Clinical application of digitally-reconstructed radiographs generated from magnetic resonance imaging for intracranial lesions

MRI-Only Radiotherapy Planning for Nasopharyngeal Carcinoma Using Deep Learning

Background

Radical radiotherapy is the main treatment modality for early and locally advanced nasopharyngeal carcinoma (NPC). Magnetic resonance imaging (MRI) has the advantages of no ionizing radiation and high soft-tissue resolution compared to computed tomography (CT), but it does not provide electron density (ED) information for radiotherapy planning. Therefore, in this study, we developed a pseudo-CT (pCT) generation method to provide necessary ED information for MRI-only planning in NPC radiotherapy.

Methods

Twenty patients with early-stage NPC who received radiotherapy in our hospital were investigated. First, 1433 sets of paired T1 weighted magnetic resonance (MR) simulation images and CT simulation images were rigidly registered and preprocessed. A 16-layer U-Net was used to train the pCT generative model and a "pix2pix" generative adversarial network (GAN) was also trained to compare with the pure U-Net regrading pCT quality. Second, the contours of all target volumes and organs at risk in the original CT were transferred to the pCT for planning, and the beams were copied back to the original CT for reference dose calculation. Finally, the dose distribution calculated on the pCT was compared with the reference dose distribution through gamma analysis and dose-volume indices.

Results

The average time for pCT generation for each patient was 7.90 ± 0.47 seconds. The average mean (absolute) error was -9.3 ± 16.9 HU (102.6 ± 11.4 HU), and the mean-root-square error was 209.8 ± 22.6 HU. There was no significant difference between the pCT quality of pix2pix GAN and that of pure U-Net (p > 0.05). The dose distribution on the pCT was highly consistent with that on the original CT. The mean gamma pass rate (2 mm/3%, 10% low dose threshold) was 99.1% ± 0.3%, and the mean absolute difference of nasopharyngeal PGTV D_99% and PTV V_95% were 0.4% ± 0.2% and 0.1% ± 0.1%.

Conclusion

The proposed deep learning model can accurately predict CT from MRI, and the generated pCT can be employed in precise dose calculations. It is of great significance to realize MRI-only planning in NPC radiotherapy, which can improve structure delineation and considerably reduce additional imaging dose, especially when an MR-guided linear accelerator is adopted for treatment.

Systematic Review of Synthetic Computed Tomography Generation Methodologies for Use in Magnetic Resonance Imaging-Only Radiation Therapy

Magnetic resonance imaging (MRI) offers superior soft-tissue contrast as compared with computed tomography (CT), which is conventionally used for radiation therapy treatment planning (RTP) and patient positioning verification, resulting in improved target definition. The 2 modalities are co-registered for RTP; however, this introduces a systematic error. Implementing an MRI-only radiation therapy workflow would be advantageous because this error would be eliminated, the patient pathway simplified, and patient dose reduced. Unlike CT, in MRI there is no direct relationship between signal intensity and electron density; however, various methodologies for MRI-only RTP have been reported. A systematic review of these methods was undertaken. The PRISMA guidelines were followed. Embase and Medline databases were searched (1996 to March, 2017) for studies that generated synthetic CT scans (sCT)s for MRI-only radiation therapy. Sixty-one articles met the inclusion criteria. This review showed that MRI-only RTP techniques could be grouped into 3 categories: (1) bulk density override; (2) atlas-based; and (3) voxel-based techniques, which all produce an sCT scan from MR images. Bulk density override techniques either used a single homogeneous or multiple tissue override. The former produced large dosimetric errors (>2%) in some cases and the latter frequently required manual bone contouring. Atlas-based techniques used both single and multiple atlases and included methods incorporating pattern recognition techniques. Clinically acceptable sCTs were reported, but atypical anatomy led to erroneous results in some cases. Voxel-based techniques included methods using routine and specialized MRI sequences, namely ultra-short echo time imaging. High-quality sCTs were produced; however, use of multiple sequences led to long scanning times increasing the chances of patient movement. Using nonroutine sequences would currently be problematic in most radiation therapy centers. Atlas-based and voxel-based techniques were found to be the most clinically useful methods, with some studies reporting dosimetric differences of <1% between planning on the sCT and CT and <1-mm deviations when using sCTs for positional verification.

Clinical evaluation of 3D/3D MRI-CBCT automatching on brain tumors for online patient setup verification - A step towards MRI-based treatment planning

BACKGROUND

Magnetic Resonance Imaging (MRI) is often used in modern day radiotherapy (RT) due to superior soft tissue contrast. However, treatment planning based solely on MRI is restricted due to e.g. the limitations of conducting online patient setup verification using MRI as reference. In this study 3D/3D MRI-Cone Beam CT (CBCT) automatching for online patient setup verification was investigated.

MATERIAL AND METHODS

Initially, a multi-modality phantom was constructed and used for a quantitative comparison of CT-CBCT and MRI-CBCT automatching. Following the phantom experiment three patients undergoing postoperative radiotherapy for malignant brain tumors received a weekly CBCT. In total 18 scans was matched with both CT and MRI as reference. The CBCT scans were acquired using a Clinac iX 2300 linear accelerator (Varian Medical Systems) with an On-Board Imager (OBI).

RESULTS

For the phantom experiment CT-CBCT and MRI-CBCT automatching resulted in similar results. A significant difference was observed only in the longitudinal direction where MRI-CBCT resulted in the best match (mean and standard deviations of 1.85±2.68 mm for CT and -0.05±2.55 mm for MRI). For the clinical experiment the absolute difference in couch shift coordinates acquired from MRI-CBCT and CT-CBCT automatching, were ≤2 mm in the vertical direction and ≤3 mm in the longitudinal and lateral directions. For yaw rotation differences up to 3.3 degrees were observed. Mean values and standard deviations were 0.8±0.6 mm, 1.5±1.2 mm and 1.2±1.2 mm for the vertical, longitudinal and lateral directions, respectively and 1.95±1.12 degrees for the rotation (n=17).

CONCLUSION

It is feasible to use MRI as reference when conducting 3D/3D CBCT automatching for online patient setup verification. For both the phantom and clinical experiment MRI-CBCT performed similar to CT-CBCT automatching and significantly better in the longitudinal direction for the phantom experiment.

MRI-based Treatment Planning with Electron Density Information Mapped from CT Images: A Preliminary Study

Nowadays magnetic resonance imaging (MRI) has been profoundly used in radiotherapy (RT) planning to aid the contouring of targets and critical organs in brain and intracranial cases, which is attributable to its excellent soft tissue contrast and multi-planar imaging capability. However, the lack of electron density information in MRI, together with the image distortion issues, precludes its use as the sole image set for RT planning and dose calculation. The purpose of this preliminary study is to probe the feasibility and evaluate an MRI-based radiation dose calculation process by providing MR images the necessary electron density (ED) information from a patient's readily available diagnostic/staging computed tomography (CT) images using an image registration model. To evaluate the dosimetric accuracy of the proposed approach, three brain and three intracranial cases were selected retrospectively for this study. For each patient, the MR images were registered to the CT images, and the ED information was then mapped onto the MR images by in-house developed software generating a modified set of MR images. Another set of MR images with voxel values assigned with the density of water was also generated. The original intensity modulated radiation treatment (IMRT) plan was then applied to the two sets of MR images and the doses were calculated. The dose distributions from the MRI-based calculations were compared to that of the original CT-based calculation. In all cases, the MRI-based calculations with mapped ED yielded dose values very close (within 2%) to that of the CT-based calculations. The MRI-based calculations with voxel values assigned with water density indicated a dosimetric error of 3–5%, depending on the treatment site. The present approach offers a means of utilizing MR images for accurate dose calculation and affords a potential to eliminate the redundant simulation CT by planning a patient's treatment with only simulation MRI and any available diagnostic/staging CT data.

Dosimetric and geometric evaluation of an open low-field magnetic resonance simulator for radiotherapy treatment planning of brain tumours

BACKGROUND AND PURPOSE

Magnetic resonance (MR) imaging is superior to computed tomography (CT) in radiotherapy of brain tumours. In this study an open low-field MR-simulator is evaluated in order to eliminate the cost of and time spent on additional CT scanning.

MATERIALS AND METHODS

Eleven patients with brain tumours are both CT and MR scanned and the defined tumour volumes are compared. Image distortions and dose calculations based on CT density correction, MR unit density and MR bulk density, bone segmentation are performed. Monte Carlo simulations using 4 and 8 MV beams on homogeneous and bone segmented mediums are performed.

RESULTS

Mean MR and CT tumour volumes of approximately the same size (V MR =55+/-34 cm3 and V CT =51+/-32 cm3) are observed, but for individual patients, small intersection volumes are observed. The MR images show negligible distortion within radial distances below 12 cm (<1.5 mm). On unit density mediums, dose errors above 2% are observed in low dose areas. Monte Carlo simulations with 4 MV photons show large deviations in dose (>2%) just behind the skull if bone is not segmented.

CONCLUSIONS

It is feasible to use an MR-simulator for radiotherapy planning of brain tumours if bone is segmented or a careful choice of beam energy (>4 MV) is selected.

Advanced image-guided external beam radiotherapy

The goal of radiation therapy is to eradicate tumor stem cells while sparing healthy tissue. Therefore, the first aim must be to delineate tumor from healthy tissue. Advanced imaging techniques will enable one to reduce the uncertainty of microscopic extension of disease. Ultimately, advanced functional imaging systems correlated with image-registered pathological specimens will allow one to delineate disease extent from normal tissue at the tumor periphery. When it is not possible to determine the CTV margin with reasonable certainty, the margins must remain generous and conformal avoidance methodology could and should be deployed to spare critical normal structures. Of equal importance to defining the CTV is the need to guarantee that this target is indeed treated. For this purpose, image guidance using a variety of systems including portal images, ultrasound devices, and CT scanners at the time of treatment has been implemented. Some image-guided methods, portal images for instance, are more amenable for use with rigid structures such as encountered in the sinus whereas others like ultrasound or CT scanners are able to account for nonrigid setup variations. Several strategies for preventing organ motion from degrading the precision that radiotherapy offers have been described. In particular, a CT scan at the time of treatment delivery can also be used as the basis to reconstruct the dose received by the patient. Dose reconstruction will allow the dose just delivered to be superimposed on the pretreatment CT scan and will allow one to compare the reconstructed delivered dose distribution with the planned dose distribution to assess discrepancies between these. Furthermore, reconstruction of the delivered dose distributions holds the promise of allowing one to accumulate dose delivered to the tumor and normal structures on a fraction per fraction basis. This will ultimately allow for the determination of treatment-specific tumor control probabilities and normal tissue complication probabilities.

Evaluation of similarity measures for reconstruction-based registration in image-guided radiotherapy and surgery

PURPOSE

A promising patient positioning technique is based on registering computed tomographic (CT) or magnetic resonance (MR) images to cone-beam CT images (CBCT). The extra radiation dose delivered to the patient can be substantially reduced by using fewer projections. This approach results in lower quality CBCT images. The purpose of this study is to evaluate a number of similarity measures (SMs) suitable for registration of CT or MR images to low-quality CBCTs.

METHODS AND MATERIALS

Using the recently proposed evaluation protocol, we evaluated nine SMs with respect to pretreatment imaging modalities, number of two-dimensional (2D) images used for reconstruction, and number of reconstruction iterations. The image database consisted of 100 X-ray and corresponding CT and MR images of two vertebral columns.

RESULTS

Using a higher number of 2D projections or reconstruction iterations results in higher accuracy and slightly lower robustness. The similarity measures that behaved the best also yielded the best registration results. The most appropriate similarity measure was the asymmetric multi-feature mutual information (AMMI).

CONCLUSIONS

The evaluation protocol proved to be a valuable tool for selecting the best similarity measure for the reconstruction-based registration. The results indicate that accurate and robust CT/CBCT or even MR/CBCT registrations are possible if the AMMI similarity measure is used.

Opacity transfer function optimization for volume-rendered computed tomography images of the prostate

RATIONALE AND OBJECTIVES

The selection of an opacity transfer function is essential for volume visualization. Computed tomography (CT) scans of the pelvis were used to determine an optimal opacity transfer function for use in radiotherapy.

MATERIALS AND METHODS

On sample datasets (a mathematical phantom and a patient pelvis CT scan), standard viewing orientations were selected to render the prostate. Opacity functions were selected via (1) trapezoidal manual selection, (2) trapezoidal semiautomatic selection, and (3) histogram volume-based selection. Using an established metric, the errors using each of these methods were computed.

RESULTS

Trapezoidal manual opacity function optimization resulted in visually acceptable images, but the errors were considerable (6.3-9.1 voxel units). These errors could be reduced with the use of trapezoidal semiautomatic selection (4.9-6.2 voxel units) or with histogram volume-based selection (4.8-7.9 voxel units). As each visualization algorithm focused on enhancing the boundary of the prostate using a different approach, the scene information was considerably different using the three techniques.

CONCLUSION

Improved volume visualization of soft tissue interfaces was achieved using automated optimal opacity function determination, compared with manual selection.

Image guidance for precise conformal radiotherapy

PURPOSE

To review the state of the art in image-guided precision conformal radiotherapy and to describe how helical tomotherapy compares with the image-guided practices being developed for conventional radiotherapy.

MATERIALS AND METHODS

Image guidance is beginning to be the fundamental basis for radiotherapy planning, delivery, and verification. Radiotherapy planning requires more precision in the extension and localization of disease. When greater precision is not possible, conformal avoidance methodology may be indicated whereby the margin of disease extension is generous, except where sensitive normal tissues exist. Radiotherapy delivery requires better precision in the definition of treatment volume, on a daily basis if necessary. Helical tomotherapy has been designed to use CT imaging technology to plan, deliver, and verify that the delivery has been carried out as planned. The image-guided processes of helical tomotherapy that enable this goal are described.

RESULTS

Examples of the results of helical tomotherapy processes for image-guided intensity-modulated radiotherapy are presented. These processes include megavoltage CT acquisition, automated segmentation of CT images, dose reconstruction using the CT image set, deformable registration of CT images, and reoptimization.

CONCLUSIONS

Image-guided precision conformal radiotherapy can be used as a tool to treat the tumor yet spare critical structures. Helical tomotherapy has been designed from the ground up as an integrated image-guided intensity-modulated radiotherapy system and allows new verification processes based on megavoltage CT images to be implemented.

Three-dimensional conformal radiation treatment planning and delivery for low- and intermediate-grade gliomas

Three-Dimensional conformal radiation treatment (3D-CRT) planning and delivery is an external beam radiation therapy modality that has the general goal of conforming the shape of a prescribed dose volume to the shape of a 3-dimensional target volume, simultaneously limiting dose to critical normal structures. 3-Dimensional conformal therapy should include at least one volumetric imaging study of the patient. This image should be obtained in the treatment position for visualizing the target and normal anatomic structures that are potentially within the irradiated volume. Most often, computed tomography (CT) and/or magnetic resonance imaging (MRI) are used; however, recently, other imaging modalities such as functional MRI, MR spectroscopy, and positron emission tomography (PET) scans have been used to visualize the clinically relevant volumes. This article will address the clinically relevant issues with regard to low- and intermediate-grade gliomas and the role of 3D-CRT planning. Specific issues that will be addressed will include normal tissue tolerance, target definition, treatment field design in regard to isodose curves and dose-volume histograms, and immobilization.