Projects

Major Projects

Quantization of Respiratory Distress Severity : A Machine Learning Approach to Prevent Unexpected Hospital Deaths

Internship Project, GE Healthcare, May 2018 - July 2018

DeepSino: Super Resolution of Undersampled Noisy 2D Sinogram using Deep Learning

Course Project-Medical Imaging, IISc, Jan 2018 - Apr 2018

Convolutional Neural Network to outperform filtered back projection and other iterative reconstruction for under-sampled data. A replica of DnCNN (Denoising Convolutional Neural Network, 20 layer residual learning newtwork) trained and tested to remove of Gaussian & Poisson noise and up-sample sinogram in a single pass.

https://github.com/rohit-pardasani/DeepSino

Intelligent Interactive Virtual Bot

Undergraduate Project, IIT Roorke, Apr 2007 - May 2008

Minor Projects

Image Denoising and Super-Resolution using Residual Learning of Deep Convolutional Network

Course Project-Data Analysis and Visualization, IISc, Jan 2018 - Apr 2018

DeblurCNN : Convolutional Neural Network to Deblur Images

Experimental Project, IISc, Aug 2018

A 20 layer DnCNN (Denoising Neural Network) trained to remove blur introduced by averaging filter.

https://github.com/rohit-pardasani/DeblurCNN

New approach to Academic Grading

Course Project-Artificial Neural Networks, IIT Roorkee, Feb 2008 - Apr 2008

An application developed using MATLAB’s Fuzzy Inference System Tools

Database Management System for Library (in C & Fox Pro Language)

Course Project-Informatics Practices, XII Grade, 2004

A library database management system in C language and in Fox Pro (with Graphical User Interface).

My Python Notebooks : Data Science in Healthcare

Predicting No-Show for Medical Appointments

https://github.com/rohit-pardasani/MedicalAppointmentsNoShow

Based on information of patient we predict whether patient will show up or medical appointment or not

Model used: Multinomial Naive Bayes Classifier

Problem Statement & Dataset: Courtsey Kaggle https://www.kaggle.com/joniarroba/noshowappointments