Makine öğrenimi kullanarak akciğer kanseri tespiti

This document is the result of our work in constructing a model for detecting lungs cancer issues carried out as part of the end of studies. To begin this, we began our work with an exciting study through exhaustive analysis and criticism, which allowed us to define the best model to implement. So, the idea of our project here, is to build a system that is capable of identifying the type of lung cancer in CT scan represented in images. The three classification classes here are benign, malignant and normal. The problem statement can be a little tricky, but we will learn how to tackle the problem and create a good performing system. First, we looked for the dataset that allowed us to build our model. Then, we structured it in the best way possible in order to feed our model by this data. We ran some data analysis methods to have a better understanding of our work. To finally move on to the implementation of our D.L Model. The results of our project are presented on the figure for the Training & Validation accuracy and loss, in the training phase with a maximum of 99% accuracy, unlike the validation phase with a maximum of 66% accuracy. In conclusion, the work carried out offers a very clear and pragmatic vision on the lung cancer detection.