Western Kentucky University
Department of Physics and Astronomy

Colloquium

WKU Physics Majors

Department of Physics and Astronomy
Western Kentucky University

"Senior project presentations"

November 20, 2017 @ 4:00 pm in TCCW 201

Abstract

Trason Carter
"Design and Construction of a Faraday Disk for Electricity and Magnetism Education"

A Faraday Disk is an induction generator first pioneered by Michael Faraday in 1831 consisting of a conductive wheel spinning perpendicularly to a magnetic field. Our team has been working to design a Faraday Disk to demonstrate the effect of electromagnetic induction for Electricity and Magnetism students. The system will have variable wheel speed and magnetic field strength, as well as induced voltage readout so that students can verify analytical models with experimental data. The design has been modeled using ANSYS Maxwell and SOLIDWORKS, and the build will begin next semester

Byron Grant
"Assessment of diagnostic image quality of computed tomography (CT) images of the lung using deep learning"

For imaging modalities such as radiography or computed tomography (CT) where the patient is exposed to ionizing radiation during imaging, it is important to optimize imaging protocols to ensure the scan is performed at the lowest dose that yields diagnostic images. To accomplish this, it is important to verify that image quality of the acquired scan is sufficient for the diagnostic task at hand. Since the image quality strongly depends on both the characteristics of the patient as well as the imager, both of which are highly variable, simplistic parameters like noise to determine the threshold is challenging. In this work, we apply deep learning using convolutional neural network (CNN) to predict whether CT scans meet the minimal image quality threshold for diagnosis. The data set consists of 64 cases of high resolution axial CT scans acquired for the diagnosis of interstitial lung disease. The quality of the images is categorized by a radiologist. While the number of cases is relatively small for deep learning tasks, each case consists of more than 200 slices, comprising a total of 9042 images. The deep learning involves extracting features from the images using 5 max-pooling layers of VGG19 pre-trained network followed by feature classification using a support vector machine. This method was compared to current methods used to determine image quality in CT images.

Carson King
"Determining the Relation between Source Length and Energy in Coronal Hard X-ray Sources"

As part of a collaborative effort involving personnel at WKU, GSFC and the University of Genoa, RHESSI images along with AIA/SDO overlays were used to determine the size and orientation of several coronal hard X-ray sources exhibiting loop-like structures, over several energy ranges. Various different imaging algorithms were used in order to ensure robust results. To reduce contamination from outside sources (which unavoidably contribute to the observed visibilities and thus parameters deduced from purely visibility-based imaging methods) and thus place greater emphasis on the main sources, direct images (in both count and electron domains) were produced and the dominant loop structures isolated using simple “mask” shapes (e.g., rectangles, parallelograms). The length of each loop was then measured as a function of electron energy in order to test the thick-target coronal source model for such events, and to deduce parameters such as acceleration region length and density.