Premium
Po‐Thur Eve General‐26: Tracking intraprostatic volumes for determining non‐linear warping algorithm variability
Author(s) -
Venugopal Niranjan,
McCurdy Boyd,
Hnatov Alex,
Dubey Arbind
Publication year - 2006
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.2244653
Subject(s) - image warping , prostate , electromagnetic coil , magnetic resonance imaging , algorithm , data set , prostate cancer , computer science , volume (thermodynamics) , nuclear medicine , artificial intelligence , medicine , physics , radiology , cancer , quantum mechanics
Magnetic resonance spectroscopy imaging (MRSI) is becoming routinely used for diagnosis and treatment planning of prostate cancer. MRSI provides physicians and physicists with information related to metabolic activity of tissues within the prostate. During the MRSI data acquisition, an endorectal radio frequency coil is utilized to boost signal strength in the localized volume, and brings about a ∼10 fold increase in the signal to noise ratio. A challenge exist is using the MRSI data collected using the endorectal coil technique. The endorectal coil, while crucially important to data acquisition, is filled with approximately with ∼100cc's of air. This pushes the prostate superiorly/anteriorly, deforming the prostate and consequently the spectroscopic imaging data in a non‐linear manner. In this application, the coil‐deformed MRS images are warped back to a non‐deformed state, using a single data set. A non‐linear warping algorithm is presented to achieve this. For this case study, physicians were able to contour both the total prostate volume, and intraprostatic nodules. While choosing anatomical tie points along the external prostate surface (needed for the non‐linear warping algorithm), analysis of the nodules revealed the algorithm accuracy ranges from 63–93%. While the algorithm accuracy itself, when reproducing the total prostate volume, achieved an accuracy of 97%.