Abstract
Medical image classification plays an essential role in disease diagnosis and clinical treatment. More and more research efforts have been dedicated to the design of effective methods for medical image classification. As an effective framework, the capsule network (CapsNet) can realize translation equivariance. Lots of current research applies capsule networks in medical image analysis. In this paper, we propose an attentive octave convolutional capsule network (AOC-Caps) for medical image classification. In AOC-Caps, an AOC module is used to replace the traditional convolution operation. The purpose of the AOC module is to process and fuse the high- and low-frequency information in the input image simultaneously, and weigh the important parts automatically. Following the AOC module, a matrix capsule is used and the expectation maximization (EM) algorithm is applied to update the routing weights. The proposed AOC-Caps and comparative methods are tested on seven datasets, including PathMNIST, DermaMNIST, OCTMNIST, PneumoniaMNIST, OrganMNIST_Axial, OrganMNIST_Coronal, and OrganMNIST_Sagittal, which are from MedMNIST. In the experiments, baselines include the traditional CNN models, automated machine learning (AutoML) methods, and related capsule network methods. The experimental results demonstrate that the proposed AOC-Caps achieves better performance on most of the seven medical image datasets.
Highlights
In the matrix capsule network (CapsNet) [13], vectorized information is replaced by matrix capsules, and routing weights are updated by the expectation-maximization (EM) algorithm
We propose a novel capsule network, named the attentive octave convolutional capsule network (AOC-CapsNet) for medical image classification
We proposed a novel attentive octave convolutional capsule network (AOC-Caps) for medical image classification
Summary
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. With the development of deep learning technology, convolutional neural networks (CNNs) [1–7] have been widely applied in computer vision tasks, such as image classification [8,9], object detection [10], semantic segmentation [11], etc. We propose a novel capsule network, named the attentive octave convolutional capsule network (AOC-CapsNet) for medical image classification. The CBAM allows AOC-CapsNet to highlight critical local regions with rich semantic details utilized as distinguishable patterns, leading to a performance gain in the medical image classification task. Matrix capsule networks with different convolutional feature extraction layers are compared to determine which type of convolution layer is more suitable for the application of capsule networks in the medical image classification of MedMNIST. By combining the novel operation with capsule networks, we design an effective classification framework named AOC-CapsNet for medical image classification.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
