Maintaining filter structure: A Gabor-based convolutional neural network for image analysis

In image segmentation and classification tasks, utilizing filters based on the target object improves performance and requires less training data. We use the Gabor filter as initialization to gain more discriminative power. Considering the mechanism of the error backpropagation procedure to learn the data, after a few updates, filters will lose their initial structure. In this paper, we modify the updating rule in Gradient Descent to maintain the properties of Gabor filters. We use the Left Ventricle (LV) segmentation task and handwritten digit classification task to evaluate our proposed method. We compare Gabor initialization with random initialization and transfer learning initialization using convolutional autoencoders and convolutional networks. We experimented with noisy data and we reduced the amount of training data to compare how different methods of initialization can deal with these matters. The results show that the pixel predictions for the segmentation task are highly correlated with the ground truth. In the classification task, in addition to Gabor and random initialization, we initialized the network using pre-trained weights obtained from a convolutional Autoencoder using two different data sets and pre-trained weights obtained from a convolutional neural network. The experiments confirm the out-performance of Gabor filters comparing to the other initialization method even when using noisy inputs and a lesser amount of training data. •This is the first time, we use Gabor Filter with all its capacity. We change all 5 parameters to initialize the network’s weights.•We initialize all the convolutional filters in all the layers with 5 parameter Gabor function.•This is the first time of being able to keep the structure of the filter during training.•The proposed method maintains the Gabor structure in all layers we initialized the filters with.•The proposed method is general and can be applied to other segmentation and classification tasks.•The proposed method is fully automated and it does not involve any manual step to do the segmentation task.•The proposed method performs significantly better in presence of noise and when training with a lesser amount of training data.