Alteration Detection of Multispectral/Hyperspectral Images Using Dual-Path Partial Recurrent Networks

Numerous alteration detection methods are designed based on image transformation algorithms and divergence of bi-temporal images. In the process of feature transformation, pseudo variant information caused by complex external factors will be highlighted. As a result, the error of divergence between the two images will be further enhanced. In this paper, we propose to fuse the variability of Deep Neural Networks’ (DNNs) structure flexibly with various detection algorithms for bi-temporal multispectral/hyperspectral imagery alteration detection. Specifically, the novel Dual-path Partial Recurrent Networks (D-PRNs) was proposed to project more accurate and effective deep features. The Unsupervised Slow Feature Analysis (USFA), Iteratively Reweighted Multivariate Alteration Detection (IRMAD), and Principal Component Analysis (PCA) were then utilized, respectively, with the proposed D-PRNs, to generate two groups of transformed features corresponding to the bi-temporal remote sensing images. We next employed the Chi-square distance to compute the divergence between two groups of transformed features and, thus, obtain the Alteration Intensity Map. Finally, threshold algorithms K-means and Otsu were, respectively, applied to transform the Alteration Intensity Map into Binary Alteration Map. Experiments were conducted on two bi-temporal remote sensing image datasets, and the testing results proved that the proposed alteration detection model using D-PRNs outperformed the state-of-the-art alteration detection model.

Eye Movement Feature Set and Predictive Model for Dyslexia

Dyslexia is a learning disorder that can cause difficulties in reading or writing. Dyslexia is not a visual problem but many dyslexics have impaired magnocellular system which causes poor eye control. Eye-trackers are used to track eye movements. This research work proposes a set of significant eye movement features that are used to build a predictive model for dyslexia. Fixation and saccade eye events are detected using the dispersion-threshold and velocity-threshold algorithms. Various machine learning models are experimented. Validation is done on 185 subjects using 10-fold cross-validation. Velocity based features gave high accuracy compared to statistical and dispersion features. Highest accuracy of 96% was achieved using the Hybrid Kernel Support Vector Machine- Particle Swarm Optimization model followed by the Xtreme Gradient Boosting model with an accuracy of 95%. The best set of features are the first fixation start time, average fixation saccade duration, the total number of fixations, total number of saccades and ratio between saccades and fixations.

Eye Movement Feature Set and Predictive Model for Dyslexia

Dyslexia is a learning disorder that can cause difficulties in reading or writing. Dyslexia is not a visual problem but many dyslexics have impaired magnocellular system which causes poor eye control. Eye-trackers are used to track eye movements. This research work proposes a set of significant eye movement features that are used to build a predictive model for dyslexia. Fixation and saccade eye events are detected using the dispersion-threshold and velocity-threshold algorithms. Various machine learning models are experimented. Validation is done on 185 subjects using 10-fold cross-validation. Velocity based features gave high accuracy compared to statistical and dispersion features. Highest accuracy of 96% was achieved using the Hybrid Kernel Support Vector Machine- Particle Swarm Optimization model followed by the Xtreme Gradient Boosting model with an accuracy of 95%. The best set of features are the first fixation start time, average fixation saccade duration, the total number of fixations, total number of saccades and ratio between saccades and fixations.

Automatic Quantification of Atherosclerosis in Contrast-Enhanced MicroCT Scans of Mouse Aortas Ex Vivo

Objective. While microCT evaluation of atherosclerotic lesions in mice has been formally validated, existing image processing methods remain undisclosed. We aimed to develop and validate a reproducible image processing workflow based on phosphotungstic acid-enhanced microCT scans for the volumetric quantification of atherosclerotic lesions in entire mouse aortas. Approach and Results. 42 WT and 42 apolipoprotein E knockout mouse aortas were scanned. The walls, lumen, and plaque objects were segmented using dual-threshold algorithms. Aortic and plaque volumes were computed by voxel counting and lesion surface by triangulation. The results were validated against manual and histological evaluations. Knockout mice had a significant increase in plaque volume compared to wild types with a plaque to aorta volume ratio of 0.3%, 2.8%, and 9.8% at weeks 13, 18, and 26, respectively. Automatic segmentation correlated with manual ( r 2 ≥ 0.89 ; p < .001 ) and histological evaluations ( r 2 > 0.96 ; p < .001 ). Conclusions. The semiautomatic workflow enabled rapid quantification of atherosclerotic plaques in mice with minimal manual work.

Improved Otsu and Kapur approach for white blood cells segmentation based on LebTLBO optimization for the detection of Leukemia

<abstract> <p>The diagnosis of leukemia involves the detection of the abnormal characteristics of blood cells by a trained pathologist. Currently, this is done manually by observing the morphological characteristics of white blood cells in the microscopic images. Though there are some equipment- based and chemical-based tests available, the use and adaptation of the automated computer vision-based system is still an issue. There are certain software frameworks available in the literature; however, they are still not being adopted commercially. So there is a need for an automated and software- based framework for the detection of leukemia. In software-based detection, segmentation is the first critical stage that outputs the region of interest for further accurate diagnosis. Therefore, this paper explores an efficient and hybrid segmentation that proposes a more efficient and effective system for leukemia diagnosis. A very popular publicly available database, the acute lymphoblastic leukemia image database (ALL-IDB), is used in this research. First, the images are pre-processed and segmentation is done using Multilevel thresholding with Otsu and Kapur methods. To further optimize the segmentation performance, the Learning enthusiasm-based teaching-learning-based optimization (LebTLBO) algorithm is employed. Different metrics are used for measuring the system performance. A comparative analysis of the proposed methodology is done with existing benchmarks methods. The proposed approach has proven to be better than earlier techniques with measuring parameters of PSNR and Similarity index. The result shows a significant improvement in the performance measures with optimizing threshold algorithms and the LebTLBO technique.</p> </abstract>

Real-Time Heart Rate Monitoring for Wearable Electrocardiography Using Filter-Based and Peak Threshold Algorithms: A Comparative Study

Wearable and wireless Electrocardiograph (ECG) enables real-time and long-term monitoring of heart rate (i.e. 24 hours). Several algorithms have been introduced to increase the accuracy of the heart rate calculation for this type of ECG. This study aims to compare the accuracy of two heart rate calculation algorithms: Filter-Based and Peak Threshold. Both algorithms were implemented into a wearable ECG prototype comprising AD8232, 8-bit microcontroller (ATMega328), and Bluetooth. ECG signals and heart rate (in Beat per Minute (BPM)) were sent via Bluetooth and displayed in real-time on a Windows-based application created using Visual C #. Experiments were conducted on 10 healthy subjects aged 20.4 ± 2.0 years and body weight of 60.8 ± 10.2 kg. The measurement results calculated by using Filter-Based and Peak Threshold were compared to a commercial wearable ECG (KineticTM) as a ‘ground truth’. The test results showed that the Filter-Based algorithm resulted in a more accurate calculation with the Root Mean Square Error (RMSE) of 1.53, compared to the Peak Threshold algorithm with RMSE of 2.69.