Conceptualising a channel-based overlapping CNN tower architecture for COVID-19 identification from CT-scan images

Abstract

Convolutional Neural Network (CNN) has been employed in classifying the COVID cases from the lungs’ CT-Scan with promising quantifying metrics. However, SARS COVID-19 has been mutated, and we have many versions of the virus B.1.1.7, B.1.135, and P.1, hence there is a need for a more robust architecture that will classify the COVID positive patients from COVID negative patients with less training. We have developed a neural network based on the number of channels present in the images. The CNN architecture is developed in accordance with the number of the channels present in the dataset and are extracting the features separately from the channels present in the CT-Scan dataset. In the tower architecture, the first tower is dedicated for only the first channel present in the image; the second CNN tower is dedicated to the first and second channel feature maps, and finally the third channel takes account of all the feature maps from all three channels. We have used two datasets viz. one from Tongji Hospital, Wuhan, China and another SARS-CoV-2 dataset to train and evaluate our CNN architecture. The proposed model brought about an average accuracy of 99.4%, F1 score 0.988, and AUC 0.99.

A framework to distinguish healthy/cancer renal CT images using the fused deep features

“…The proposed research study also implements well-known pretraining procedures, such as VGG16, VGG19, ResNet50, ResNet101, DenseNet121, and DenseNet201, to examine the KC in RCT slices. The complete evidence concerning the preferred schemes can be found in the literature (28)(29)(30)(31)(32), and in this study, these schemes are considered along with chosen binary classifiers. The following initial parameters are assigned for these models: learning rate = 1×10 −5 , training with linear dropout rate (LDR), Adam optimization, ReLu activation, total iteration = 2000, total epochs = 150, and classification with a SoftMax unit using a 5-fold cross-validation.…”

Distributed autoencoder classifier network for small‐scale and scattered COVID‐19 dataset classification

“…The final results show that the 7L-CNN-CD model, 14-way data augmentation, and stochastic pooling method have a good effect on improving the precision. Wang et al 13 16 designed the CNN tower architecture to extract features from each channel. The features extracted from specific channels are connected in series to CNN tower, and the features extracted from specific channels are connected in series to generate an intermediate feature map.…”

“…Finally, in the COVID‐19 classification task, the classification accuracy of this model reaches 96.7%. Tiwari et al 16 designed the CNN tower architecture to extract features from each channel. The features extracted from specific channels are connected in series to CNN tower, and the features extracted from specific channels are connected in series to generate an intermediate feature map.…”