@MASTERSTHESIS\{IMM2012-06405,
    author       = "H. M. Kjer",
    title        = "{MRI} in Radiation Therapy Planning: The Use of Ultrashort Echo Time Imaging",
    year         = "2012",
    school       = "Technical University of Denmark, {DTU} Informatics, {E-}mail: reception@imm.dtu.dk",
    address      = "Asmussens Alle, Building 305, {DK-}2800 Kgs. Lyngby, Denmark",
    type         = "",
    note         = "{DTU} supervisor: Rasmus R. Paulsen, rrp@imm.dtu.dk, {DTU} Informatics",
    url          = "http://www.imm.dtu.dk/English.aspx",
    abstract     = "Background: Radiation therapy is a technique used in the treatment of a number of cancers. The radiation therapy planning process is currently based on a CTscan of the patient because these images are geometrically accurate and contain the information of the electron densities required for calculating dose plans. Radiation therapy based only on {MRI} would be advantageous in many aspects. The superior soft tissue contrast would for example make the delineations of tumours and organs at risk more accurate. It requires that the information from {CT} can be estimated from {MRI}. Conventional {MRI} sequences lack the ability to visualise compact bone, which hinders proper conversion of {MR} images into {CT} images. Visualisation of compact bone is however made possible with a newly installed {MRI} sequence where images are obtained with ultra short echo times (UTE). The objective of this report is to explore the potential of this sequence for enabling {MRI} based radiation therapy.

Materials and Methods: It should be established how {UTE} acquisitions should be conducted at the department’s 1 Tesla open {MR-}scanner dedicated for radiation therapy. A cut-off knee from a calf was scanned with a range of varying parameters. Regions of muscle and bone tissue were manually annotated on the images. Optimal parameters could be investigated by calculating a contrast-tonoise ratio. Strategies for estimation of {CT} from {MRI} should further be explored. The overall chosen strategy was to segment the {MRI} into different tissue groups and assign an appropriate bulk electron density to each of them with . Four different approaches were investigated and compared to each other. The evaluation was based both on the geometric and dosimetric accuracy compared to {CT,} and was tested on both data from the knee-phantom and on human head anatomy with data from a single patient.

Results: It was shown that the first echo time should be minimized, the second echo time should be placed close to 4 ms, and that the flip angle of the sequence was optimal at 25 degrees. A classification method based on Markov Random Fields was shown to have the overall best performance of the compared methods.

Conclusion: Proper acquisition parameters for {UTE} imaging were established and a relatively successful segmentation approach was implemented. There is room for improvement both concerning image acquisition and segmentations, but the results are promising and good enough for conducting further studies into radiation therapy based only on {MRI}."
}