DME-Deep: A Computerize Tool for Detection of Diabetic Macular Edema Grading Based on Multilayer Deep Learning and Transfer Learning

Diabetic macular edema (DME) is a common disease of diabetic retinopathy (DR). Due to the infection of DME disease, many patients’ vision is lost. To cure DME eye disease, early detection and treatment are very important and vital steps. To automatically diagnosis DEM disease, several studies were developed by detection of the macula center which is dependent on optic disc (OD) location. In this paper, a novel features pre-training based model was proposed based on dense convolutional neural network (DCNN) to diagnose DME related disease. As a result, a computerize tool “DME-Deep” for detection of DME-based grading system was implemented through a new dense deep learning model and feature’s transfer learning approaches. This DCNN model was developed by adding new five convolutional and one dropout layers to the network. The DME-Deep system was tested on three different datasets, which obtained from online sources. To train the DCNN model for features learning, the 1650 retinal fundus images were utilized from the Hamilton HEI-MED, ISBI 2018 IDRiD and MESSIDOR datasets. On datasets, the DME-Deep achieved 91.2% of accuracy, 87.5% of sensitivity and 94.4% of specificity. Compare to obtain hand-crafted features, the automatic feature’ learning it provided favorable results. Hence, the experimental results also indicate that this DME-Deep system can automatically assist ophthalmologists in finding DEM eye-related disease.

Structure Prediction of ncRNA genes: A review on Computational Intelligence Algorithms

Noncoding RNAs (ncRNAs) are an important part of genes and having an important role in human cellular activities and serious diseases. To predict ncRNAs structure, there are many computational intelligence algorithms (CIAs) that are developed in past studies. However, many studies suggested that there were still many structures that are still unpredictable by researchers. In this paper, CIAs algorithms were comprehensively reviewed to predict ncRNAs structures. The advantages and disadvantages of CIA algorithms are briefly mentioned related to ncRNA genes. Moreover, the latest software tools are also compared and reviewed to identify the structure of ncRNAs for mining deep sequencing data. In this study, conventional machine learning algorithms are mainly focused and future trends are also described to predict ncRNAs structure. This paper concludes that there is a need for improving CIA algorithms by using deep learning architectures in terms of layers and computational complexity to predict ncRNAs structures.

Current Challenges of Computational Intelligent Techniques for Functional Annotation of ncRNA Genes

The limited understanding of functional annotation of non-coding RNAs (ncRNAs) has been largely due to the complex functionalities of ncRNAs. They perform a vital part in the operation of the cell. There are many ncRNAs available such as micro RNAs or long non-coding RNAs that play important functions in the cell. In practice, there is a strong binding of the function of RNAs that must be considered to develop computational intelligent techniques. Comprehensive modeling of the structure of an ncRNA is essential that may provide the first clue towards an understanding of its functions. Many computational techniques have been developed to predict ncRNAs structures but few of them focused on the functions of ncRNA genes. Nevertheless, the accuracy of the functional annotation of ncRNAs is still facing computational challenges and results are not satisfactory. Here, many computational intelligent methods were described in this paper to predict the functional annotation of ncRNAs. The current literature review is suggested that there is still a dire need to develop advanced computational techniques for functional annotating of ncRNA genes in terms of accuracy and computational time.