# Advantage of CNN
###### tags: `賴`
### [Localization and diagnosis framework for pediatric cataracts based on slit-lamp images using deep features of a convolutional neural network](https://docs.google.com/file/d/1oW7SR41t-VXsFB00g-IJIexF_UX9ZvDY/edit?ts=612da5e5)
Recently, deep learning convolutional neural network (CNN) methods have gained consid-
erable popularity since they offer superior performance in the field of image recognition tasks[18–22]. The CNN is an end-to-end learning model that avoids image pre-processing, requires no expert knowledge and extracts relevant high-level features directly from the raw image. The CNN architecture is inspired by the visual cortex of cats in Hubel’s and Wiesel’s early work[23].

### V [Diagnosis of Ophthalmic Diseases in Fundus Image Using various Machine Learning Techniques](https://docs.google.com/file/d/1gZXHUvMpGZxd_hVhVZebC816Xdo_4qvv/edit?ts=612da642)
**table 1** - looks like CNN accuracy better than the others

### [Cataract Detection Using Convolutional Neural Network with VGG-19 Model](https://docs.google.com/file/d/1cWYRs1rT2TovEAjSV2gKpAURXcYyspB6/edit?ts=612da6b6)
Deep learning (DL) is a subfield of machine learning. DL
uses multiple layers to extract higher levels of features from raw input. Because of its feature extraction ability, deep learning is highly used in medical image analysis. To be more specific Convolutional Neural Networks (CNN) are used for medical image analysis. Without the need for human interference, the CNN model extracts local features directly from the fundus images.

### V [Cataract Detection and Classification Systems Using Computational Intelligence: A Survey](https://docs.google.com/file/d/1siIi_fEBRuH9I7GpIN-4MZ4XRr87OcVt/edit?ts=612da755)
The strength of CNN’s lies in their weight sharing, exploiting the intuition that similar structures occur in different locations in an image. When seeing an image as a vectorized image, weights can be shared in such a way that it results in a convolution opera- tion. This drastically reduces the number of parameters that need to be learned and renders the network equivariant with respect to translations of the input.
Convolutional layers are typically alternated with pool- ing layers where pixel values of neighborhoods are aggre- gated using some permutation invariant function which induces a certain amount of translation invariance.
At the end of the convolutional stream of the network, fully-connected layers are usually added to act as classifi- cation layers, where weights are no longer shared. CNN’s are typically trained end-to-end in a completely supervised way.

### [A Hybrid Global-Local Representation CNN Model for Automatic Cataract Grading](https://docs.google.com/file/d/1CTQ7be_2HJELvV148RRuiq4q-51f9ADn/edit?ts=612ef707)
This paper uses convolutional neural networks (CNN) to learn useful features directly from input data, and deconvolution network method is employed to investigate how CNN characterizes cataract layer-by-layer. We found that compared to the global feature set, the detail vascular information, which is lost after multi-layer convolution calculation also plays an important role in cataract grading task.

### [A survey on deep learning in medical image analysis](https://drive.google.com/drive/u/1/folders/1FBOkHb_6KgEeOWgNBOe3INkRrpzpOQU0)

### [Detection and Classification of Lung Abnormalities by Use of Convolutional Neural Network (CNN) and Regions with CNN Features (R-CNN)](https://drive.google.com/drive/u/1/folders/1FBOkHb_6KgEeOWgNBOe3INkRrpzpOQU0)
Moreover, convolutional neural network (CNN) has
brought about breakthrough in pattern recognition of images including medical images. In usual CAD algorithms, designing an image-feature extractor is important. However, this task is difficult. On the other hand, a CAD algorithm by use of CNN does not necessarily require the image-feature extractor.