Md. Biddut Hossain

Sejong University | sejong · Artificial Engineering and Robotics

Edge computing, difference between Edge and cloud computing and several security issues are described in this presentation.

Subjective visual examination by human dermatologists is associated with inter-observer variability and error. To address this problem, we present a method to accurately diagnose the 23 most common skin conditions, using an adaptive GrabCut approach with the EfficientNetB3 model, for accurate segmentation and classification, respectively. Using a c...

We present a deep learning-based method that corrects motion artifacts and thus accelerates data acquisition and reconstruction of magnetic resonance images. The novel model, the Motion Artifact Correction by Swin Network (MACS-Net), uses a Swin transformer layer as the fundamental block and the Unet architecture as the neural network backbone. We...

Citation: Hossain, M.B.; Shinde, R.K.; Oh, S.; Kwon, K.-C.; Kim, N. A Systematic Review and Identification of the Challenges of Deep Learning Techniques for Undersampled Magnetic Resonance Image Reconstruction. Sensors 2024, 24, 753. Abstract: Deep learning (DL) in magnetic resonance imaging (MRI) shows excellent performance in image reconstruction...

Citation: Hossain, M.B.; Shinde, R.K.; Oh, S.; Kwon, K.-C.; Kim, N. A Systematic Review and Identification of the Challenges of Deep Learning Techniques for Undersampled Magnetic Resonance Image Reconstruction. Sensors 2024, 24, 753. Abstract: Deep learning (DL) in magnetic resonance imaging (MRI) shows excellent performance in image reconstruction...

This article describes a novel approach for enhancing the three-dimensional (3D) point cloud reconstruction for light field microscopy (LFM) using U-net architecture-based fully convolutional neural network (CNN). Since the directional view of the LFM is limited, noise and artifacts make it difficult to reconstruct the exact shape of 3D point cloud...

We propose a dual-domain deep learning technique for accelerating compressed sensing magnetic resonance image reconstruction. An advanced convolutional neural network with residual connectivity and an attention mechanism was developed for frequency and image domains. First, the sensor domain subnetwork estimates the unmeasured frequencies of k-spac...

We propose generated a high-quality 3D model using a deep neural network by acquiring high-resolution 3D data from real-world objects and generated a high-speed and high-quality 3D CGH based on integrated hybrid map layering for use in holographic display.

Simple Summary Skin cancer is a life-threatening condition. It is difficult to diagnose in its early stages; therefore, we proposed an easy-to-use telemedicine device to tackle skin cancer without expert intervention. The deep learning model automatically detects skin cancer patches on lesions with a credit-card-sized device named Raspberry Pi and...

When sparsely sampled data are used to accelerate magnetic resonance imaging (MRI), conventional reconstruction approaches produce significant artifacts that obscure the content of the image. To remove aliasing artifacts, we propose an advanced convolutional neural network (CNN) called fully dense attention CNN (FDA-CNN). We updated the Unet model...

A holographic display provides natural-view three-dimensional (3D) images with exact depth and parallax information for given objects and reconstructs the actual 3D visualization. The depth map-based methods have the advantages that rapid calculation of CGHs than the point cloud-based methods; and provides a better image quality. However, it has di...

Air-writing is a growing research topic in the field of gesture-based writing systems. This research proposes a unified, lightweight, and general-purpose deep learning algorithm for a trajectory-based air-writing recognition network (TARNet). We combine a convolutional neural network (CNN) with a long short-term memory (LSTM) network. The architect...

Using a microlens array (MLA), integral imaging microscopy (IIM) captures angular information in three dimensions. In this paper, we provide a three-dimensional display system using a deep neural network and augmented reality (AR) for the IIM that renders a true 3D image. There are three parts to this procedure: image capture using the IIM system [...

We proposed an end-to-end MRI reconstruction method based on the generative adversarial networks (GAN) where the generator network maps the relationship between under-sampled and fully sampled k-data. The discriminator network judges whether the reconstructed image is real or fake and it is determined by adversarial loss. The perceptual, pixel-wise...

This study proposes a robust depth map framework based on a convolutional neural network (CNN) to calculate disparities using multi-direction epipolar plane images (EPIs). A combination of three-dimensional (3D) and two-dimensional (2D) CNN-based deep learning networks is used to extract the features from each input stream separately. The 3D convol...

The epipolar plane images (EPIs) that represent the disparity of a 3D point are widely used to estimate depth in light fields. In this study, we use an approach that extracts EPIs in all possible directions from an elemental image array (EIA) of an integral imaging microscopy (IIM) system. Our work presents an occlusion-aware depth estimation netwo...

In this paper, we proposed a Swin transformer-based magnetic resonance image reconstruction method with Unet architecture to improve image quality and speed up the reconstruction from highly under-sampled k-space data. To remove checkboard artifacts, we presented a dual up-sample block design that includes both subpixel and bilinear up-sample algor...

Convolutional neural networks (CNNs) have recently shown impressive results for estimating depth-map from microscopic images. This study presents a depth estimation network using CNNs and microscopic epipolar plane images (EPIs) based on synthetic light field datasets. Multiple directions of microscopic images are chosen as inputs, and convolutiona...

In this paper, we propose a dual-domain architecture for magnetic resonance imaging (MRI) reconstruction method which performs both the k-space and the image domains to reduce the image reconstruction time. In order to improve the resolution of reconstructed images, we applied two separate convolutional neural networks (CNNs), one for k-domain anot...

The waveguide-type full-color 3D-AR display system based on the integral imaging technique using the holographic mirror array is proposed. In the experiment, the AR feature has been successfully verified that the real-world scene and reconstructed virtual full-color 3D image were observed simultaneously.

Holographic optical element (HOE) recording technique using holographic printing is presented. Holographic printer has the potential to realize complex optical functions with high diffraction efficiency and provides wide field of view for augmented-reality near-eye display.

Magnetic resonance imaging (MRI) is a widely used non-ionizing medical imaging modality. In MRI the data samples are not being directly connected with image rather than k-space. K-space contains spatial frequency information that is acquired line-by-line and converted to images by inverse Fast Fourier Transform (IFFT). For high-quality reconstructe...

Integral imaging microscopy (IIM) has a wide range of applications, although it requires some pre-processing of the raw images before advancing operation1. This paper presents a methodology for the rotational calibration of the microlens array (MLA) in the IIM system. Rotational errors caused by the assembly of defects can result in 3D information...

In an integral-imaging microscopy (IIM) system, a microlens array (MLA) is the primary optical element; however, surface errors impede the resolution of a raw image’s details. Calibration is a major concern with regard to the incorrect projection of the light rays. A ray-tracing-based calibration method for an IIM camera is proposed, to address fou...

In this work, a touchless user interactive high-resolution light field display system using a three-dimensional tracking camera has been proposed. The whole process is divided into two principal parts: LF capturing and the display system. An electrically controllable camera slider is used to capture 71 horizontal views of a 3D object. Moreover, the...

In this research, a trajectory-based air writing character identification system using a fusion of convolution neural network (CNN) and long short-term memory (LSTM) named CNN-LSTM is proposed. Two publicly available datasets realsense trajectory digit (RTD) and realsense trajectory digit (RTC) were employed. RTD and RTC dataset contains 20,000 dig...

This research focuses on the higher frame rates, smooth motion parallax, and high-resolution light field display. The frame rate and motion parallax problems are solved by the tracking system and depth camera. An automatic electric motor is used to capture the views. Finally, the view slices are viewed in a stereo display. Thus, the observer can pe...

At present, the development of global cloud computing is booming. So many governments and large enterprises have their different development strategies and directions for the cloud computing development. To upgrade the present ICT cell in public as well as private sectors Virtual Desktop Infrastructure based cloud desktop may be one of the best app...

Lung cancer is one of the major health issues for both developed and developing countries. In the year 2000, more than one million lung cancer deaths have been reported worldwide. It has been also estimated that death for lung cancer will reach around 10 million in the year 2030. Different classifiers, segmentation techniques and features used in d...

31 Abstract— The Uses of Information Technology in day-today activities, the necessity for online services such as storage space, software, platforms is generating rapidly. This trend to generate a new concept of Cloud Computing. Security of the Cloud Storage is the major concern of cloud computing. When it appears to security of the data stored in...

Data security has been a major concern in the today’s information technology era. Especially it becomes serious in the cloud environment because the data is located in different places all over the world. Encryption has come up as a solution and different encryption algorithms play an important role in data security on cloud. Encryption algorithms...