%0 Journal Article %1 noauthororeditor %D 2020 %E Saif, Amin %J International Journal of Advanced Networking and Applications %K Cancer Convolutional Deep Detection, Learning Learning, Lung Networks, Neural Transfer myown %N 02 %P 4549-4554 %R 10.35444/IJANA.2020.12202 %T Multi Paths Technique On Convolutional Neural Network For Lung Cancer Detection Based On Histopathological Images %U https://www.ijana.in/download12-2-2.php?file=V12I2-2.pdf %V 12 %X Lung cancer is the leading cancer cause of death and it's survival rate is very small. An early diagnosis is a good solution to increase the survival rate for lung cancer. To diagnose lung cancer using deep learning we present a convolutional neural network to diagnose three types of lung cancer(Adenocarcinoma, Benign and Squamous) based on histopathological images. The proposed model consists of a main path and three sub-paths. The main path works to extract the small features and creates feature maps at low-level. As for the sub-paths is responsible for transferring the medium and high levels feature maps to fully connected layers to complete the classification process, also the VGG16 was prepared to compare it with the performance of the proposed. After training the models and testing them on 1500 images, we obtained an overall accuracy of 98.53% for the proposed model and 96.67% for the VGG16 model. The proposed model achieved a sensitivity of 97.4%, 99.6% and 98.6% for Adenocarcinoma, Benign and Squamous respectively. We can say that it is not always necessary for the used to be very deep to diagnose histopathological images, and the most important thing is to create a sufficient number of feature maps at different levels.

@article{noauthororeditor, abstract = {Lung cancer is the leading cancer cause of death and it's survival rate is very small. An early diagnosis is a good solution to increase the survival rate for lung cancer. To diagnose lung cancer using deep learning we present a convolutional neural network to diagnose three types of lung cancer(Adenocarcinoma, Benign and Squamous) based on histopathological images. The proposed model consists of a main path and three sub-paths. The main path works to extract the small features and creates feature maps at low-level. As for the sub-paths is responsible for transferring the medium and high levels feature maps to fully connected layers to complete the classification process, also the VGG16 was prepared to compare it with the performance of the proposed. After training the models and testing them on 1500 images, we obtained an overall accuracy of 98.53% for the proposed model and 96.67% for the VGG16 model. The proposed model achieved a sensitivity of 97.4%, 99.6% and 98.6% for Adenocarcinoma, Benign and Squamous respectively. We can say that it is not always necessary for the used to be very deep to diagnose histopathological images, and the most important thing is to create a sufficient number of feature maps at different levels.}, added-at = {2021-01-27T11:31:10.000+0100}, biburl = {https://www.bibsonomy.org/bibtex/23636553da850018526c5c27d78e80d3e/editor.ijana}, doi = {10.35444/IJANA.2020.12202}, editor = {Saif, Amin}, interhash = {ee0c9460fe66d2551efb7a418660c3cc}, intrahash = {3636553da850018526c5c27d78e80d3e}, issn = {0975-0282}, journal = {International Journal of Advanced Networking and Applications}, keywords = {Cancer Convolutional Deep Detection, Learning Learning, Lung Networks, Neural Transfer myown}, language = {English}, month = oct, number = 02, pages = {4549-4554}, timestamp = {2021-01-27T11:31:10.000+0100}, title = {Multi Paths Technique On Convolutional Neural Network For Lung Cancer Detection Based On Histopathological Images}, url = {https://www.ijana.in/download12-2-2.php?file=V12I2-2.pdf}, volume = 12, year = 2020 }