Patch‐based generation of a pseudo CT from conventional MRI sequences for MRI‐only radiotherapy of the brain

Abstract

Purpose: In radiotherapy (RT) based on magnetic resonance imaging (MRI) as the only modality, the information on electron density must be derived from the MRI scan by creating a so-called pseudo computed tomography (pCT). This is a nontrivial task, since the voxel-intensities in an MRI scan are not uniquely related to electron density. To solve the task, voxel-based or atlas-based models have typically been used. The voxel-based models require a specialized dual ultrashort echo time MRI sequence for bone visualization and the atlas-based models require deformable registrations of conventional MRI scans. In this study, we investigate the potential of a patch-based method for creating a pCT based on conventional T 1 -weighted MRI scans without using deformable registrations. We compare this method against two state-of-the-art methods within the voxel-based and atlas-based categories. Methods: The data consisted of CT and MRI scans of five cranial RT patients. To compare the performance of the different methods, a nested cross validation was done to find optimal model parameters for all the methods. Voxel-wise and geometric evaluations of the pCTs were done. Furthermore, a radiologic evaluation based on water equivalent path lengths was carried out, comparing the upper hemisphere of the head in the pCT and the real CT. Finally, the dosimetric accuracy was tested and compared for a photon treatment plan. Results: The pCTs produced with the patch-based method had the best voxel-wise, geometric, and radiologic agreement with the real CT, closely followed by the atlas-based method. In terms of the dosimetric accuracy, the patch-based method had average deviations of less than 0.5% in measures related to target coverage. Conclusions: We showed that a patch-based method could generate an accurate pCT based on conventional T 1 -weighted MRI sequences and without deformable registrations. In our evaluations, the method performed better than existing voxel-based and atlas-based methods and showed a promising potential for RT of the brain based only on MRI. C 2015 American Association of Physicists in Medicine. [http://dx

Monte Carlo calculation based on hydrogen composition of the tissue for MV photon radiotherapy

“…Current methods to derive electron density from MRI data by generating a pseudo‐CT are divided into two approaches: the anatomy‐based and the voxel‐based methods. In the anatomy‐based methods, deformable registration of reference paired MRI/CT scan (atlas or reference patient) to the MRI scan of the patient under consideration is used to build a pseudo‐CT from the warped reference CT scan 19 , 20 , 21 . Although mathematical or statistical operations could be used to smooth registration errors due to interpatient anatomical differences, 19 , 20 these methods fail for patients presenting high anatomy dissimilarities.…”

“…In the anatomy‐based methods, deformable registration of reference paired MRI/CT scan (atlas or reference patient) to the MRI scan of the patient under consideration is used to build a pseudo‐CT from the warped reference CT scan 19 , 20 , 21 . Although mathematical or statistical operations could be used to smooth registration errors due to interpatient anatomical differences, 19 , 20 these methods fail for patients presenting high anatomy dissimilarities. The voxel‐based methods convert MRI intensity into pseudo‐CT number using either a bulk density assignment after segmenting tissues or using a regression model to compute continuous values of CT number from a set of MRI data 22 , 23 , 24 , 25 .…”

Do we need a new CT scan for retreatment of intracranial SRS patients?

“…In this case, localization was typically provided by stereotactic head frame. Efforts have been made to create “synthetic” or “pseudo” CTs from MR images that can be used for dose calculation and localization 7, 8, 9, 10, 11, 12, 13, 14. These methods show promise in terms of geometric and dosimetric accuracy, but are not in widespread clinical use at this time.…”

CT metal artifact reduction using MR image patches

“…This renders the method fully automatic and optimises the algorithm for each individual patient. Since the resulting method only requires data from the patient targetted for MAR, it avoids any inter-subject issues that arise during pCT generation 14,16 such as the requirement for MR intensity normalisation [17][18][19] or time-consuming atlas registration of the target patient. We will refer to our method as "kernel regression MAR" (kerMAR) in the remainder.…”

“…We therefore propose in this paper a novel MR-based MAR algorithm that combines MR information in larger larger spatial neighborhoods with the local, corrupted CT values to make more accurate CT value predictions. Our approach is based on methods developed for pseudo-CT generation in MR-only radiotherapy 14 and PET/MR attenuation correction 15 applications. Here, CT values are predicted from image patches (clusters of neighboring voxels) in coregistered MR scans, using regression models learned from a database of matching MR patches and CT values.…”