A Progressively-Trained Scale-Invariant and Boundary-Aware Deep Neural Network for the Automatic 3D Segmentation of Lung Lesions.
Bo ZhouAntong ChenRandolph CrawfordBelma DogdasGregory GoldmacherPublished in: WACV (2019)
Keyphrases
- scale invariant
- neural network
- multiscale
- lesion segmentation
- natural images
- scale space
- keypoints
- image analysis
- affine invariant
- multilayer perceptron
- scale selection
- image segmentation
- scale invariance
- region features
- back propagation
- segmentation algorithm
- medical images
- texture images
- level set
- segmentation method
- image structure
- skin lesion
- ct images
- local binary pattern
- edge detection
- interest point detectors
- pattern recognition
- invariant features
- computer aided diagnosis
- medical imaging
- pulmonary nodules
- training set
- translation and scale invariant
- active contours
- lung nodules
- input image
- scale invariant features