Automatic segmentation of cerebral infarcts in follow-up computed tomography images with convolutional neural networks.
Renan Sales BarrosManon L TolhuisenAnna Mm BoersIvo JansenElena PonomarevaDiederik W J DippelAad van der LugtRobert J van OostenbruggeWim H van ZwamOlvert A BerkhemerMayank GoyalAndrew M DemchukBijoy K MenonPeter MitchellMichael D HillTudor G JovinAntoni DavalosBruce C V CampbellJeffrey L SaverYvo B W E M RoosKeith W MuirPhil WhiteSerge BracardFrancis GuilleminSilvia Delgado OlabarriagaCharles B L M MajoieHenk A MarqueringPublished in: Journal of neurointerventional surgery (2019)
Convolutional neural networks are valuable and accurate in the quantitative assessment of infarct volumes, for both subtle and severe hypodense infarcts in follow-up CT images. Our proposed three-CNNs approach strongly outperforms a more straightforward single CNN approach.
Keyphrases
- convolutional neural network
- deep learning
- computed tomography
- dual energy
- positron emission tomography
- image quality
- contrast enhanced
- magnetic resonance imaging
- machine learning
- acute myocardial infarction
- high resolution
- heart failure
- brain injury
- atrial fibrillation
- cerebral ischemia
- pet ct
- coronary artery disease