DOI

https://doi.org/10.25772/45J6-TZ62

Defense Date

2021

Document Type

Thesis

Degree Name

Master of Science

Department

Radiation Oncology

First Advisor

Lulin Yuan

Second Advisor

Laura Padilla

Abstract

Visualization of liver tumors on simulation CT scans is challenging even with contrast-enhancement, due to the sensitivity of the contrast enhancement to the timing of the CT acquisition. Image registration to magnetic resonance imaging (MRI) can be helpful for delineation, but differences in patient position, liver shape and volume, and the lack of anatomical landmarks between the two image sets makes the task difficult. This study develops a U-Net based neural network for automated liver and tumor segmentation for purposes of radiotherapy treatment planning. Non-contrast simulation based abdominal CT axial scans of 52 patients with primary liver tumors were utilized. Preprocessing steps included HU windowing to isolate livers from the scan and creating masks for liver and tumor using the radiotherapy structure set (RTSTRUCT) DICOM file, and converting the images to a PNG format. The RTSTRUCT file contained the ground truth contours that were manually labelled by the physician for both liver and tumor. The image slices were split into 1400 for training and 600 for validation. Two fully convolutional neural networks with a U-Net architecture were used in this study. The first U-Net segments the livers. The second U-Net segments the tumor from the liver segments produced from the first network. The dice coefficient for liver segmentation was 89.5% and the dice coefficient for liver tumor segmentation was 44.4%. The results showed that the proposed algorithm had good performance in liver segmentation and shows areas for improvement for liver tumor segmentation.

Rights

© The Author

Is Part Of

VCU University Archives

Is Part Of

VCU Theses and Dissertations

Date of Submission

8-13-2021

Share

COinS