PURPOSE: To investigate the feasibility of fully automated detection of fiducial markers implanted into the prostate using portal images acquired with an electronic portal imaging device. METHODS AND MATERIALS: We have made a direct comparison of 4 different methods (2 template matching-based methods, a method incorporating attenuation and constellation analyses and a cross correlation method) that have been published in the literature for the automatic detection of fiducial markers. The cross-correlation technique requires a-priory information from the portal images, therefore the technique is not fully automated for the first treatment fraction. Images of 7 patients implanted with gold fiducial markers (8 mm in length and 1 mm in diameter) were acquired before treatment (set-up images) and during treatment (movie images) using 1MU and 15MU per image respectively. Images included: 75 anterior (AP) and 69 lateral (LAT) set-up images and 51 AP and 83 LAT movie images. Using the different methods described in the literature, marker positions were automatically identified. RESULTS: The method based upon cross correlation techniques gave the highest percentage detection success rate of 99% (AP) and 83% (LAT) set-up (1MU) images. The methods gave detection success rates of less than 91% (AP) and 42% (LAT) set-up images. The amount of a-priory information used and how it affects the way the techniques are implemented, is discussed. CONCLUSIONS: Fully automated marker detection in set-up images for the first treatment fraction is unachievable using these methods and that using cross-correlation is the best technique for automatic detection on subsequent radiotherapy treatment fractions.
PURPOSE: To investigate the feasibility of fully automated detection of fiducial markers implanted into the prostate using portal images acquired with an electronic portal imaging device. METHODS AND MATERIALS: We have made a direct comparison of 4 different methods (2 template matching-based methods, a method incorporating attenuation and constellation analyses and a cross correlation method) that have been published in the literature for the automatic detection of fiducial markers. The cross-correlation technique requires a-priory information from the portal images, therefore the technique is not fully automated for the first treatment fraction. Images of 7 patients implanted with gold fiducial markers (8 mm in length and 1 mm in diameter) were acquired before treatment (set-up images) and during treatment (movie images) using 1MU and 15MU per image respectively. Images included: 75 anterior (AP) and 69 lateral (LAT) set-up images and 51 AP and 83 LAT movie images. Using the different methods described in the literature, marker positions were automatically identified. RESULTS: The method based upon cross correlation techniques gave the highest percentage detection success rate of 99% (AP) and 83% (LAT) set-up (1MU) images. The methods gave detection success rates of less than 91% (AP) and 42% (LAT) set-up images. The amount of a-priory information used and how it affects the way the techniques are implemented, is discussed. CONCLUSIONS: Fully automated marker detection in set-up images for the first treatment fraction is unachievable using these methods and that using cross-correlation is the best technique for automatic detection on subsequent radiotherapy treatment fractions.
Authors: Robert A Ochs; Fereidoun Abtin; Raffi Ghurabi; Ajay Rao; Shama Ahmad; Matthew Brown; Jonathan G Goldin Journal: Acad Radiol Date: 2009-02 Impact factor: 3.173
Authors: Markus K A Herrmann; Tereza Kertesz; Tammo Gsänger; Eugen Bloch; Gerhard Pollul; Mohamed Bouabdallaoui; Arne Strauss; Mareike Herrmann; Hans Christiansen; Hendrik A Wolff; Clemens F Hess; Andrea Hille Journal: Radiat Oncol Date: 2012-02-20 Impact factor: 3.481