DRU‐net: a novel U‐net for biomedical image segmentation

A Stacked Generalization U-shape network based on zoom strategy and its application in biomedical image segmentation

Image segmentation is the first step in image analysis and pattern recognition, and it is one of the most difficult tasks in image processing, and determines the quality of the final result of analysis. Image segmentation requires the improvement because most image segmentation solution is problem based. Biomedical image segmentation is different than normal image segmentation. Because normal image is taken from camera which is free of noise. While biomedical image is taken from MRI, CT, Microscopic Instrument or X-ray which is it noisy and fuzzy. Medical image segmentation methods generally have restrictions because medical images have very similar gray level and texture among the interested. While extracting region of interest in biomedical image, region of interest can avoid the processing of irrelevant image point. Biomedical Image segmentation is necessary when we want the computer to make decision, surgical planning and robotically assisted surgical invention. Biomedical image segmentation is nothing but the classification of similar pixel and their following extraction into separate segment. There are different methods to do the Biomedical image segmentation but they are slow. he method which is described in this paper is fast and efficient. This paper gives comparative research of different method of each of category. The method described in the paper is for fast biomedical image segmentation

Many biomedical image analysis applications require segmentation. Convolutional neural networks (CNN) have become a promising approach to segment biomedical images; however, the accuracy of these methods is highly dependent on the training data. We focus on biomedical image segmentation in the context where there is variation between source and target datasets and ground truth for the target dataset is very limited or non-existent. We use an adversarial based training approach to train CNNs to achieve good accuracy on the target domain. We use the DRIVE and STARE eye vasculture segmentation datasets and show that our approach can significantly improve results where we only use labels of one domain in training and test on the other domain. We also show improvements on membrane detection between MIC-CAI 2016 CREMI challenge and ISBI2013 EM segmentation challenge datasets.

Left ventricle (LV) segmentation is crucial to assess left ventricle global function. U-Net; a Convolutional Neural Network (CNN); boosted the performance of many biomedical image segmentation tasks. In LV segmentation, U-Net suffered from accurately extracting small objects such as the apical short-axis slices. In this paper, we propose a fully automated left ventricle segmentation method for both short-axis and long-axis views. The proposed model utilizes U-Net architecture and Multi-Gate input block to enhance the performance by aggregating multi-scale features and adding different vision scopes providing a more robust model against scale variance of objects within the images. The proposed approach was validated against left ventricle segmentation challenge (LVSC) and Automated cardiac diagnosis challenge (ACDC). For LV myocardium segmentation, the proposed approach achieved mean dice index 82% on LVSC and 90% and 91% for end-diastole (ED) and end-systole (ES) time frames respectively on ACDC outperforming other published methods measurements at ED. For LV blood pool segmentation, mean dice index was 97% and 92.3% for the ED and ES time frames using ACDC, outperforming other methods’ ED measurements.

Biomedical images have complex tissue structures, and there are great differences between images of the same part of different individuals. Although deep learning methods have made some progress in automatic segmentation of biomedical images, the segmentation accuracy is relatively low for biomedical images with significant changes in segmentation targets, and there are also problems of missegmentation and missed segmentation. To address these challenges, we proposed a biomedical image segmentation method based on dense atrous convolution. First, we added a dense atrous convolution module (DAC) between the encoding and decoding paths of the U-Net network. This module was based on the inception structure and atrous convolution design, which can effectively capture multi-scale features of images. Second, we introduced a dense residual pooling module to detect multi-scale features in images by connecting residual pooling blocks of different sizes. Finally, in the decoding part of the network, we adopted an attention mechanism to suppress background interference by enhancing the weight of the target area. These modules work together to improve the accuracy and robustness of biomedical image segmentation. The experimental results showed that compared to mainstream segmentation networks, our segmentation model exhibited stronger segmentation ability when processing biomedical images with multiple-shaped targets. At the same time, this model can significantly reduce the phenomenon of missed segmentation and missegmentation, improve segmentation accuracy, and make the segmentation results closer to the real situation.

As a popular convolutional neural network, U-net developed by Ronneberger et al. at the University of Freiburg, Germany, has been widely applied in biomedical image segmentation. The network is an extension of the fully convolutional network with multi-resolution encoder and decoder structures and skip layers at each resolution. It can achieve effective segmentation results with a limited training dataset. Initially developed for multi-channel 2D images, U-net has been extended to segmenting 3D biomedical images and even multi-channel or serial 3D images by using 3D convolution kernels, down-sampling, and up-sampling blocks, and various activation functions. In this chapter, using the lung CT segmentation challenge datasets, the performance of organ segmentation is compared between 2D and 3D U-nets trained at different spatial resolutions. The network structures used herein provide intuitive, hands-on experiences about their advantages and disadvantages for multiple organ segmentation. The comparative study could be helpful to facilitate research and development of deep learning-based biomedical image segmentation algorithms.

Transfer learning can realize cross-domain fault diagnosis of rotating machinery, where the model trained on many labeled samples collected in one working condition can be transferred to insufficient samples collected in the target working condition. Currently, the data features cannot be completely extracted by existing methods when the data distribution gap of the samples collected in different working conditions is quite large. In order to fully extract the data features of rotating machinery to achieve cross-domain fault diagnosis, this paper investigated a cross-domain fault diagnosis model of rotating machinery based on graph feature extraction. The proposed method can realize unsupervised fault diagnosis on rotating machinery running under different working conditions by extracting the numerical and structural features of source and target domains. First of all, data features with large data distribution gaps need to be fully extracted, so a convolutional network based on a deformable convolutional network (De-conv) is designed to extract the features with large differences in data distribution under various working conditions. Secondly, features are extracted based on a convolutional neural network for data values in existing domain adaptation (DA) methods while the structure features of machine monitoring data are ignored. Therefore, a composite spectral-based graph convolutional network is designed to extract structural features of data. Thirdly, fully extracted features are input into a universal DA network to achieve cross-domain fault diagnosis of unknown faults in rotating machinery under changing working conditions. Finally, a benchmarking data set and a data set collected from a practical experimental platform are used to verify the effectiveness of the proposed model, and the results show that it is more suitable for cross-domain fault diagnosis of rotating machinery than other comparison models.

With the rapid growth of medical and biomedical image data, energy-efficient solutions for analyzing such image data that can be processed fast and accurately on platforms with low power budget are highly desirable. This paper uses segmenting glial cells in brain microscopy images as a case study to demonstrate how to achieve biomedical image segmentation with significant energy saving and minimal comprise in accuracy. Specifically, we design, train, implement, and evaluate Fully Convolutional Networks (FCNs) for biomedical image segmentation on IBM's neurosynaptic DNN processor - TrueNorth (TN). Comparisons in terms of accuracy and energy dissipation of TN with that of a low power NVIDIA TX2 mobile GPU platform have been conducted. Experimental results show that TN can offer at least two orders of magnitude improvement in energy efficiency when compared to TX2 GPU for the same workload.

It is described a method of biomedical images segmentation by discrimination of textures based on their morphological spectra. Basic notions concerning morphological spectra are given. Their properties making possible to characterize basic morphological structures independently on spatial orientation or shifts of the analyzed specimens are described. It is shown that spectral components can be chosen and used in aggregated form so as to make discrimination of textures invariant with respect to scale changing and to basic geometrical image transformations. Analysis of two types of biomedical images: aorta tissue and pancreas tissue, based on comparison of histograms of selected spectral components values illustrate the methods presented in the paper.

Recently, the introduction of Convolutional Neural Network (CNNs) has advanced the way of solving image segmentation tasks. Semantic image segmentation has considerably benefited from employing various CNN models. The most widely used network in this field is U-Net and its different variations. However, these models require significant number of trainable parameters, floating-point operations per second, and great computational power to be trained. These factors make real-time semantic segmentation in low powered devices very hard. Therefore, in the present paper, we aim to modify particular aspects of the U-Net model to improve its performance through developing a fast U-Net (FU-Net) relying on bottleneck convolution layers in the contraction and expansion paths of the model. The proposed model can be utilized in semantic segmentation applications even on the devices with limited computational power and memory by ensuring the state-of-the-art performance. The amount of memory required by the proposed model is reduced by 23 times when compared with the original U-Net. Moreover, the modifications allowed achieving better performance. In conducted experiments, we assessed the performance of the proposed model on two biomedical image segmentation datasets, namely 2018 Data Science Bowl and ICIS 2018: Skin Lesion Analysis Towards Melanoma Detection. FU-Net demonstrated the state-of-the-art results in biomedical image segmentation, requiring the number of trainable parameters reduced by eight times compared with the original U-Net model. In addition, using bottleneck layers decreased the number of computations, resulting in nearly 30% speed-up at the training, validation and test stages. Furthermore, despite relying on fewer parameters FU-Net achieved a slight improvement of the performance in terms of pixel accuracy, Jaccard index, and dice coefficient evaluation metrics.

In the real-time test fault diagnosis algorithm based on deep learning, it is difficult to guarantee that the training and testing data come from the same time series distribution. Inconsistent distribution will lead to a decline in diagnostic performance. In addition, the convolutional neural network is limited by the fixed shape of its convolution kernel, which makes it difficult to fully extract the spatial constraint features between fault data. To solve the above problems, this paper proposes a bearing fault diagnosis method based on inverted Mel-scale frequency cepstrum coefficients and deformable convolution networks. The core of traditional Mel-scale frequency cepstrum coefficients is to construct a non-uniformly distributed frequency-domain filter bank. It is characterized by the dense distribution of low-frequency regions and the sparse distribution of high-frequency regions. Considering that the features that can well characterize fault information are concentrated in the high-frequency part, we reconstruct the traditional Mel-scale frequency cepstrum coefficients filter bank and propose a feature extraction method of inverted Mel-scale frequency cepstrum coefficients. This method can obtain the frequency-domain characteristics of bearing vibration signals, highlight the fault information contained in the high-frequency region, and reduce the influence of time series distribution inconsistency between training samples and testing samples on the diagnosis accuracy. In order to further improve the spatial discrimination between different fault categories, the deformable convolution networks model is introduced to extract the spatial distribution information of fault features and improve the accuracy of fault diagnosis. Finally, two public data sets and data from an experimental platform verify that the method can achieve high-precision fault diagnosis, and that inverted Mel-scale Frequency cepstrum coefficients are effective in solving the problem of inconsistent distribution.

Evaluation of the benchmark datasets for testing the efficacy of deep convolutional neural networks

Intense interest in applying convolutional neural networks (CNNs) in biomedical image analysis is wide spread, but its success is impeded by the lack of large annotated datasets in biomedical imaging. Annotating biomedical images is not only tedious and time consuming, but also demanding of costly, specialty-oriented knowledge and skills, which are not easily accessible. To dramatically reduce annotation cost, this paper presents a novel method called AIFT (active, incremental fine-tuning) to naturally integrate active learning and transfer learning into a single framework. AIFT starts directly with a pre-trained CNN to seek "worthy" samples from the unannotated for annotation, and the (fine-tuned) CNN is further fine-tuned continuously by incorporating newly annotated samples in each iteration to enhance the CNN's performance incrementally. We have evaluated our method in three different biomedical imaging applications, demonstrating that the cost of annotation can be cut by at least half. This performance is attributed to the several advantages derived from the advanced active and incremental capability of our AIFT method.

Multiobjective optimization methods in image analysis are one of the active research domains in the current years. These methods are used for the decision-making process in case of image segmentation. Multiobjective techniques are popular and suitable model for many difficult optimization problems. In various practical problems, different objectives are to be considered. Now, most of the problems have some objectives those are conflicting in nature. Hence, only one objective cannot be optimized or prioritize because it can result in some adverse effect on other objective, and can produce some undesired results in terms of the other objectives. The main goal of this chapter is to give a comprehensive study of multiobjective optimization techniques in biomedical image analysis problem. The different models are categorized depending on the relevant features. For example, the different aspects of the optimization methods employed, different formulations of the problems, categories of data, and the domain of the application. This study mainly focuses on the multiobjective optimization techniques that can be used to analyze digital images specially biomedical images. Here, some of the problems, and challenges related to images are diagnosed and analyzed with multiple objectives. It is a comprehensive study that consolidated some of the recent works along with future directions.

The convolutional neural networks with supervised training solve problems of semantic segmentation in a wide area of biomedical applications. However, this approach requires a significant size of the training dataset. Unsupervised learning can provide segmentation of some images without training dataset. Joint unsupervised learning is an approach that combines neural networks with another algorithm of data clusterization. In this preliminary study, we test the joint unsupervised learning approach with minor modifications for the problem of semantic segmentation at several datasets of biomedical images.