Japanese Journal of Clinical Oncology Advance Access published online on July 16, 2007
Japanese Journal of Clinical Oncology, doi:10.1093/jjco/hym050
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2007 Foundation for Promotion of Cancer Research
Feasibility of using MRI alone for 3D Radiation Treatment Planning in Brain Tumors
Department of Radiotherapy, Institute Rotary Cancer Hospital, All India Institute of Medical Sciences, New Delhi, India
For reprints and all correspondence: Ramachandran Prabhakar, Medical Physicist, Department of Radiotherapy, Institute Rotary Cancer Hospital, All India Institute of Medical Sciences, New Delhi 110029, India. E-mail: rampraba{at}rediffmail.com
Received December 4, 2006; accepted January 28, 2007
 |
Abstract |
|---|
 TOP Abstract INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONConflict of interest statementReferences |
|---|
Objective: The aim of this study was to establish whether radiation treatment planning using MRI alone could replace CT-based planning for brain tumors while retaining the dosimetric accuracy. This would help to provide a single imaging modality for both target delineation as well as treatment planning, thus saving time and resources.
Methods: Twenty-five patients with brain tumors were scanned on a spiral CT scanner and 1.5 T MRI scanner. Three treatment plans were generated for all patients. The first plan was generated using the CT scan images with inhomogeneity correction (CT + IC); the second plan used the CT scan without inhomogeneity correction (CT – IC) and the third plan was generated using the MRI scan (MRI alone).
Results: The maximum distortion in the MRI phantom study was less than 1 mm. There were no statistically significant differences in any of the target coverage parameters analysed in this study. Similarly, the maximum antero-posterior and lateral dimensions for the CT-based and MRI-based planning did not show any statistical difference.
Conclusion: MRI-based treatment planning for brain lesions is feasible and gives equivalent dosimetric results compared to CT-based treatment planning.
Key Words: MRI • computed tomography • 3D-CRT • brain tumors
|
INTRODUCTION |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONConflict of interest statementReferences |
|---|
With the introduction of advanced radiation therapy techniques like 3D conformal radiotherapy (3D-CRT) and intensity modulated radiotherapy (IMRT), it has become important to delineate the target volume precisely in order to avoid a geographical miss. Currently, the most suitable sources of detailed 3D anatomic information for treatment planning are the CT and MRI devices and 3D-CRT has been almost entirely based on the images provided by CT. CT images are very useful in providing information regarding the pixel-by-pixel variation of tissue densities. Dose calculation formalism requires anatomic information in terms of electron density ratios, which can be obtained only from CT images in the form of Hounsfield values. Three-dimensional anatomic information obtained from CT images forming a 3D matrix of CT numbers is essential for 3D dose calculations. Although CT has been used as a primary imaging modality for all treatment sites in radiotherapy, the introduction of MRI has challenged the utilization of CT for many tumor sites. One area where MRI targeting has seriously challenged CT is intracranial lesions and its effectiveness in this respect needs to be evaluated. MRI has introduced several added imaging benefits that may confer an advantage over the use of CT, such as improved soft tissue definition and unrestricted multi-planar and volumetric imaging. However, MRI has not yet seriously challenged CT in radiation treatment planning. The reasons for this include: (i) poor imaging of bone; (ii) the lack of electron density information, which is required for dosimetric calculations; (iii) the presence of intrinsic system-related and object-induced MR image distortions (1,2). MRI image data represent quantities such as proton density and nuclear spin relaxation times, which cannot be simply converted to electron density or elemental composition of the body. Although CT scanning provides geometrically precise scans, it gives less detailed tumor and normal tissue anatomic definition in comparison to MRI (3). It was reported about a decade ago that by using MRI in the radiation treatment planning of brain tumors, there was a mean reduction of 30% in field size when compared to CT-based planning (1). However, others have reported a greater tumor volume seen on MRI as opposed to CT images alone (2,4). It can therefore be concluded that MRI shows greater sensitivity in defining target volumes. This will in turn lead to a better dose delivery to the tumor as well as more sparing of normal tissues (5). The greater sensitivity of MRI to variations in tissue proton density and in T1 and T2 relaxation times can be highly valuable for imaging CNS lesions. MRI markedly increases the apparent macroscopic tumor volume from that seen on contrast-enhanced CT alone (6). MRI is a superior modality to assess soft tissue involvement and skull-based lesions whereas CT provides better tumor visualization within bony regions (7). These studies show that MRI is a highly sensitive and preferred imaging modality for most of the brain tumors. Image registration may prove useful in those instances where information from multiple modalities or multiple scans of the same modality is important in planning a given case. CT and MRI fusion for brain itself introduces some error during image registration. CT-MRI fusion needs special software for image fusion and has the potential of introducing errors in the target delineation.
Except in Gamma Knife surgery, MRI-based radiation treatment planning has not been used as an independent procedure and there are no systematic reports published on brain tumor patients using MRI images alone for both target delineation as well as treatment planning. Unlike other sites, the effect of inhomogeneity is comparatively less in brain tumors. By assigning a uniform density to the MRI dataset, MRI alone can be used for radiotherapy treatment planning. A study is thus needed to establish whether MRI-based planning can replace CT-based planning in brain tumors. If so, this would help provide a single imaging modality for both the delineation of target as well as treatment planning and avoid errors in fusing CT and MRI. Before using MRI images alone for radiotherapy treatment planning, it is necessary to examine whether geometrical distortion affects the calculated dose distribution as this may lead to small variations in the dose maximum (±0.5%) (8). Although CT and MRI are the standard imaging modalities for brain tumors, some radiotherapy centers use CT alone for treatment planning in brain tumors. This study was planned to assess the feasibility of using MRI alone as an independent imaging modality for treatment planning.
|
METHODS |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONConflict of interest statementReferences |
|---|
Prior to start of treatment planning study, a phantom study was performed to evaluate the MRI image distortion. The slice thickness used was 2.5 mm. The Leksel MRI phantom was used for the initial quality assurance test on MRI scanner to check for the potential image distortion in the MRI images. It has a plastic container, which simulates the head of the patient. There is an axial grid with an inter-distance of 16 mm between the plastic rods. The linear distortion can be checked on the MRI image by measuring the distance between dots as seen on the MRI slices and comparing them against the geometrically correct values. Non-linearity is present if the dots in the phantom are not aligned vertically and horizontally to each other. The phantom was scanned in the MRI and the image datasets were transferred to the 3D treatment planning system. The inter-distance between the rods was measured with the help of the measuring tool available in the treatment planning system and compared with the known values.
Twenty-five patients with brain tumors were included in this study after obtaining informed consent. The diagnosis of the cases selected is shown in Table 1. For all the patients, immobilization was achieved with a custom-molded thermoplastic sheet (Orfit—Med-Tec). Images for treatment planning were acquired on both a Spiral CT (Siemens Volume Zoom CT) and a 1.5 T MRI scanner (Sonata-1.5 Tesla, Siemens Medical Systems). Three fiducial markers were placed (one anterior and two lateral) as reference markers to define a reference plane. Non-ionic contrast and gadolinium was administered intravenously for all the patients for CT and MRI scanning respectively. The slice thickness for CT and MRI imaging was 2.5 mm and the field of view (FOV) was kept as 280 mm. Using the sequential scanning mode, 60–80 slices were taken for each patient. The CT and the MRI datasets were then transferred to the Eclipse treatment planning system (Varian Medical system) through the DICOM network. Both CT-based and MRI-based treatment planning were performed on the same planning system. Three treatment plans were generated for all patients: first (plan 1) based on the CT images with inhomogeneity correction factor applied to account for the varying electron density (CT + IC), the second (plan 2) based on CT images without any inhomogeneity correction factor (CT–IC), and the third (plan 3) based on the MRI images alone. All three plans used a pencil beam convolution (PBC) algorithm for dose computation. For the first two plans, the CT and MRI images were fused together with the help of the image fusion software (pixel matching technique) available in the Eclipse planning system. The target volume was delineated on the MRI and the planning was done on the 3D CT dataset. In the MRI-based plan, the target was delineated on the MRI image datasets and a uniform density value of zero (Hounsfield value for water) was assigned thereby neglecting the inhomogeneity.
|
The target delineated on the MRI slices was superimposed on the CT dataset such that in all the three arms, the contour and the shape of the target remained exactly the same. The body contour was done separately on CT and MRI-based planning. The maximum possible antero-posterior (AP) and lateral dimensions of the body contour were recorded independently for the CT and the MRI images. These were subsequently compared to look for any appreciable differences between the CT and MRI body contours as this would influence dose calculations. Reliability analysis was applied to the MRI and CT measured maximum AP and lateral dimensions to assess their similarity. All three plans (plan 1, plan 2 and plan 3) were compared using various dosimetric parameters such as mean dose, modal dose, maximum dose, minimum dose, monitor units, IV80 (volume covered by 80% isodose) and IV50 (volume covered by 50% isodose). Plan normalization was done at the isocentre. Two-way ANOVA with repeated measures was used for comparing the dosimetric parameters between the three plans.
|
RESULTS |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONConflict of interest statementReferences |
|---|
The inter-distance values between the rods as measured in the Eclipse planning system after scanning the MR Leksel phantom in the MRI scanner were found to lie between 15.6 and 16.4 mm and thereby validating the accuracy of the MRI-defined anatomy. This indicates that the MRI image distortion was within ±1 mm. The tumor location, total dose delivered and tumor volume delineated on MRI for the 25 patients are shown in Table 1. The mean, median, minimum and maximum volumes of tumors were 36.10, 22.1, 3.25 and 97.3 cm3 respectively.
The reliability test was applied for comparison between the maximum AP and the maximum lateral dimensions recorded for the CT and MRI images. Similarity between the measured AP value on CT and MRI showed the P value to be highly significant (r = 0.9933, 95% CI 0.9848–0.9971, P < 0.00001). Similarly, the P value for the maximum lateral dimension was also highly significant (r = 0.9962, 95% CI 0.9914–0.9983, P < 0.0001) (Table 2). Figures 1 and 2 show the graphical representation of the comparison of the maximum AP and lateral separation distances for the skull contours in individual patients. As can be seen from Figs 1 and 2, the maximum difference between CT- and MRI-defined maximum antero-posterior dimension was 3.6 mm and for the maximum lateral dimension it was 2.7 mm.
|
|
|
Table 3 shows the mean and the range value for the mean dose, modal dose, maximum dose, minimum dose, IV80 (volume covered by 80% isodose) and IV50 (volume covered by 50% isodose). Figure 3 depicts the percentage differences in monitor units for plans 2 and 3 with respect to plan 1 and shows that the maximum difference was 2.3%. The two-way ANOVA by repeated measures was done for mean dose (P value: 0.492), modal dose (P value: 0.416) and IV80 (P value: 0.525). The differences between the three plans were statistically insignificant.
|
|
The comparison of mean values for different dosimetric parameters such as mean dose, median dose, modal dose, maximum dose, minimum dose, IV80, IV50 and monitor units is displayed in Fig. 4. The mean differences between plan 1 and plan 2 for mean dose, modal dose, minimum dose, maximum dose, IV80, IV50 and monitor units were 0.12, 0.27, 0.39, 0.21, 0.25, 0.77 and 1.13 respectively. Similarly the mean differences between plan 1 and plan 3 for mean dose, modal dose, minimum dose, maximum dose, IV80, IV50 and monitor units were 0.05, 0.11, 0.22, 0.39, 0.52, 4.78 and 0.93 respectively. Except for IV50 in plan 3, the mean differences in the dosimetric parameters were statistically insignificant. All the studied dosimetric parameters for plans 2 and 3 were within ±2% of plan 1.
|
|
DISCUSSION |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONConflict of interest statementReferences |
|---|
MRI is generally considered superior to CT for target delineation in brain tumors. CT imaging however continues to be widely used for 3D radiation treatment planning mainly because it provides necessary electron density information for accurate dosimetric calculations. Clinicians and physicists have various concerns for using MRI as the sole planning tool. MRI alone has been used for treatment planning in Gamma Knife radiosurgery. However, in routine radiotherapy treatment planning, either CT alone or CT fused with MRI images is used for treatment planning in brain tumors. The fusion of two different imaging modalities require special software for image fusion. An important consideration is that of the inaccuracy in matching the two imaging modalities (CT and MRI), which may lead to errors in target localization (9). If MRI alone can be used for treatment planning, it may reduce inaccuracies found in CT and MRI fusion, avoid unnecessary financial and physical burden to the patients and reduce the overall time for treatment planning. In our study, we have tried to determine whether dosimetric accuracy can be maintained in treatment plans generated using MRI images alone when electron density information and inhomogeneity correction tools are unavailable.
Because the brain tissue is relatively homogeneous and target volumes are relatively small, lack of tissue heterogeneity data and image distortion is unlikely to significantly affect the dosimetric accuracy. For all 25 patients, the dose parameters were recorded while keeping the target volumes the same in each of the plans. The body contour, however, was drawn independently in CT and MRI slices in this study using the auto-contouring tool available in the planning system. The body contour and its dimensions can significantly influence the dose distributions in 3D treatment planning. In our study, we found that the contours (as judged by the maximum AP and lateral dimensions) did not differ appreciably in the CT and the MRI images (similarity test between CT and MRI: P < 0.0001) and a very appreciable intra-class correlation (Table 2). This finding is important because it means that MRI images alone can be safely used for all stages of radiation treatment planning in brain tumors including body contouring, target delineation and subsequent dose characteristic evaluation.
In the case of pituitary tumors, both MRI and contrast-enhanced CT are effective in defining the target volume. However MRI may provide more information about the precise extent of the lesion and involvement of adjacent structures. In many instances, the sensitivity of MRI exceeds that of CT. With the advancements in MRI, regions of gadolinium enhancement on MRI T1-weighted scans can be compared to the contrast enhancing CT tumor volumes, while abnormalities detected on MRI T2-weighted scans are the counterpart of the CT-defined edema. Zhao et al. (10) have postulated that it may be feasible to detect tumor hypoxia in vivo with MRI studies. Thus MRI appears to be a promising imaging modality for target delineation and radiation therapy planning for brain tumors.
While CT images are usually regarded as geometrically correct, MRI images are known to suffer from geometrical distortion. Advantages of MRI over CT include (i) superior soft tissue contrast (ii) better definition of tumor margins (iii) multi-planar imaging (iv) study of physiology without disturbing the patient and (v) no radiation exposure. Practical issues that limit wider applicability of MRI include availability, increased time and cost, motion artifacts. Besides it cannot be used in patients with metallic devices or in those who suffer from claustrophobia. As far as tissue inhomogeneity correction is concerned, it can be neglected by assigning the Hounsfield value 0 (water) as in our study.
Currently, CT and MRI fusion has been used extensively in treatment planning for brain tumors (2). The main disadvantages of using MRI for radiation treatment planning are the lack of electron density information and the potential presence of distortions resulting in geometrical inaccuracies. Bone is only imaged negatively, though an absence of signal may be distinguished from air spaces. To date, no method has been found to derive electron densities from MRI imaging data. Ramsey et al. (11) used the RANDO head phantom to compare CT-based and MRI-based treatment planning and concluded that MRI images provided homogeneous treatment planning with a dosimetric accuracy of ±2% but for treatment plans in areas where the beam passes through large air cavities such as maxillary sinus it provides an unacceptable dosimetric error of 2–4%. In our case, most of the tumors were treated with two lateral and one vertex fields and none of our fields passed through large air cavities such as maxillary sinus. In the case of pituitary tumors, a portion of the two lateral fields passed through the posterior ethmoid sinus and even in this case our dosimetric parameters were within 2% for all three plans.
MRI image distortion is one of the serious concerns when MRI is used for treatment planning and several authors have addressed this issue (12–16). Image distortions arise from both system-related effects and object-induced effects. Any potential image distortions must be quantified and corrected for, if necessary (2). Spatial distortions in MRI images vary with magnetic field strength and with image acquisition protocol and spatial accuracy generally decreases as the distance from the magnet isocenter increases. Imaging protocols with high gradient bandwidths should be used to reduce the spatial distortions (7). System-related distortions arise as a result of inhomogeneities in the main magnetic and gradient fields, and have been found to be inversely proportional to the gradient field strength (17). System-induced distortions are more pronounced at the edges of the radial field and therefore can be expected to be minimal if only the central region of the magnet bore is utilized. This situation can easily be arranged in RT planning of the brain. Maximum distortion errors of 2 mm for FOV less than 20 cm have been reported (14,18). In this situation, image correction could be deemed unnecessary. For larger FOVs, image distortions are much more appreciable, but this situation rarely arises in treatment planning in brain tumors. Yanke et al. (19) have also discussed the above issue and have reported similar findings. Major et al. (20) used a humanoid phantom for CT and MRI scanning and concluded that geometric distortion was negligible (<2 mm). In a study by Moerland et al. (14) for small FOV less than 20 cm (brain) maximum distortion of 2 mm was observed, which correlated with the current study.
To the best of our knowledge, this is the first study of its kind reporting a systematic comparative analysis of CT-based and MRI-based radiation treatment planning in brain tumors aimed at assessing the effect of lack of tissue inhomogeneity data on dosimetric parameters. MRI-based treatment planning has also been studied for other sites such as the prostate (9,21). These studies have reported that radiation treatment planning for prostate cancer using magnetic resonance imaging alone is feasible.
Our results showed that the difference between the CT and MRI-based treatment plans for mean dose, maximum dose and modal dose was within ±2%. Differences in the isodose volumes and monitor units, while using MRI alone in radiotherapy planning of brain tumors, were statistically insignificant. This study clearly shows that MRI alone is feasible for radiotherapy treatment planning for brain tumors.
|
CONCLUSION |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONConflict of interest statementReferences |
|---|
The introduction of MRI in radiotherapy treatment planning has seriously challenged CT. The comparison of dosimetric parameters such as maximum dose, minimum dose, mean dose, modal dose, antero–posterior separation, lateral separation, volume covered by 50% isodose and 80% isodose shows that MRI alone can be used for treatment planning in brain tumors. We conclude that for small volumes, the image distortion in MRI is less than 1 mm and the inhomogeneity corrections can be neglected for brain tumors. The difference in dose volume parameters between CT-based and MRI-based planning was not statistically significant and the dosimetric variations were within ±2%. Therefore, the utilization of MRI alone for radiotherapy treatment planning in brain tumors is a practically feasible option.
|
Conflict of interest statement |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONConflict of interest statementReferences |
|---|
None declared.
|
References |
|---|
|
TOPAbstractINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONConflict of interest statementReferences |
|---|
1 Websy G, Adamis MK, Edelmann RR. Artifacts in MRI: description, causes and solutions. In: Clinical Magnetic Resonance Imaging—Edelmann RR, Hesselink JK, Zlatkin MB, eds. (1996) Philadelphia, PA: Saunders. 88–144.
2 Khoo VS, Dearnaley DP, Finnigan DJ, Padhani A, Tanner SF, Leach MO. Magnetic resonance imaging (MRI): considerations and applications in radiotherapy treatment planning. Radiother Oncol (1997) 42:1–15.[CrossRef][ISI][Medline]
3 Kent DL, Haynor DR, Longstreth WT, Larson EB. The clinical efficacy of magnetic resonance imaging in neuroimaging. Ann Intern Med (1994) 120:856–71.
4 Ten Haken RK, Thornton AF, Sandler HM, LaVigne ML, Quint DJ, Fraass BA, et al. A quantitative assessment of the addition of MRI to CT-based, 3-D treatment planning of brain tumors. Radiother Oncol (1992) 25:121–33.[CrossRef][ISI][Medline]
5 Kessler ML, Ten Haken RK, Fraas BA, McShan DL. Expanding the use and effectiveness of dose volume histograms for 3D treatment planning. I: Integration of 3-D dose-display. Int J Radiat Oncol Biol Phys (1994) 29:1125–31.[ISI][Medline]
6 Thornton AF, Sandler HM, Ten Haken RK, McShan DL, Fraass BA, La Vigne ML, et al. The clinical utility of magnetic resonance imaging in 3-dimensional treatment planning of brain neoplasms. Int J Radiat Oncol Biol Phys (1992) 24:767–75.[ISI][Medline]
7 Khoo VS, Adams EJ, Saran F, Bedford JL, Perks JR, Warrington AP, et al. A comparison of clinical target volumes determined by CT and MRI for the radiotherapy planning of base of skull meningiomas. Int J Radiat Oncol Biol Phys (2000) 46:1309–17.[CrossRef][ISI][Medline]
8 Beavis AW, Gibbs P, Dealey RA, Whitton VJ. Radiotherapy treatment planning of brain tumors using MRI alone. In: Br J Radiol (1998) 71:544–8.[Abstract]
9 Chen L, Price RA, Wang L, Li J, Qin L, McNeeley S, et al. MRI-based treatment planning for radiotherapy: dosimetric verification for prostate IMRT. Int J Radiat Oncol Biol Phys (2004) 60:636–47.[CrossRef][ISI][Medline]
10 Zhao D, Ran S, Constantinescu A, Hahn EW, Mason RP. Tumor oxygen dynamics: correlation of in vivo MRI with histological findings. Neoplasia (2003) 5:308–18.[ISI][Medline]
11 Ramsey CR, Oliver AL. Magnetic resonance imaging based digitally reconstructed radiographs, virtual simulation, and three-dimensional treatment planning for brain neoplasms. Med Phys (1998) 25:1928–34.[CrossRef][ISI][Medline]
12 Fransson A, Andreo P, Potter R. Aspects of MR image distortions in radiotherapy treatment planning. Strahlenther Onkol (2001) 177:59–73.[CrossRef][ISI][Medline]
13 Prott FJ, Haverkamp U, Eich H, Resch A, Micke O, Fischedick AR, et al. Effect of distortions and asymmetry in MR images on radiotherapeutic treatment planning. Int J Cancer (2000) 90:46–50.[CrossRef][ISI][Medline]
14 Moerland MA, Beersma R, Bhagwandien R, Wijrdeman HK, Bakker CJ. Analysis and correction of geometric distortions in 1.5 T magnetic resonance images for use in radiotherapy treatment planning. Phys Med Biol (1995) 40:1651–4.[CrossRef][ISI][Medline]
15 Mizowaki T, Nagata Y, Okajima K, Murata R, Yamamoto M, Kokubo M, et al. Development of an MR simulator: experimental verification of geometric distortion and clinical application. Radiology (1996) 199:855–60.
16 Petersch B, Bogner J, Fransson A, Lorang T, Potter R. Effects of geometric distortion in 0.2T MRI on radiotherapy treatment planning of prostate cancer. Radiother Oncol (2004) 71:55–64.[CrossRef][ISI][Medline]
17 O'Donnell M, Edelstein WA. NMR imaging in the presence of magnetic field inhomogeneities and gradient field inhomogeneities. Med Phys (1985) 12:20–6.[CrossRef][ISI][Medline]
18 Hill DLG, Hawkes DJ, Gleeson MJ, Cox TC, Strong AJ, Wong WL, et al. Accurate frameless registration of MR and CT images of the head: applications in planning surgery and radiation therapy. Radiology (1994) 191:447–54.
19 Yanke BR, Ten Haken RK, Aisen A, Fraass BA, Thornton AF Jr. Design of MRI scan protocols for use in 3-D, CT-based treatment planning. Med Dosim (1991) 16:205–11.[Medline]
20 Major T, Petranyi A, Varjas G, Nemeth G. The possibility of the use of MRI images in the three-dimensional external radiotherapy treatment planning. Magy Onkol (2002) 46:239–45.[Medline]
21 Lee YK, Bollet M, Edwards GC, Flower MA, Leach MO, McNair H, et al. Radiotherapy treatment planning of prostate cancer using magnetic resonance imaging alone. Radiother Oncol (2003) 66:203–16.[CrossRef][ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||