Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


Thesis_Kovid.pdf (6.95 MB)

ETD Abstract Container

Abstract Header

Liver Segmentation by Geometric and Texture features using Support Vector Machine (LiGTS)

Bhatnagar, Kovid
http://rave.ohiolink.edu/etdc/view?acc_num=osu1452224451

Abstract Details

2016, Master of Science, Ohio State University, Electrical and Computer Engineering.
Magnetic resonance elastography (MRE) is a novel, non-invasive technique to measure the stiffness of soft tissue. In MRE, mechanical vibrations are induced in the region of interest (ROI), which are then encoded in the phase of an MR image. These wave images are further processed to obtain a spatial stiffness map. The work presented in this thesis is focused on segmenting the liver to facilitate stiffness quantification for staging liver fibrosis. The current practice requires an expert radiologist to segment the liver in MRE scans so that the stiffness computation can be performed within the liver. In addition to being labor intensive, this manual intervention causes inter- and intra-observer variability in drawing the boundaries of the liver and thus impacts the computed stiffness values. Hence, there is a need to increase the consistency of the stiffness computation process, which will enhance the clinical effectiveness of the MRE-based stiffness computation. Automating the segmentation process can improve repeatability and eliminate inter- and intra-observer variability, providing the desired consistency for MRE-based stiffness quantification. In this work, an automated method for liver segmentation is presented. The method is based on supervised learning and employs a combination of geometric and texture-based features in conjunction with support vector machine-based classification. For validation, liver MRE is performed in 14 healthy volunteers. For performance evaluation, results from automated segmentation are analyzed in terms of error rate, sensitivity, specificity, and precision, with manual segmentation providing the ground truth. The results show that the proposed automated segmentation method has an error rate of less than 4%. Also, the mean stiffness values derived from the proposed method are in good agreement with the values obtained from a manually drawn ROI. In summary, this study demonstrates and validates a new segmentation method for liver MRE.
Arun Kolipaka, Dr. (Advisor)
Rizwan Ahmad, Dr. (Advisor)
67 p.
Electrical Engineering

Recommended Citations

Citations

  • Bhatnagar, K. (2016). Liver Segmentation by Geometric and Texture features using Support Vector Machine (LiGTS) [Master's thesis, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1452224451

    APA Style (7th edition)

  • Bhatnagar, Kovid. Liver Segmentation by Geometric and Texture features using Support Vector Machine (LiGTS). 2016. Ohio State University, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1452224451.

    MLA Style (8th edition)

  • Bhatnagar, Kovid. "Liver Segmentation by Geometric and Texture features using Support Vector Machine (LiGTS)." Master's thesis, Ohio State University, 2016. http://rave.ohiolink.edu/etdc/view?acc_num=osu1452224451

    Chicago Manual of Style (17th edition)

osu1452224451
883
© 2016, some rights reserved.Creative Commons License Liver Segmentation by Geometric and Texture features using Support Vector Machine (LiGTS) by Kovid Bhatnagar is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by The Ohio State University and OhioLINK.