Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Semi-Automated Diagnosis of Pulmonary Hypertension Using PUMA, a Pulmonary Mapping and Analysis Tool

Berty, Holly (2013) Semi-Automated Diagnosis of Pulmonary Hypertension Using PUMA, a Pulmonary Mapping and Analysis Tool. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Primary Text

Download (7MB) | Preview


Pulmonary Arterial Hypertension (PAH) is a progressive, potentially fatal disease that results in the remodeling of the pulmonary vasculature. Currently the gold standard for diagnosis of pulmonary hypertension is through right heart catheterization, an invasive and costly procedure where pressure measurements are made directly within the affected vessels. Since PAH is associated with the remodeling of the pulmonary arteries, others have proposed quantifying the vessel geometry depicted in computed tomography (CT) images as a non-invasive technique for diagnosis of PAH. The work presented here proposes a similar method of diagnosis by defining and incorporating techniques that are both manual in nature in reference to the segmentation process and automated with the modeling and anatomic measurement quantification steps. Data comprised of both normal and disease cases were gathered and the vessel geometry (specifically the pulmonary trunk, right main pulmonary artery and the left main pulmonary artery) were measured both manually and automatically. A comparison of the automated measurements of the vessel geometry to the manual measurements showed no significant difference between the means of the two groups. A significant difference was found between the cases and the controls leading to the possibility of classifying images based on the vessel geometry. Logistic regression and naïve Bayes models were constructed from the data for discriminating the cases from the controls. Overall, the Naïve Bayes model performed better with a higher sensitivity of 42.9% compared to 19% and a small decrease in specificity of 90.9% from 96.6%, and the model is able to classify correctly more of the patients with disease. Due to the permanent nature of the disease a type I error is acceptable; we prefer to classify patients that do not have the disease as positives than vice versa. We found that the segmenting of additional branches of the pulmonary vasculature could provide additional information for the improvement of the models presented here. In conclusion, we were able to quantify the vessel geometry depicted in CT images as a non-invasive technique for diagnosing PAH and we have shown that the two classes of measurements are not significantly different.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Berty, Hollyhlp8@pitt.eduHLP8
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairCooper, Gregory Fgfc@cbmi.pitt.eduGFC
Committee MemberVisweswaran, Shyamshv3@pitt.eduSHV3
Committee MemberSimon, Marc
Thesis AdvisorChapman,
Date: 11 April 2013
Date Type: Publication
Defense Date: 20 November 2012
Approval Date: 11 April 2013
Submission Date: 4 April 2013
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 217
Institution: University of Pittsburgh
Schools and Programs: School of Medicine > Biomedical Informatics
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: Pulmonary hypertension Vascular segmentation Prediction modeling Right-heart catheterization Computed tomography pulmonary angiography
Date Deposited: 11 Apr 2013 12:13
Last Modified: 15 Nov 2016 14:11


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item