scholarly journals Optimizing the cervix cytological examination based on deep learning and dynamic shape modeling

2017 ◽  
Vol 248 ◽  
pp. 28-40 ◽  
Author(s):  
Afaf Tareef ◽  
Yang Song ◽  
Heng Huang ◽  
Yue Wang ◽  
Dagan Feng ◽  
...  
2012 ◽  
Author(s):  
Alison M. Pouch ◽  
Paul A. Yushkevich ◽  
Benjamin M. Jackson ◽  
Joseph H. Gorman III ◽  
Robert C. Gorman ◽  
...  

Author(s):  
Joris S. M. Vergeest ◽  
Chensheng Wang ◽  
Yu Song ◽  
Imre Horva´th

Automatic processing of shape information requires the selection of a representation form of shape. Shape modeling is based on a choice of shape type, which is the joint specification of representation form and a set of operations. In shape applications, such as shape design and shape optimization, it is not sufficient to maintain a static shape type. Depending on the specific needs during the application, i.e. depending on the modeling context, the appropriate shape type might be continuously varying. Programmed systems can handle static shape types relatively well. However, to support dynamic shape typing a number of fundamental problems need to be understood and solved. An approach to dynamic typing of freeform shapes is presented. Theoretical issues will be described and some concrete examples and initial experimental results will be presented.


2020 ◽  
Vol 7 (3) ◽  
pp. 296
Author(s):  
Rulisiana Widodo ◽  
Tessy Badriyah ◽  
Iwan Syarif

<p>Lung cancer is one of the most dangerous cases with the largest number of new cases in the world. The number of Lung Cancer in Indonesia is increasing rapidly every day until it is ranks 8th position in Southeast Asia, experiencing an increase in the last five years by 10.85 percent. This study aims to build a tool to detect lung cancer using the Deep Learning classification method with the Convolutional Neural Network (CNN) Algorithm. The tools that are made can be used for consideration in detecting from the results of cytological examinations, can be classified into normal (negative) and abnormal (positive) types of cancer. The experiment was carried out by performing hyperparameter optimization. The results show that the hyperparameter optimization has superior results compared to others, using the hyperparameter Gradient Boosted Regression Tree method. Experiments without hyperparameters give an accuracy value of 97%, while with the Gaussian Process it gives 98% accuracy and with a hyperparameter gradient boosted regression tree gives 99% accuracy, which is the best accuracy.</p><p><strong>Keywords</strong> : Lung Cancer, Cytological Examinations, Deep Learning, Convolutional Neural Network (CNN)</p><p><em>terbanyak di dunia. J</em><em>umlah penderita </em><em>Kanker Paru</em><em> di Indonesia semakin </em><em>hari semakin meningkat</em><em> pesat hingga menduduki urutan ke-8 di Asia Tenggara, mengalami peningkatan dalam lima tahun terakhir sebanyak 10.85 persen</em><em>. </em><em>Penelitian ini bertujuan untuk </em><em>me</em><em>mbangun </em><em>alat </em><em>pen</em><em>deteksi </em><em>kanker paru menggunakan metode klasifikasi</em><em> Deep Learning dengan Algoritma Convolutional Neural Network (CNN). Alat yang dibuat dapat digunakan sebagai pertimbangan dalam mendeteksi Kanker Paru</em><em> dari hasil pemeriksaan sitologi</em><em>, diklasifikasikan menjadi jenis</em><em> normal (negati</em><em>f</em><em>) dan abnormal (positi</em><em>f</em><em>) kanker</em><em>. </em><em>Percobaan dilakukan dengan melakukan optimasi hyperparameter. Hasil penelitian menunjukkan bahwa optimasi hyperparameter memiliki hasil yang lebih unggul yaitu dengan menggunakan metode hyperparameter Gradient Boosted Regression Tree. Percobaan tanpa hyperparameter memberikan nilai akurasi 97%, sedangkan dengan Gaussian Process memberikan akurasi 98% dan dengan hyperparameter Gradient Boosted Regression Tree memberikan akurasi terbaik yaitu 99%.</em></p><p><strong><em>Kata Kunci</em></strong> : <em>Kanker Paru, Pemeriksaan Sitologi, Deep Learning, </em><em>Convolutional Neural Network (CNN)</em><em></em></p>


Author(s):  
Stellan Ohlsson
Keyword(s):  

2019 ◽  
Vol 53 (3) ◽  
pp. 281-294
Author(s):  
Jean-Michel Foucart ◽  
Augustin Chavanne ◽  
Jérôme Bourriau

Nombreux sont les apports envisagés de l’Intelligence Artificielle (IA) en médecine. En orthodontie, plusieurs solutions automatisées sont disponibles depuis quelques années en imagerie par rayons X (analyse céphalométrique automatisée, analyse automatisée des voies aériennes) ou depuis quelques mois (analyse automatique des modèles numériques, set-up automatisé; CS Model +, Carestream Dental™). L’objectif de cette étude, en deux parties, est d’évaluer la fiabilité de l’analyse automatisée des modèles tant au niveau de leur numérisation que de leur segmentation. La comparaison des résultats d’analyse des modèles obtenus automatiquement et par l’intermédiaire de plusieurs orthodontistes démontre la fiabilité de l’analyse automatique; l’erreur de mesure oscillant, in fine, entre 0,08 et 1,04 mm, ce qui est non significatif et comparable avec les erreurs de mesures inter-observateurs rapportées dans la littérature. Ces résultats ouvrent ainsi de nouvelles perspectives quand à l’apport de l’IA en Orthodontie qui, basée sur le deep learning et le big data, devrait permettre, à moyen terme, d’évoluer vers une orthodontie plus préventive et plus prédictive.


Sign in / Sign up

Export Citation Format

Share Document