highest accuracy result was obtained with the Support Vector Machines-based Kernel method at 75.26%. In the classification performed with deep learning models, an accuracy rate of 79.31% was achieved with the VGG16 model.