Automatic cervical lymphadenopathy segmentation from CT data using deep learning - 16/11/21
, Diana L.F. Cabrera a, b, Nicolas Gogin a, Loic Gaillandre c, Geoffrey Rico d, Jules Zhang-Yin e, Mickael Elhaik f, François Bidault g, h, Imad Bousaid f, Nathalie Lassau g, h
Bienvenue sur EM-consulte, la référence des professionnels de santé.
Article gratuit.
Connectez-vous pour en bénéficier!
Highlights |
• | Despite limited available data and partial annotations, a convolutional neural network-based approach provides encouraging results for the segmentation of cervical lymphadenopathy. |
• | The quality of the provided ground-truth proved to be crucial for the convolutional neural network segmentation performance. |
• | A lead for future studies could be to distinguish the segmentation from the detection task. |
Abstract |
Purpose |
The purpose of this study was to develop a fast and automatic algorithm to detect and segment lymphadenopathy from head and neck computed tomography (CT) examination.
Materials and methods |
An ensemble of three convolutional neural networks (CNNs) based on a U-Net architecture were trained to segment the lymphadenopathies in a fully supervised framework. The resulting predictions were assessed using the Dice similarity coefficient (DSC) on examinations presenting one or more adenopathies. On examinations without adenopathies, the score was given by the formula M/(M+A) where M was the mean adenopathy volume per patient and A the volume segmented by the algorithm. The networks were trained on 117 annotated CT acquisitions.
Results |
The test set included 150 additional CT acquisitions unseen during the training. The performance on the test set yielded a mean score of 0.63.
Conclusion |
Despite limited available data and partial annotations, our CNN based approach achieved promising results in the task of cervical lymphadenopathy segmentation. It has the potential to bring precise quantification to the clinical workflow and to assist the clinician in the detection task.
Le texte complet de cet article est disponible en PDF.Keywords : Deep learning, Lymphadenopathy, Tomography, X-ray computed, Image processing, Computer-assisted, Artificial intelligence
Abbreviations : AI, CNN, CT, DL, DSC, HU, MIP, MRI, SD
Plan
Vol 102 - N° 11
P. 675-681 - novembre 2021 Retour au numéro
Article précédent
- Three artificial intelligence data challenges based on CT and ultrasound
- Nathalie Lassau, Imad Bousaid, Emilie Chouzenoux, Antoine Verdon, Corinne Balleyguier, François Bidault, Elie Mousseaux, Sana Harguem-Zayani, Loic Gaillandre, Zoubir Bensalah, Isabelle Doutriaux-Dumoulin, Michèle Monroc, Audrey Haquin, Luc Ceugnart, Florence Bachelle, Mathilde Charlot, Isabelle Thomassin-Naggara, Tiphaine Fourquet, Héloise Dapvril, Joseph Orabona, Foucauld Chamming's, Mickael El Haik, Jules Zhang-Yin, Marc-Samir Guillot, Mickaël Ohana, Thomas Caramella, Yann Diascorn, Jean-Yves Airaud, Philippe Cuingnet, Umit Gencer, Littisha Lawrance, Alain Luciani, Anne Cotten, Jean-François Meder
- Automatic coronary artery calcium scoring from unenhanced-ECG-gated CT using deep learning
- Nicolas Gogin, Mario Viti, Luc Nicodème, Mickaël Ohana, Hugues Talbot, Umit Gencer, Magloire Mekukosokeng, Thomas Caramella, Yann Diascorn, Jean-Yves Airaud, Marc-Samir Guillot, Zoubir Bensalah, Caroline Dam Hieu, Bassam Abdallah, Imad Bousaid, Nathalie Lassau, Elie Mousseaux
Bienvenue sur EM-consulte, la référence des professionnels de santé.