First published online October 26, 2006
British Journal of Radiology (2007) 80, 347-354
© 2007 British Institute of Radiology
doi: 10.1259/bjr/65349468
Use of MR spectroscopy and functional imaging in the treatment planning of gliomas
A Narayana, MD
1
J Chang, PhD
2
S Thakur, PhD
2
W Huang, PhD
2,3
S Karimi, MD
3
B Hou, PhD
2,3
A Kowalski, MS
2
G Perera, MS
2
A Holodny, MD
3 and
P H Gutin, MD
4,5
Departments of 1 Radiation Oncology, 2 Medical Physics, 3 Radiology and 4 Surgery, Memorial Sloan-Kettering Cancer Center and 5 Department of Neuro-Surgery, Weill Medical College of Cornell University, 1275 York Avenue, New York, NY 10021, USA
Correspondence: Ashwatha Narayana, Department of Radiation Oncology, New York University Medical Center, 550, 1st Avenue, New York, NY 11016, USA. E-mail: ashwatha.narayana{at}nyumc.org
 |
Abstract
|
|---|
Routine anatomical imaging with CT and MRI does not reliably indicate the true extent or the most malignant areas of gliomas and cannot identify the functionally critical parts of the brain. The aim of the study was to see if the use of MR spectroscopic imaging (MRSI) along with functional MRI (fMRI) can better define both the target and the critical structures to be avoided to improve radiation delivery in gliomas. 12 patients with gliomas underwent multivoxel MRS and functional imaging using GE processing software. The choline to creatine ratio (Cho:Cr), which represents the degree of abnormality for each individual voxel on MRSI, was derived, converted into a grayscale grading system, fused to the MRI images and then transferred to the planning CT images. An intensity-modulated radiation therapy (IMRT) plan was developed using the dose constraints based on both the anatomical and the functionally critical regions. Cho:Cr consistently identified the gross tumour volume (GTV) within the microscopic disease (clinical target volume, CTV) and allowed dose painting using IMRT. No correlation between MRSI based Cho:Cr 
2 and MR defined CTV nor their location was noted. However, MRSI defined Cho:Cr 3 was smaller by 40% compared with post-contrast T1 weighted MRI defined GTV volumes. fMRI helped in optimizing the orientation of the beams. In conclusion, both MRSI and fMRI provide additional information to conventional imaging that may guide dose painting in treatment planning of gliomas. A Phase I IMRT dose intensification trial in gliomas using this information is planned.
|
Introduction
|
|---|
The local control for patients with gliomas remains poor following either conventional or three-dimensional (3D) conformal external beam radiation therapy [1]. Intensity-modulated radiation therapy (IMRT) that relies on advanced accelerator and multileaf collimator technology to deliver non-uniform beam intensities may have the ability to increase dose to tumour volume without increasing the dose to normal tissue and holds the promise of improving local control [2]. To maximize its benefit, areas of high grade tumour must be identified separately from low grade tumour and normal brain tissue. The conventional treatment approach in gliomas is to deliver 45–60 Gy to the MRI defined volume plus a margin of 2–3 cm [3, 4]. However, target definition based on MRI information has its limitations due to the presence of non-enhancing tumour or contrast-enhancing necrosis and may contribute to local failure [5].
Magnetic resonance spectroscopic imaging (MRSI) is an emerging powerful tool for target definition in brain tumours [6, 7]. It provides regional information about tumour pathology based on the levels of cellular metabolites, including choline (Cho), creatine (Cr), N-acetylaspartate (NAA) and lactate/lipid (Lac). Studies have shown that changes in metabolite levels may help to differentiate normal from abnormal tissue in patients with glioma [8, 9]. These studies have shown that gliomas exhibit a high resonance in the spectral region of Cho and a low NAA and Cr resonance, implying increases in the Cho:Cr and/or Cho:NAA ratios [8]. The ability of MRSI to show areas of high grade tumour within low grade gliomas has also been noted [9].
Functional magnetic resonance imaging (fMRI) is an imaging technique that involves using a gradient echo-echo planar imaging (GE-EPI) sequence to define the location of functionally eloquent cortices, such as the motor cortex, Broca's area, Wernicke's area, the visual cortex etc. in the brain [10]. A preliminary study by Liu et al in three patients who underwent stereotactic radiosurgery demonstrated that the radiation dose can be reduced to an fMRI defined eloquent cortex, without changing the dose to the tumour [11]. The use of fMRI can possibly allow a radiation oncologist to properly plan and deliver an adequate radiation dose while avoiding damage to the adjacent functional cortices in the treatment of gliomas.
The variability in tumour cell distribution and inability to visualize functionally important regions within the brain make it difficult to accurately define appropriate target volumes and normal tissue regions for purposes of treatment planning using routine MRI alone. The primary goal of this study was to examine the feasibility of incorporating functional and spectroscopic imaging into the treatment planning process. The secondary aim was to examine the resulting changes to the image defined radiation target volume when compared with conventional treatment planning.
|
Methods and materials
|
|---|
MRI
MRI examinations were performed on a 1.5 Tesla GE LX or Excite MR scanner (General Electric Medical Systems, Milwaukee, WI) using a quadrature head coil. The diagnostic MRI protocol included pre-contrast axial T2 and T1 weighted MRI, axial fluid-attenuated inversion recovery (FLAIR) MRI with either 3 mm or 4.5 mm slice thickness and 0 mm gap and post-contrast T1 weighted sequences in axial, sagittal and coronal planes.
MR spectroscopy
Following FLAIR image acquisition, but prior to contrast administration, 3D MRSI data were collected with a point-resolved spectroscopy technique (PRESS) with an echo time (TE) of 144 ms and repetition time (TR) of 1000 ms and 8x8x8 phase encoding. Thin section FLAIR images were used as scouts for the placement of the rectangular volume of interest for MRSI acquisition, which extended beyond the suspected disease (hyperintense areas on FLAIR images) to include normal appearing brain, allowing recording of control spectra. It was also positioned to avoid areas of subcutaneous lipid, bone and varying magnetic susceptibility that might have compromised the quality of the spectra. Typical MRSI acquisition had a nominal spatial resolution of 1x1x1 cm (1 cm3 voxel–1), although it ranged from 0.8x0.8x0.8 cm to 1.2x1.2x1.2 cm based on the size of the tumour.
The raw spectral data were reconstructed using GE's Functool software (General Electric Medical Systems). The peak parameters (height and area) for Cho and Cr were estimated on a voxel-by-voxel basis within the excited region. An automated statistical analysis served to identify a control population of spectra acquired from normal tissue within the image volume from the studied patient. The degree of spectral abnormality, on a voxel-by-voxel basis, was determined by evaluating the number of standard deviations of difference between the relative Cho and Cr levels within a given voxel and that of the control voxels. A grading system was developed to grade the Cho:Cr ratio (Table 1
). Although there is technically no upper limit to Cho:Cr, our preliminary data correlating selected biopsy and radiological progression with Cho:Cr calculated for the exact sample location have found a Cho:Cr of 3 or higher to be the closest value corresponding to the high grade tumour or the gross tumour volume (GTV) [12]. Again, our preliminary data have indicated that a Cho:Cr of 1–2 correlates best with microscopic disease or the clinical target volume (CTV) [12]. By interpolation, a Cho:Cr of 2–3 would correlate with non-enhancing anaplastic tumour. A Cho:Cr of less than 1 would be considered as normal brain tissue.
Text files containing the interpolated Cho:Cr ratios for each MRSI voxel overlaid on each FLAIR image were generated by the Functool software, as described previously [12]. A new set of MRSI images was then generated using the Cho:Cr grades to replace the Cho:Cr ratios of the MRSI voxels inside the MRS acquisition box on the FLAIR images, with intensities as per Table 1
. The screen-dumped MRSI (SDMRSI) images, in which the MRSI data are overlaid with each FLAIR image, were then generated (Figure 1). We then manually translated the SDMRSI set to fuse with the FLAIR images. The position of the MRS box on the new MRSI images was then compared with that on the SDMRSI image, and shifted until the agreement was within a predetermined tolerance. The fused MRI data set were then contoured jointly by the radiation oncologist and the radiologist using an interactive image analysis program developed at the Memorial Sloan-Kettering Cancer Center (MSKCC). Three CTVs were then contoured on the FLAIR images corresponding to the Cho:Cr grades.
View larger version (56K):
[in this window]
[in a new window]
|
Figure 1. Magnetic resonance spectroscopic imaging(MRSI) incorporation in treatment planning. (a) Fluid attenuated inversion recovery (FLAIR) image of a right frontal glioma. (b) Superimposition of multivoxel MRSI over the image. (c) Conversion of Cho:Cr ratio into a grayscale.
|
|
Functional MRI
The fMRI paradigm was designed to define the motor areas, language areas (including Broca's area and Wernicke's area) or both. Prior to fMRI scanning, these motor and language paradigms were practised outside the scanner environment until it was clear that subjects understood the task and were able to comply. The motor paradigm consisted of self-paced finger tapping. Four different language paradigms were then applied: (1) generation of action word for specific nouns; (2) word generation from categories; (3) generation of words beginning with a given alphabetic character; and (4) naming pictures. The paradigm was presented as a block design, consisting of 60 images, with 6 intervals (5 images) of paradigm execution alternated by 6 intervals (5 images) of rest. For these tasks, the patients were asked to perform the tasks silently, avoiding mouth and tongue movement. Stimuli were presented using a Brainwave software (Medical Numerics; Sterling, VA) and the stimulus was projected onto LCD (liquid crystal display) goggles worn by the patient. The stimulus presentation was synchronized with the scanner. Using the Brainwave software, the results of BOLD (blood oxygenation level dependent) activation were visualized in real time during scanning.
Functional images were acquired with a repetition time (TR) of 4000 ms; echo time (TE) of 40 ms; 90° flip angle; 128x128 matrix; 240 mm field of view (FOV); and 4.5 mm thickness. 21 axial slices were selected covering lesions. 2D and 3D T1 weighted anatomical images were acquired with a spin echo and a spoiled GRASS sequence, respectively.
Functional images were analysed and overlaid onto structure images with the Analysis of Functional Neuroimaging (AFNI) program. Statistical maps of activation were generated using a cross-correlation algorithm (Figure 2). A modelled box-car waveform corresponding to the stimulus presentation (task performance block) was cross-correlated with all pixel time courses on a pixel-by-pixel basis to identify stimulus locked responses. The standard deviation of the distribution of the resting-state correlation coefficients is typically somewhat less than 0.1. A threshold of 0.5, five times the standard deviation, which corresponds to a p<0.0001, after Bonferroni correlation, threshold was applied to all pixels to generate a colour-based activation map displaying all pixels surpassing this threshold. Images were then visually inspected for gross artefacts and viewed in a cine loop to detect residual motion. These regions were then superimposed on the CT data set after image registration of the functional set (an MRI FLAIR (first the functional areas will be overlaid in the T1 data set, then registration is to be done on the FLAIR images)) to that of the planning CT set as previously described.
View larger version (56K):
[in this window]
[in a new window]
|
Figure 2. Functional MRI(fMRI) incorporation in treatment planning. (a) Text file of motor cortical mapping in a patient with frontal glioblastoma. (b) three-dimensional reconstruction of motor cortical mapping in the same patient.
|
|
IMRT planning
All patients were immobilized using an Aquaplast (Aquaplast, Wycoff Heights, NJ) face mask. Patients then underwent CT simulation with intravenous contrast and images were obtained from the vertex to approximately the level of C2. The contoured FLAIR images with three CTVs based on the Cho:Cr grades were then fused with the simulation CT images using a mutual-information algorithm available on the CT Simulator (AcQSim; Philips Medical Systems, Best, The Netherlands). The planning target volume (PTV) was determined by adding an additional 0.5 cm margin to the CTV to account for treatment uncertainties.
The normal anatomical critical tissues included the spinal cord, brain stem, eyes, optic chiasm, optic nerves and the functional regions of the cortex as defined by the fMRI. The "normal brain" was then defined as all brain tissue outside the PTV. The dose limits that defined an acceptable plan included a maximum point dose of 4500 cGy to the spinal cord, 5400 cGy to the optic nerves and the chiasm, 4000 cGy to the retina and 5940 cGy to the brain stem. An attempt was made to keep the dose to the adjacent functional cortices as low as possible.
An IMRT plan that delivered a uniform dose to the PTV defined by the conventional imaging using inverse planning was generated using the MSKCC in-house developed treatment planning system and then used for the actual treatment. A dose of 54 Gy was prescribed to the low grade gliomas while high grade gliomas were treated with 59.4 Gy using 1.8 Gy per fraction prescribed to the 100% isodose line. Multiple non-coplanar beams were used. An attempt was made to limit the number of treatment fields to three to five in consideration of both treatment complexity and the volume of normal brain tissue irradiated to low dose levels. Acceptable target coverage was defined by a D95 (dose received by 95% of the PTV) of at least 95% of the prescription. Treatment was delivered using dynamic multileaf collimation (DMLC). All patients were treated using 6 MV X-rays from Varian accelerators equipped with a Millennium DMLC. Separately, an IMRT plan that could simultaneously deliver three different dose levels of 54 Gy/59.4 Gy/70 Gy (column 4 of Table 1) to the three MRSI defined CTVs using inverse planning was then generated using the same treatment planning system with similar dosimetric constraints and similar beam arrangement (Figure 3). Although this plan tested the feasibility of using dose painting, it was not used in the actual treatment of the patients. This IMRT dose painting plan using MRSI/fMRI defined volumes was then compared with the IMRT plan using MRI defined volumes. GTV (defined as Cho:Cr 2 with image definition and contrast enhancement with conventional imaging), CTV (defined as Cho:Cr >1 with image definition and 1 cm around FLAIR defined volume with conventional imaging) and PTV (defined as 0.5 cm around the CTV as described earlier) were compared.
View larger version (91K):
[in this window]
[in a new window]
|
Figure 3. Intensity-modulated radiation therapy (IMRT) planning in glioma. (a) Fluid attenuated inversion recovery (FLAIR) image of a patient with brain stem glioma. (b)Multivoxel MR spectroscopic imaging (MRSI). (c) Conversion into grayscale of Cho:Cr ratio to define the target volume. (d) Intensity-modulated radiation therapy (IMRT) plan with dose painting.
|
|
|
Results
|
|---|
The conversion of raw MRSI data into a grayscale and the MRI–MRSI image fusion protocol that was developed [12] were found to be consistently reliable and reproducible. We were able to successfully define the target volume in all the 12 cases based on the Cho:Cr ratio. The use of MRSI allowed the identification of high grade elements defined by a Cho:Cr ratio of >3 within the low grade gliomas in three of the six patients indicating the need for higher dose coverage that otherwise would have been missed with conventional imaging. Dose painting using IMRT could be done in all the patients based on the intensity of abnormality as defined by the Cho:Cr ratio on MRSI.
We noted that the location of MRSI based PTV defined by Cho:Cr 2 extended beyond MRI FLAIR defined volumes in six of the 12 patients (Figure 4). On the other hand, FLAIR sequences overestimated the PTV in six other patients compared with MRSI defined volumes indicating overtreatment of normal brain tissue (Figure 5). There was no correlation between MRI defined PTV and MRSI based Cho:Cr >2 volumes. However, MRSI defined GTV (Cho:Cr >3) was smaller by 40% compared with post-contrast T1 weighted imaging defined GTV volumes (Figure 6).
View larger version (110K):
[in this window]
[in a new window]
|
Figure 4. Comparison of target volumes with(a,b) MR spectroscopic imaging (MRSI) information and fluid attenuated inversion recovery (FLAIR) defined CTV, and (c,d) MRSI defined CTV showing both overestimation (upper arrow) and underestimation (lower arrow) with MRI defined volumes.
|
|
View larger version (31K):
[in this window]
[in a new window]
|
Figure 5. Comparison of treatment plans with MR spectroscopic imaging(MRSI) information. (a) MRI defined plan. (b) MRSI defined plan.
|
|
View larger version (17K):
[in this window]
[in a new window]
|
Figure 6. Comparison of planned target volumes(PTV) with MRI and MR spectroscopic imaging (MRSI) defined volumes.
|
|
We also found that the functional MRI information could be successfully fused to the MRI/CT images and could be used to define secondary critical regions of the brain in all cases. Use of functional imaging helped in designing the orientation of the beams while maintaining similar PTV coverage in some patients (Figure 7). However, due to proximity to target tissue with which it overlapped in some cases or in brain stem location, it did not help in further improvization of the treatment plan in others.
View larger version (40K):
[in this window]
[in a new window]
|
Figure 7. Use of functional MRI(fMRI) in treatment planning. (a) MR spectroscopic imaging (MRSI) defined plan. (b) MRSI with fMRI defined plan.
|
|
With a median follow up of 1 year, six of these patients have failed (high grade –4, low grade –2), all in the previously treated region. Five of the six failures are in the MRSI defined Cho:Cr >3 region within the tumour (Figure 8). The other patient recurred in a MRSI defined Cho:Cr >2 region in a tumour which mostly had Cho:Cr 1–2 region. Three patients underwent resection, all of which showed glioblastoma in the pathology specimen. However, since they were not stereotactic guided procedures, localization to previously noted Cho:Cr >3 region on imaging could not be done.
View larger version (37K):
[in this window]
[in a new window]
|
Figure 8. Pattern of failure in a patient with brain stem glioma.(a) FLAIR image of a low grade brain stem glioma. (b) Superimposition of multi-voxel MR spectroscopic imaging (MRSI) over the image showing areas of high (upper arrow) and low (lower arrow) Cho:Cr ratios. (c) Recurrence 4 months later in the area of high Cho:Cr ratio.
|
|
|
Discussion
|
|---|
Radiation therapy of gliomas is challenging. In spite of conventional radiation therapy to doses of 50–60 Gy to MR defined tumour with margins, 90% of the recurrences occur within the radiation volume [13]. One explanation could be that both target definition and grading of the tumour based on conventional MRI information on which radiation therapy is planned has its limitations. The gadolinium-enhancing lesion, as seen on T1 weighted MRI, reflects regions where there has been a breakdown of the blood–brain barrier. This may not be a reliable indicator of active tumour due to the presence of non-enhancing tumour tissue or contrast-enhancing necrosis [5]. Similarly, T2 defined volume either overestimates or underestimates the microscopic or non-enhancing disease in a majority of patients [14]. The other explanation could be that the dose of 50.4–54 Gy for low grade lesions and 59.4–60 Gy using 1.8–2 Gy per fraction may not be sufficient to control the tumour, due to the limitations of biopsy material or non-enhancing tumour on imaging techniques to detect the high grade tumour adequately. It is also possible that we may need doses in excess of 70 Gy for the high grade tumours for achieving local control.
We found that MRSI is a much better tool than conventional MRI alone in both defining the true extent of the tumour as well as defining the grade of the tumour. Conventional MRI failed to detect the tumour adequately in 50% of the patients that were noted in the MRSI images. Similarly we found that post-contrast T1 weighted with imaging overestimated the GTV defined by Cho:Cr ratio of >3 by 40%. We also noted that the T2 based imaging overestimated the CTV defined by Cho:Cr ratio of >1 by 30% in half the patients indicating overtreatment of normal brain tissue. We were able to validate the MRSI defined target definition by observing the patterns of relapse. In all the cases, patients failed in MRSI defined high Cho:Cr region which indicated the active high grade tumour. The tumour recurrence was biopsy confirmed to be glioblastoma in half the cases. It was interesting that two of these cases were so-called low grade gliomas prior to the re-resection, indicating the effectiveness of MRSI in predicting the grade of the tumour. The histopathological validation of MRSI as a predictor of tumour presence has been described previously by McKnight et al [15]. Importantly, in our opinion, it takes away the myth of separating gliomas into low grade and high grade "water tight" compartments and recognizes it as a continuous spectrum of disease progression as demonstrated by genetic analysis. We plan to further validate Cho:Cr ratio with histological grading with frameless stereotactic biopsies done at the time of surgery soon on a clinical protocol.
Although MRSI is a powerful tool, it has its limitations. Poor quality spectroscopy due to the presence of clips, radioactive seeds, clotted blood and skull bone within the measured volume may make it very difficult to interpret the findings [16]. Decreasing the size of the individual voxels improves the spatial resolution of MRSI data but can increase the scan time and/or decrease the signal-to-noise ratio (quality of the spectroscopy). Use of 3 T MRI and defining the volume of interest tightly around the FLAIR abnormality while avoiding the ventricles and bone can significantly improve the target definition. We have used Cho:Cr ratio as a marker for tumour definition and found it to be consistently reproducible and repeatable. Pirzkall et al used both Cho:Cr and Cho:NAA ratio, categorised based on abnormality index and demonstrated metabolically active tumour outside MR defined volumes in a non-uniform manner in both low grade and high grade tumours [9, 17]. Our findings validate their conclusion and indicate the need to custom design the treatment planning in gliomas based on the true extent of the disease and not just with an empirical margin around T1 and/or T2 anatomical based imaging.
An improvement in local control and survival has been clearly shown in high grade gliomas when the radiation dose has been escalated from 40 Gy to 60 Gy [18]. Despite a lack of consistent evidence for a strong dose–response curve beyond 60 Gy when using radiation to treat high grade gliomas, there is interest in employing higher doses, due to the benefit seen in other locations [19, 20]. However, dose escalation carries with it an increased incidence of side effects [21]. The two possible approaches would be to use IMRT and functional imaging to spare the critical structures. Our own preliminary data with IMRT in high grade gliomas indicated that it is unlikely that the use of IMRT with conventional doses and volume definition would improve the local control in high grade gliomas [22]. However, it did result in decreased acute and late toxicities associated with radiation therapy [22]. Similarly, functional imaging may selectively help in some cases where the tumour is in proximity to critical structures. It is imperative that every effort should be made to spare both the anatomical and the functional critical areas when planning dose escalation for gliomas.
|
Conclusion
|
|---|
We found that incorporating functional and spectroscopic imaging into the treatment planning process was feasible. Use of MRSI resulted in significant change in target location and volume compared with MRI defined volumes. It also resulted in identification of high grade elements within the low grade gliomas. Use of Cho:Cr ratio enabled dose painting using IMRT. Use of fMRI may help in designing the orientation of the beams to decrease the dose to critical areas. Based on these promising results, we are planning to open a Phase I dose escalation trial using MRSI and fMRI defined volumes with IMRT dose painting in gliomas.
Received for publication May 10, 2006.
Revision received September 15, 2006.
Accepted for publication September 15, 2006.
|
References
|
|---|
- Medical Research Council Brain Tumor Working Party. Randomized trial of procarbazine, lomustine, and vincristine in the adjuvant treatment of high grade astrocytoma: a Medical Research Council trial. J Clin Oncol 2001;19:509–18.[Abstract/Free Full Text]
- Shaw EG, Stieber V, Tatter S, Ellis T, Hinson W, Kearns W, et al. A Phase I dose escalating study of intensity modulated radiation therapy (IMRT) for the treatment of glioblastoma multiforme (GBM). Int J Radiat Oncol Biol Phys 2002;54: 206Abstract
- Wallner KE. Radiation therapy treatment planning for malignant astrocytomas. Semin Radiat Oncol 1991;1:17–25.[Medline]
- Thornton AF Jr, Sandler HM, Ten Haken RK, McShawn 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.[Medline]
- Byrne TN. Imaging of gliomas. Semin Oncol 1994;21:162–71.[Medline]
- Duijn JH, Matson GB, Maudsley AA, Weiner MW. 3D phase encoding 1H spectroscopic imaging of human brain. Magn Reson Imaging 1992;10:315–19.[CrossRef][Medline]
- Moonen CT, von Kienlin M, van Zijl PC, Cohen J, Gillen J, Daly P, et al. Comparison of single-shot localization methods (STEAM and PRESS) for in vivo proton NMR spectroscopy. NMR Biomed 1989;2:201–8.[Medline]
- De Stefano N, Caramanos Z, Preul MC, Francis G, Antel JP, Arnold DL. In vivo differentiation of astrocytic brain tumors and isolated demyelinating lesions of the type seen in multiple sclerosis using 1H magnetic resonance spectroscopic imaging. Ann Neurol 1998;44:273–8.[CrossRef][Medline]
- Pirzkall A, McKnight TR, Graves EE, Carol MP, Sneed PK, Wara WW, et al. MR-spectroscopy guided target delineation for high-grade gliomas. Int J Radiat Oncol Biol Phys 2001;50:915–28.[CrossRef][Medline]
- Moonen CTW, Bandettini PA, editors. Functional MRI. Berlin, Germany: Springer, 1999
- Liu WC, Schulder M, Narra V, Kalnin AJ, Kathcart C, Jacobs A, et al. Functional magnetic resonance imaging aided radiation treatment planning. Med Phys 2000;27:1563–72.[CrossRef][Medline]
- Chang J, Thakur S, Perera G, Kowalski A, Huang W, Karimi S, et al. Image-fusion of MR spectroscopic images for treatment planning of gliomas. Med Phys 2006;33:32–40.[CrossRef][Medline]
- Lee SW, Fraass BA, Marsh LH, Herbort K, Gebarski SS, Martel MK, et al. Patterns of failure following high-dose 3-D conformal radiotherapy for high-grade astrocytomas: a quantitative dosimetric study . Int J Radiat Oncol Biol Phys 1999;43:79–88.[Medline]
- Earnest F 4th, Kelly PJ, Scheithauer BW, Kall BA, Cascino TL, Ehman RL, et al. Cerebral astrocytomas: histopathologic correlation of MR and CT-contrast enhancement with stereotaxic biopsy. Radiology 1998;166:823–7.
- McKnight TR, von dem Bussche MH, Vigneron DB, Lu Y, Berger MS, McDermott MW, et al. Histopathological validation of a three-dimensional magnetic resonance spectroscopy index as a predictor of tumor presence. J Neurosurg 2002;97:794–802.[Medline]
- Narayana A, Leibel SA. Adult brain tumors. In: Leibel SA, Philips TL, editors. Textbook of radiation oncology, 2nd edn. Philadelphia, PA: Saunders, 2004:463–95
- Pirzkall A, Nelson SJ, McKnight TR, Takahashi MM, Li X, Grave EE, et al. Metabolic imaging of low-grade gliomas with three-dimensional magnetic resonance spectroscopy. Int J Radiat Oncol Biol Phys 2002;53:1254–64.[CrossRef][Medline]
- Bleehen NM, Stenning SP. A Medical Research Council trial of two radiotherapy doses in the treatment of grades 3 and 4 astrocytoma. The Medical Research Council Brain Tumor Working Party. Br J Cancer 1991;64:769–74.[Medline]
- Hunt MA, Zelefsky MZ, Wolden S, Chui CS, LoSasso T, Rosenzweig K, et al. Treatment planning and delivery of intensity-modulated radiation therapy for primary nasopharynx cancer. Int J Radiat Oncol Biol Phys 2001;49:623–32.[CrossRef][Medline]
- Huang E, Teh BS, Strother DR, Davis QG, Chiu JK, Lu HH, et al. Intensity-modulated radiation therapy for pediatric medulloblastoma: early report on the reduction of ototoxicity. Int J Radiat Oncol Biol Phys 2002;52:599–605.[CrossRef][Medline]
- Roman DD, Sperduto PW. Neuropsychological effects of cranial radiation: current knowledge and future directions. Int J Radiat Oncol Biol Phys 1995;31:983–8.[CrossRef][Medline]
- Narayana A, Yamada Y, Hunt M, Shah P, Gutin PH, Leibel SA. Intensity modulated radiotherapy in high grade gliomas: clinical and dosimetric results. Int J Radiat Oncol Biol Phys 2006;64:892–7.[CrossRef][Medline]
This article has been cited by other articles:
|
|
|
|
 
L.-G. Vezina
Imaging of Central Nervous System Tumors in Children: Advances and Limitations
J Child Neurol,
October 1, 2008;
23(10):
1128 - 1135.
[Abstract]
[PDF]
|
|
|
|
|
|
|
C. Plathow and W. A. Weber
Tumor Cell Metabolism Imaging
J. Nucl. Med.,
June 1, 2008;
49(Suppl_2):
43S - 63S.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
BJR review of the year -- 2007
Br. J. Radiol.,
April 1, 2008;
81(964):
265 - 269.
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
