Diabetic retinopathy is a common complication of diabetes, that affects blood vessels in the light-sensitive tissue called the retina. It is the most common cause of vision loss among people with diabetes and the leading cause of vision impairment and blindness among working-age adults. Recent progress in the use of automated systems for diabetic retinopathy diagnostics has offered new challenges for the industry, namely the search for a less resource-intensive architecture, e.g., for the development of low-cost embedded software. This paper proposes a comparison between two widely used conventional architectures (DenseNet, ResNet) with the new optimized one (EfficientNet). The proposed methods classify the retinal image as one of 5 class cases based on the dataset obtained from the 4th Asia Pacific Tele-Ophthalmology Society (APTOS) Symposium.
Satellite remote sensing of near real-time reservoir filling has important implications for the monitoring of territorial hydrological resources. These tasks are particularly relevant for reservoirs whose ground hydrological characteristics are not accessible. The monitoring of reservoir filling demands bathymetric data or satellite altimetry. However, bathymetry and satellite altimetry data, which are essential for estimating water storage variations, are only available for a limited number of reservoirs. Another possibility is to combine satellite monitoring of the water mirror with data from the Digital Terrain Model (DEM). In this research, we suggest using global DEM to estimate the slope of a reference alluvial inflow cone entering a reservoir, such as a riverbed. Satellite monitoring of the water mirror creates a set of coastlines at different reservoir fills. The spatial location of coastlines on a reference alluvial cone with a known slope allows altimetric linking of water mirrors. The resulting set of altimetrically referred water mirrors gives the possibility to reconstruct the reservoir 3D bed characteristics. The 3D model of the reservoir bed is constructed as a set of layers. Each layer is a truncated pyramid. Pyramid bases are the neighboring altimetric reference water mirrors. The Kapshagay reservoir (Northwest China) on the River Tekes which is the main tributary of the cross-border (Kazakhstan-China) River Ili was considered as an example. Landsat-5,7,8 (resolution 30 m) data for the period 2007–2018 were used to monitor the water mirror area of the Kapshagay reservoir, which varies from 22 to 59 km2. Shuttle Radar Topography Mission (SRTM) was used as DEM. The nine-layer 3D model of the reservoir bed was created. The operation volume of the reservoir is estimated at 1,37 ± 0,08 km3, which is close to the official 1,4 km3. The empirical equation of the relationship between the mirror area and the water storage in the operating part of the reservoir, received as a result of 3D model calculations, is the basis for operational hydrological monitoring of objects that do not have access to ground data.
30 March 2023 in Scientific Journal of Astana IT University
Premedical diagnostics is the process of examining survey results. Correct premedical diagnostics can improve the process of patient management and reduce the burden on the medical sector. Diagnostics of medical images such as computed tomography and X-ray are obligatory steps for further treatment. However, the shortage of clinicians causes delays in this step. We observed two state-of-the-art algorithms proposed for medical image segmentation: TransUnet and Swin-Unet. We conducted a theoretical comparison of algorithms in terms of the applicability of pre-hospital diagnostics according to quality and speed of training. The comparison is based on the original source of code provided by the authors of the original articles. We chose these two algorithms because they have similar U-form architecture, a high level of citation, and show competitive DICE scores on pictures of various human organs. Some architectural features were also important. Both models inherit key elements of U-net. TransUnet is a hybrid Transformer and CNN model. It consists of Transformer encoder and a convolutional decoder. Some additional computations are required in the bottleneck. Swin-Unet is a fully Transformer-based model. These architectural differences give rise to a difference in the number of trainable parameters. Generally, deeper architectures with a bigger number of parameters usually show better performance, however, according to our review, Swin-Unet has a smaller number of parameters and shows better DICE and Hausdorff Distance. It should be noted that the distribution between false positive and false negative predictions is important in medical image processing. It is crucial to avoid overloading the medical sector while also not missing any sick patients. Precision and recall can be used to evaluate the ratio of incorrect predictions. Therefore, we also observed the results of caries segmentation where precision and DICE were provided. In this specific case, TransUnet shows better DICE and recall values but worse precision.
Natural language processing (NLP) field has been developing rapidly recently. This article consists mainly of literature review of the basic understanding and solving the causality problem in natural language processing field. Existing models may benefit from the concept of causality because conventional language models are brittle and spurious [10]. Incorporating the principle of causality could assist in resolving this issue. Since this issue affects seriously on the accuracy value of NLP methods and algorithms, it is worth paying attention to. Content of the article includes the authors who have been covered this topic and have made researches respecting mentioned problem, the results that have been achieved, the methods and approached that have been used and the data that was used in researches.