Tuberculosis (TB) is is an infectious disease that generally attacks the lungs and causes death for millions of people annually. Chest radiography and deep-learning-based image segmentation techniques can be utilized for TB diagnostics. Convolutional Neural Networks (CNNs) has shown advantages in medical image recognition applications as powerful models to extract informative features from images. Here, we present a novel hybrid method for efficient classification of chest X-ray images. First, the features are extracted from chest X-ray images using MobileNet, a CNN model, which was previously trained on the ImageNet dataset. Then, to determine which of these features are the most relevant, we apply the Artificial Ecosystem-based Optimization (AEO) algorithm as a feature selector. The proposed method is applied to two public benchmark datasets (Shenzhen and Dataset 2) and allows them to achieve high performance and reduced computational time. It selected successfully only the best 25 and 19 (for Shenzhen and Dataset 2, respectively) features out of about 50,000 features extracted with MobileNet, while improving the classification accuracy (90.2% for Shenzen dataset and 94.1% for Dataset 2). The proposed approach outperforms other deep learning methods, while the results are the best compared to other recently published works on both datasets.
Accomplishing high recognition performance is considered one of the most important tasks for handwritten Arabic character recognition systems. In general, Optical Character Recognition (OCR) systems are constructed from four phases: pre-processing, feature extraction, feature selection, and classification. Recent literature focused on the selection of appropriate features as a key point towards building a successful and sufficient character recognition system. In this paper, we propose a hybrid machine learning approach that utilizes neighborhood rough sets with a binary whale optimization algorithm to select the most appropriate features for the recognition of handwritten Arabic characters. To validate the proposed approach, we used the CENPARMI dataset, which is a well-known dataset for machine learning experiments involving handwritten Arabic characters. The results show clear advantages of the proposed approach in terms of recognition accuracy, memory footprint, and processor time than those without the features of the proposed method. When comparing the results of the proposed method with other recent state-of-the-art optimization algorithms, the proposed approach outperformed all others in all experiments. Moreover, the proposed approach shows the highest recognition rate with the smallest consumption time compared to deep neural networks such as VGGnet, Resnet, Nasnet, Mobilenet, Inception, and Xception. The proposed approach was also compared with recently published works using the same dataset, which further confirmed the outstanding classification accuracy and time consumption of this approach. The misclassified failure cases were studied and analyzed, which showed that they would likely be confusing for even Arabic natives because the correct interpretation of the characters required the context of their appearance.
23 September 2022 in Expert Systems with Applications
Multilevel thresholding is one of the most effective image segmentation methods, due to its efficiency and easy implementation. This study presents a new multilevel thresholding method based on a modified artificial ecosystem-based optimization (AEO). The differential evolution (DE) is applied to overcome the shortcomings of the original AEO. The main idea of the proposed method, artificial ecosystem-based optimization differential evolution (AEODE), is to employ the operators of the DE as a local search of the AEO to improve the ecosystem of solutions. We used benchmark images to test the performance of the AEODE, and we compared it to several existing approaches. The proposed AEODE achieved a high performance when evaluated by the structural similarity index (SSIM), peak signal-to-noise ratio (PSNR), and fitness values. Moreover, the AEODE outperformed the basic version of the AEO concerning SSIM and PSNR by 78% and 82%, respectively, which reserves the best features for each of AEO and DE.
Arabic handwritten letter recognition (AHLR) is a challenging field in pattern recognition due to the different font styles, size, and noises. The font style is changed according to several criteria such as human skills, age, and culture. So, till now there is no such a system with sufficient accuracy. In the current paper, a new approach is introduced based on Moth-flame optimization (MFO) for AHLR, called (MFO-AHLR). The main goal of the proposed approach is to improve the accuracy of AHLR with a least number of features. MFO-AHLR consists of the following phases: pre-processing (including binarization and noise removal), feature extraction (including several kinds of features) and classification (which are kNN, RF, and LDA in this work). Between feature selection and classification, we adopt MFO as a feature selector. A benchmark dataset for Arabic handwritten letter images (CENPARMI) was used. The achieved results showed superior results for the selected features in all experiments. Also, for all classifiers, the selected feature sets outperformed the non-selected features; and the processing time was improved. The MFO-AHLR achieved 99.25% of classification accuracy which is the highest among the other published works. To the best of our knowledge, there is no AHLR achieved it.