Meeting Banner
Abstract #0093

Derivation and Comparison of Site Specific Peripheral and Transition Zone Quantitative DCE MRI Logistic Regression Models for Prostate Cancer Detection: Does Cancer Location Matter?

Nikolaos Dikaios1, Mohamed Abd Alazeez2, Mark Emberton2, Taiki Fujiwara3, David Atkinson4, Shonit Punwani4

1Centre for Medical Imaging, University College London , London, Greater London, United Kingdom; 2Department of Urology, University College London Hospital, London, Greater London, United Kingdom; 3Department of Radiology, University College London Hospital, London, Greater London, United Kingdom; 4Centre for Medical Imaging, University College London, London, Greater London, United Kingdom

Dynamic contrast enhancement (DCE) MRI has been advocated for detection of prostate cancer, but studies predominantly concentrate on peripheral zone tumours. Biological differences are known to exist between tumours within transition and peripheral zones due to their different microenvironment. This study aims to compare and contrast quantitative DCE MRI parameters of prostate PZ and TZ cancer and non-cancer regions; to identify those best able to classify cancer tissue at each site and thereby improve DCE MRI based diagnostic models used within computer aided detection software.

Keywords

ability able affect aided aims analyst apical application arterial audience benign best biological biomedical biopsy blood bound cancer classification classify clinical college combination comprehensive computer computing concentrate confidence consensus constant contoured contrast core curve decreased density derived described detection diagnostic differentiate discriminatory disease dynamic early enhanced enhancement exist experienced facilitates field fitted formation funded funding gland grants greater health hospital hypertrophy identify illustrates improvement improving included individual initial institute intensity international interquartile interstitial intervals involved kingdom known length likely localization location logistic majority mapping mark median medical model models modified muscle national operator parametric patient patients peripheral permeability phenotype physicists placed placement plasma power predominantly presence previously proportion prostate providing quantitative radiological radiologists radiology received receiver reduced reflecting regression relatively remainder resolution segregated significance significantly site slice slope software space studies subsequent suggested suggesting surrounding table target template temporal thereby tissue tools transition type underwent univariate useful utility view zone zones