Detection of fibrosis regions in the lungs based on CT scans
| Parent link: | Информационные технологии в науке, управлении, социальной сфере и медицине: сборник научных трудов IV Международной научной конференции, 5-8 декабря 2017 г., Томск/ Национальный исследовательский Томский политехнический университет (ТПУ).— , 2017 Ч. 2.— 2017.— [С. 4-9] |
|---|---|
| Yazar: | |
| Müşterek Yazar: | |
| Diğer Yazarlar: | |
| Özet: | Заглавие с титульного экрана The main aim in the article was to provide an accurate, simple and fast algorithm that can increase the performance of the system and thereby the efficiency. Accurate results for lung images have not been accurate as the edges form in many diverse ways. Thereby, a universally applicable edge detection algorithm cannot comply with the purpose of detecting fibrosis. Thus by considering and furthermore introducing a deep convolutional neural network with pixel manipulation, the detection of fibrosis can be made easy, efficient and even accurate unlike the traditional learning structures. By implementing this we are free from extraction of features or even computation of multiple channels and thus suggesting a very straight forward method in terms of the detection and output accuracy. |
| Baskı/Yayın Bilgisi: |
2017
|
| Konular: | |
| Online Erişim: | http://earchive.tpu.ru/handle/11683/46959 |
| Materyal Türü: | Elektronik Kitap Bölümü |
| KOHA link: | https://koha.lib.tpu.ru/cgi-bin/koha/opac-detail.pl?biblionumber=626581 |
| Özet: | Заглавие с титульного экрана The main aim in the article was to provide an accurate, simple and fast algorithm that can increase the performance of the system and thereby the efficiency. Accurate results for lung images have not been accurate as the edges form in many diverse ways. Thereby, a universally applicable edge detection algorithm cannot comply with the purpose of detecting fibrosis. Thus by considering and furthermore introducing a deep convolutional neural network with pixel manipulation, the detection of fibrosis can be made easy, efficient and even accurate unlike the traditional learning structures. By implementing this we are free from extraction of features or even computation of multiple channels and thus suggesting a very straight forward method in terms of the detection and output accuracy. |
|---|