Bayesian Functional Spatial Partitioning Methods for Prostate Cancer Lesion Detection

Loading...
Thumbnail Image

Persistent link to this item

Statistics
View Statistics

Journal Title

Journal ISSN

Volume Title

Title

Bayesian Functional Spatial Partitioning Methods for Prostate Cancer Lesion Detection

Published Date

2022-05

Publisher

Type

Thesis or Dissertation

Abstract

Manual protocols to predict the number, size, and location of cancerous lesions in the prostate using imaging data are highly dependent on reader experience and expertise. Existing computer-aided voxel-wise classifiers do not directly provide estimates of lesion boundaries, which are clinically important. Spatial partitioning methods estimate boundaries separating regions of local stationarity in spatially registered data, but existing methods are inadequate for the application of lesion detection because the boundaries are restricted to be linear or piecewise linear. We first introduce a novel Bayesian functional spatial partitioning method (BFSP-1) which estimates the partitioning boundary around an anomalous region of data with a distinct distribution or spatial process. Our algorithm transitions between a fixed Cartesian and a moving polar coordinate system to model the boundary with functional estimation tools. Using adaptive Metropolis-Hastings, the BFSP-1 algorithm simultaneously estimates the partitioning boundaries and the parameters of the spatial distributions within each region. BFSP-1 assumes the data contain one and only one anomalous region. To create a more clinically useful tool, we build upon our original boundary estimation framework and propose BFSP-M for multiple region discovery. This method uses reversible jump Markov chain Monte Carlo to jointly estimate the number of lesions, their boundaries, and their distinct spatial processes. Finally, we discuss BFSP-3D, which extends the BFSP framework to three dimensions. Through simulation, we show that our methods are robust to the shape of the target zone and region-specific spatial processes. Our methods prove to be a clinically useful tool for automatic boundary drawing of cancerous lesions using non-invasive prostate imaging data.

Keywords

Description

University of Minnesota Ph.D. dissertation. May 2022. Major: Biostatistics. Advisors: Lin Zhang, Joseph Koopmeiners. 1 computer file (PDF); viii, 97 pages.

Related to

Replaces

License

Collections

Series/Report Number

Funding information

Isbn identifier

Doi identifier

Previously Published Citation

Other identifiers

Suggested citation

Masotti, Maria. (2022). Bayesian Functional Spatial Partitioning Methods for Prostate Cancer Lesion Detection. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/241432.

Content distributed via the University Digital Conservancy may be subject to additional license and use restrictions applied by the depositor. By using these files, users agree to the Terms of Use. Materials in the UDC may contain content that is disturbing and/or harmful. For more information, please see our statement on harmful content in digital repositories.