Embedded Files
My Ph.D. research focuses on computational imaging methods to improve the quality and speed of tomographic imaging. During my Ph.D., I have worked on various research collaborations with Eli Lily, Canadian Light Source, Argonne National Laboratory, General Electric, Northup Grumman, and Air Force Research Laboratory. These collaborations have lead to several publications and open-source software packages.
Multi-Slice Fusion
Multi-Slice Fusion
Collaboration between Purdue and Eli Lily
Inverse problems spanning four or more dimensions such as space, time and other independent parameters have become increasingly important
Convolutional neural networks (CNNs) have produced state of the art results in many cases but are limited to 2D or 3D images
Multi-Slice Fusion provides a framework to create high-dimensional prior models for iterative reconstruction by combining CNNs along different planes
Reconstruction results with 90° limited angle views per time-point:
The multi-slice fusion result does not suffer from major limited-angle artifacts in contrast to FBP and MBIR+4D-MRF.
CodEx: Coded Exposure CT Reconstruction
CodEx: Coded Exposure CT Reconstruction
Collaboration between Purdue and Argonne national laboratory
Multi-Pose-Fusion
Multi-Pose-Fusion
Collaboration between Purdue and Eli Lily
X-ray computed tomography (CT) is used to reconstruct the internal morphology of a three dimensional object from a collection of projection images, most commonly using a single rotation axis.
For objects containing dense materials like metal, the use of a single rotation axis may leave some regions of the object obscured by the metal
In this work, we propose pixel-weighted Multi-pose Fusion: a framework to reduce metal artifacts by fusing the information from complementary measurement poses into a single reconstruction
Separable Models for cone-beam MBIR Reconstruction
Separable Models for cone-beam MBIR Reconstruction
Collaboration between Purdue, GE Aviation, Northrop-Grumman, and Air-force Research Laboratory
The proposed separable cone-beam projection operator allows faster computation of tomographic reconstruction due to more efficient computation and improved parallelism and cache efficiency
The proposed efficient tomographic projector allows us to implement model based iterative reconstruction with advanced prior models leading to an improved reconstruction quality
A Model Based Neuron Detection Approach Using Sparse Location Priors
A Model Based Neuron Detection Approach Using Sparse Location Priors
Page updated
Google Sites
Report abuse