A5018477867- AI in cancer detection
- COVID-19 diagnosis using AI
- Medical Image Segmentation Techniques
- Advanced Neural Network Applications
- Precipitation Measurement and Analysis
- Image Enhancement Techniques
- Photonic and Optical Devices
- Handwritten Text Recognition Techniques
- Climate variability and models
- Advanced Image and Video Retrieval Techniques
- Imbalanced Data Classification Techniques
- Image and Object Detection Techniques
- Irrigation Practices and Water Management
- Brain Tumor Detection and Classification
- Radiation Effects in Electronics
- VLSI and Analog Circuit Testing
- Biodiesel Production and Applications
- Plasmonic and Surface Plasmon Research
- Soil Moisture and Remote Sensing
- Retinal Imaging and Analysis
- Neural Networks and Applications
- Hydrology and Watershed Management Studies
- Phytoplasmas and Hemiptera pathogens
- Cell Image Analysis Techniques
- Colorectal Cancer Screening and Detection
Purdue University Northwest
2021-2022
Tohoku University
2020
University of Naples Federico II
2018
Virginia Commonwealth University
2013-2016
Bangladesh University of Engineering and Technology
2009-2012
In this paper, we propose a U-net architecture that integrates residual skip connections and recurrent feedback with EfficientNet as pretrained encoder. Residual help feature propagation in deep neural networks significantly improve performance against similar number of parameters while ameliorate gradient learning. We also second model utilizes densely connected layers aiding deeper networks. is family powerful encoders streamline network design. The proposed are evaluated state-of-the-art...
A completely new scheme for quaternary logic is proposed. Instead of conventional Fuzzy or Galois Field theory, the system based on extension Boolean algebra. The capable handling both and coupled-binary inputs, where binary operands are coupled in pairs to form entities. All necessary operators defined several theorems properties derived develop a way expressing arbitrary truth tables with sum-of-product functions. To demonstrate functionality this novel scheme, some useful blocks such as...
Using a novel quaternary algebra, several sequential circuits are designed. The logic scheme used here is obtained by extending Boolean algebra into domain. A set of operators capable handling both coupled-binary and ordinary inputs to design the circuit elements such as latches, flip-flops, registers counters. Sufficient conditions for stable operation blocks derived mathematically demonstrated graphically using simulated timing diagrams.
Mosquito traps offer researchers and health officials a reasonable estimate of mosquito abundances to assess the spatial temporal occurrences mosquito-transmitted pathogens. Existing traps, however, have issued efficient design detect energy consumption device. We designed novel collection device that sensitively detects presence via fiber-optic sensor. In this prototype, pushing capture mechanism selectively powers efficiently captures live mosquitoes without destroying identifying...
U-Net is a deep learning technique that has made significant contributions to medical image segmentation. Although the accomplishments of algorithms in terms processing are evident, many challenges still need be overcome achieve human-like performance. One main building deeper U-Nets black-box problems, such as vanishing gradients. Overcoming this problem allows us develop neural networks with advanced network designs.We propose three variants, namely efficient R2U-Net, dense U-Net, and...
Cloud Colonography is proposed in this paper, using different types of cloud computing environments. Databases from the Computed Tomographic (CTC) screening tests among several hospitals are explored. These networked databases going to be available near future via technologies. Associated Multiple (AMD) was developed study handle multiple CTC databases. When AMD used for assembling databases, it can achieve very high classification accuracy. The has potential play a role core classifier...
This paper describes the technique of extracting vehicle license-plate and interpretation registration code from a captured image in real-time. The developed algorithm is divided into three stages: extraction license plate image, segmentation components recognition characters for code. A control also to choose between processes reconstruct recognized characters. Most are implemented using mostly logical operations achieve lower execution time.
In this contribution, design of simple label-free biosensors based on surface plasmon ring resonators was discussed. Eight sensors with distinct geometry were proposed and the effect resonator sensor performance numerically investigated using FDTD scheme by MATLAB. The region wavelength chosen for operation biological window (990nm-1100nm). results provided insight to choose proper designing within aforementioned spectrum.