BACKGROUND: The advantage of MRI-based radiotherapy planning is the superior soft tissue differentiation. However, for accurate patient dose calculations, a conversion of the MR images into Hounsfield CT maps is necessary. The aim of the present study was to investigate the dose accuracy that can be achieved with segmented MR-images derived from the planning CT images by assigning fixed densities to different classes of tissues. METHODS: Treatment plans for ten prostate cancer patients were selected. A collapsed cone algorithm was used to calculate patient dose distributions. The dose calculations were based on four different image sets: (1) the original CT-series (DD(DP)), (2) a simulated MR series with all tissue set to a homogenous water equivalent material of density 1.02 g/cm(3) (DD(W)), (3) a simulated MR series with soft tissue set to a water equivalent material with density 1.02 g/cm(3) and the bone set to a density of 1.3 g/cm(3) (DD(W+B1.3)), and (4) a simulated MR series identical to (3) but with a bone density equal to 2.1 g/cm(3) (DD(W+B2.1)). The dose distributions were compared by analysing dose difference histograms as well as through a visual display of spatial dose deviations. RESULTS: The population based minimum, mean and maximum dose difference between the DD(DP) and DD(W) in the target volume was -2.8, -1.0 and 0.6%, respectively. Corresponding differences between DD(DP) and DD(W+B1.3) were -1.6, 0.2 and 1.5%, respectively, and between DD(DP) and DD(W+B2.1) -4.3, 4.2 and 9.7%, respectively. For the rectum, the differences between CT(DP) and the other image sets were in the range of -19.5 to 8.8%. For the bladder, the differences were in the range of -9.6 to 7.0%. CONCLUSIONS: A systematic study using segmented MR images was undertaken. To achieve an acceptable accuracy in the CTV dose, the MR images should be segmented into bone and water equivalent tissue. Still, significant dose deviation for the organs at risk may be present. As tissue segmentation in real MR images is introduced, segmentation errors and errors that stem from geometrical non-linearities may further reduce the accuracy.
BACKGROUND: The advantage of MRI-based radiotherapy planning is the superior soft tissue differentiation. However, for accurate patient dose calculations, a conversion of the MR images into Hounsfield CT maps is necessary. The aim of the present study was to investigate the dose accuracy that can be achieved with segmented MR-images derived from the planning CT images by assigning fixed densities to different classes of tissues. METHODS: Treatment plans for ten prostate cancerpatients were selected. A collapsed cone algorithm was used to calculate patient dose distributions. The dose calculations were based on four different image sets: (1) the original CT-series (DD(DP)), (2) a simulated MR series with all tissue set to a homogenous water equivalent material of density 1.02 g/cm(3) (DD(W)), (3) a simulated MR series with soft tissue set to a water equivalent material with density 1.02 g/cm(3) and the bone set to a density of 1.3 g/cm(3) (DD(W+B1.3)), and (4) a simulated MR series identical to (3) but with a bone density equal to 2.1 g/cm(3) (DD(W+B2.1)). The dose distributions were compared by analysing dose difference histograms as well as through a visual display of spatial dose deviations. RESULTS: The population based minimum, mean and maximum dose difference between the DD(DP) and DD(W) in the target volume was -2.8, -1.0 and 0.6%, respectively. Corresponding differences between DD(DP) and DD(W+B1.3) were -1.6, 0.2 and 1.5%, respectively, and between DD(DP) and DD(W+B2.1) -4.3, 4.2 and 9.7%, respectively. For the rectum, the differences between CT(DP) and the other image sets were in the range of -19.5 to 8.8%. For the bladder, the differences were in the range of -9.6 to 7.0%. CONCLUSIONS: A systematic study using segmented MR images was undertaken. To achieve an acceptable accuracy in the CTV dose, the MR images should be segmented into bone and water equivalent tissue. Still, significant dose deviation for the organs at risk may be present. As tissue segmentation in real MR images is introduced, segmentation errors and errors that stem from geometrical non-linearities may further reduce the accuracy.
Authors: Sharif Amit Kamran; Khondker Fariha Hossain; Hussein Moghnieh; Sarah Riar; Allison Bartlett; Alireza Tavakkoli; Kenton M Sanders; Salah A Baker Journal: iScience Date: 2022-04-21
Authors: Jidi Sun; Jason A Dowling; Peter Pichler; Joel Parker; Jarad Martin; Peter Stanwell; Jameen Arm; Fred Menk; Peter B Greer Journal: J Appl Clin Med Phys Date: 2015-03-08 Impact factor: 2.102
Authors: F Guerreiro; N Burgos; A Dunlop; K Wong; I Petkar; C Nutting; K Harrington; S Bhide; K Newbold; D Dearnaley; N M deSouza; V A Morgan; J McClelland; S Nill; M J Cardoso; S Ourselin; U Oelfke; A C Knopf Journal: Phys Med Date: 2017-02-24 Impact factor: 2.685
Authors: Jae Hyuk Choi; Danny Lee; Laura O'Connor; Stephan Chalup; James S Welsh; Jason Dowling; Peter B Greer Journal: Front Oncol Date: 2019-10-02 Impact factor: 6.244