Hi, I am TahsinI'm a PhD student at Vanderbilt University in the Electrical and Computer Science Department. I have done my BSc and MSc in Electrical Engineering at Bangladesh University of Engineering and Technology. My current research focuses on the application of Machine Learning on skin cGVHD. I have been with Bengali.AI since February 2018. I play my part in planning the projects along with my teammates. Since I am big a fan of Kaggle competitions, I have the most fun designing the competition phase of the projects. I spend my free time either playing chess or studying music.
My current research interest is in the applications of machine learning algorithms in biomedical signals. Previously I have worked in 5G communication technologies.
Heart Disease Detection From ECG Signals
Myocardial Infarction is one of the leading causes of death worldwide. This project investigates the development of a Convolutional Neural Network architecture which distinguishes between inferior myocardial infarction (IMI) and healthy signals.
Skin Lesion Segmentation in GVHD patients
Chronic graft-versus-host-disease (cGVHD) is a common occurrence after hematopoietic stem cell transplantation (HCT) and the primary affected organ is skin. The extent of the disease is measured in terms of affected body surface area (BSA). The aim of this project is to develop an automated method to compute BSA.
Cross Modal Synthesis of CT Images
Magnetic resonance imaging (MRI) is increasingly favored for use in radiotherapy treatment planning (RTP) due to its excellent soft-tissue contrast and lack of ionizing radiation, as compared to computed tomography (CT). However, MRI currently plays a limited role in RTP due to its limited ability to provide electron density information, a requirement for calculation of tissue attenuation and dose distribution during treatment. As a result, there is a strong incentive to synthesize CT images from MRI, which would enable MRI-only RTP. We investigate a new approach for CT image synthesis, where the network is trained to predict the Fourier transform of CT images based on supervised learning
Cognitive radio (CR) is considered to be a promising technology for future wireless networks to make opportunistic utilization of the unused or underused licensed spectrum. Meanwhile, coordinated multipoint joint transmission (CoMP JT) is another promising technique to improve the performance of cellular networks. In this work, we propose a CR system with CoMP JT technique. We develop an analytical model of the received signal-to-noise ratio at a CR to determine the energy detection threshold and the minimum number of required samples for energy detection–based spectrum sensing in a CR network (CRN) with CoMP JT technique. Additionally, we formulate an optimization problem for a CRN with CoMP JT technique to configure the channel allocation and user scheduling for maximizing the minimum throughput of the users.
Reasat, Tahsin, and Celia Shahnaz. “Detection of inferior myocardial infarction using shallow convolutional neural networks.” In 2017 IEEE Region 10 Humanitarian Technology Conference (R10-HTC), pp. 718-721. IEEE, 2017. link
Reasat, Tahsin, Abir Saha, and Md Forkan Uddin. “Cognitive radio network with coordinated multipoint joint transmission.” International Journal of Communication Systems 30, no. 16 (2017): e3310. link
Alam, Samiul, Tahsin Reasat, Rashed Mohammad Doha, and Ahmed Imtiaz Humayun. “NumtaDB-Assembled Bengali Handwritten Digits.” arXiv preprint arXiv:1806.02452 (2018). link