Navigation-Guided Surgery of Suspected Low-Grade Gliomas
Navigation-Guided Surgery of Suspected Low-Grade Gliomas
In the present study, altogether 51 gliomas with nonsignificant contrast enhancement on MRI were included to define a standardized multimodality imaging protocol (n = 40) that was subsequently prospectively validated during advanced navigation-guided glioma surgery (n = 11).
Initially, we integrated the multiple imaging data obtained from the first 40 glioma patients into our planning station and retrospectively reviewed these data to define standardized settings of structural, metabolic, and fiber tract data and 3D brain visualization. We aimed at enabling an optimal concurrent visualization of all relevant multimodality image data in a single picture. Details of the standardized imaging settings for each modality are provided in Table 3 .
Structural Imaging. T1-Weighted Images With Contrast Medium, T2-Weighted, and FLAIR Sequences The planning station recognizes structural MR images and automatically selects the "grayscale" color map resulting in predefined level and width settings. Additional adjustments can be made depending on the intended visualization of the tumor and surrounding white and gray matter.
Contour Visualization. The planning station is capable of outlining only the contour of specific pathological or functional objects. For contour visualization of tumor borders, we propose selecting T2-weighted or FLAIR images to create a specific "object" by using the 3D model tool. Subsequently, the contour of this object is then superimposed on the reference study (T1-weighted contrast-enhanced MRI), enabling the concurrent visualization of the exact tumor border and all additional coregistered image data.
Metabolic Imaging. PET. As the PET scan is not automatically recognized by the planning station, the first step is to define it as "PET" in the image modality, which leads to the "rainbow" mode in the color map and thus enables color-coded visualization of the PET data (Fig. 1). To highlight only the intratumoral area with the highest metabolic activity corresponding to the PET hotspot, we propose to increase the level value to 25,000 ± 1000 and adjust the width value to 6000 ± 1000. Furthermore, for a simultaneous visualization of the PET hotspot with the structural images in the background, it was found to be optimal to decrease the transparency of the PET scan from 100% to 20%.
(Enlarge Image)
Figure 1.
Illustration of the standardized image settings in metabolic images using PET and MRS CSI. PET (upper) and CSI images (lower) on the planning station. A and B: The raw data of the PET scan and the specific CSI hybrid data set both in black and white (A) are converted into color-coded images (B). C: By specific adjustment of the level and width value, only the intratumoral PET and CSI hotspot can be visualized. D: Decreasing the transparency to 20% allows a concurrent visualization of structural images using contrast-enhanced T1-weighted images and the PET and CSI hotspot for navigation-guided tissue sampling.
CSI. This time, "SPECT" has to be selected within the image modality to allow a "rainbow" color-coded visualization of the CSI data (Fig. 1). The optimal adjustments for CSI data were defined as follows: the level value has to be increased to maximum and the width value should be adjusted to 150 ± 100 and the transparency to 20%.
Fiber Tracking. After image fusion, the specific objects created on the planning station were automatically extracted out of the fiber tract data by using the 3D model and import models tool. The tracts were visualized as 3D fiber bundles within the 3D brain surface model or as 2D fiber contours (contour visualization) superimposed on the reference image.
3D Brain Visualization. An anatomical 3D model of the brain can be created using the 3D model tool of the planning station. After selection of the T1-weighted images without contrast medium the lower threshold has to be increased by stages until only the brain tissue is visible and the surrounding bone, fat, and skin disappears (lower threshold 50–300 and upper threshold 1200). To obtain a detailed 3D brain model with detailed gyral and sulcal structures, we prefer the "basic" mode in the color map within the 3D model display. In contrast, for 3D brain vessel visualization we apply T1-weighted images with contrast medium that enable 3D brain models with visualization of superficial vascular structures. For an optimal visualization of the 3D vascular pattern either the "basic" or "shaded" mode in the color map can be selected. A concurrent visualization of both 3D brain models is possible.
Second, the expected benefit of the various image modalities was retrospectively analyzed in each of the 40 cases by 2 experienced neurosurgeons (S.W. and G.W.) to define specific indications for the different imaging modalities. For each imaging modality, we applied the standardized imaging settings described above. A detailed overview of these indications is provided in Table 4, and the multimodality image data are illustrated in Fig. 2. 1) Impact of T2-weighted or FLAIR images for delineation of the tumor borders. The application of T2-weighted or FLAIR images in the context of the contour visualization function was considered useful in all cases (n = 36; 90% of cases) that are treated by resection for precise definition of the tumor margin. 2) Impact of metabolic imaging (PET or CSI) for identification of an intratumoral hotspot. The application of metabolic imaging using PET or CSI was considered useful in all cases; PET was capable of identifying an intratumoral hotspot (T/N ratio ≥ 1.2) in 33 patients. CSI was able to detect a CSI hotspot in all cases of a negative PET (T/N ratio < 1.2; n = 7). 3) Impact of fiber tracking for visualization of relevant white matter tracts adjacent to gliomas. Forty percent of the gliomas (n = 16) were localized close to eloquent cortical or subcortical structures (motor or speech area). Consequently, fiber tracking was considered useful in these patients, especially to limit glioma resection and thus preserve important functional structures. 4) Impact of 3D brain visualization. 3D brain surface visualization was considered useful for precise localization of all gliomas (n = 20 patients; 50% of cases) situated at the brain convexity. In contrast, only minor benefit of 3D brain surface visualization was expected in cases of deep-seated tumors and repeated surgeries.
(Enlarge Image)
Figure 2.
Illustration of the multimodality imaging data on the planning station that were obtained in a patient with a suspected LGG. A: Structural images consisting of contrast-enhanced T1-weighted images (reference image for navigation, upper) and FLAIR sequence (lower) reveal a lesion in the left frontal premotor region. B: The tumor volume was segmented using FLAIR images and is superimposed as red contour on the reference image. C: Furthermore, the color-coded PET "hotspot" is added. D: In the next step, fiber tracking data consisting of the corticospinal tract (green contour) and the arcuate fascicle (yellow contour) are included. E: Finally, the 3D brain surface model (upper) based on T1-weighted images without contrast medium administration clearly shows the exact topographic localization of the tumor (red) in the middle frontal gyrus. The 3D brain vessel model (lower) based on additional contrast-enhanced T1-weighted images enables visualization of the vascular brain surface anatomy important for approach planning and intraoperative orientation.
The 3D brain vessel visualization was deemed beneficial for approach planning and intraoperative orientation in 30 patients (75% of cases). To this end, especially the identification of bridging veins, vein of Labbé, and specific cortical vessels that serve as anatomical landmarks was considered useful.
The standardized imaging settings together with the specific indications for each imaging modality defined our new multimodality imaging protocol for gliomas with nonsignificant contrast enhancement (Table 3 and Table 4). This newly established protocol including structural images, metabolic data, fiber tracking, and/or 3D surface visualization was then prospectively applied during navigationguided surgery in 11 glioma patients (Table 2). Intraoperative monitoring with either cortical/subcortical stimulation (n = 4) or awake surgery (n = 3) was additionally applied to preserve neurological function if indicated. According to the appraisal of the neurosurgeons performing the surgery, this protocol was feasible and was rated surgically relevant in all cases. By using this protocol all surgically important imaging modalities were concurrently visualized in each case. In all gliomas that were considered to be resectable according to preoperative imaging (n = 5), a complete tumor resection was achieved by using contour visualization with T2-weighted or FLAIR images. Moreover, all tumor samples from the metabolic hotspot (PET or CSI) of each WHO Grade III or IV glioma (n = 5) revealed malignant tissue according to the histopathological WHO criteria. Additionally, the use of fiber tracking was a powerful tool to intraoperatively define brain areas adjacent to the glioma in which relevant white matter tracts are expected. The application of intraoperative monitoring such as cortical/subcortical stimulation or awake surgery provided the neurosurgeon with additional "real-time" information that is unaffected by brain shift regarding the integrity of the language/motor fiber tracts during tumor removal. A permanent postoperative neurological deficit was not observed in any of these 11 patients (unchanged neurological findings in 7 cases; transient neurological deficit in 4 cases), whereby DTI (10 cases) and/or intraoperative monitoring (7 cases) were applied during surgery in the vast majority of patients. Finally, the 3D brain surface and vessel models showed a significant intraoperative topographical correlation with gyral anatomy and superficial vessels in all cases in which this imaging technique was considered to be useful. Further details are provided in Table 2.
The additional application of electromagnetic instrument tracking enabled continuous navigation guidance during glioma surgery. For this purpose, a flexible "stylet" integrated in the suction device allowed "real-time" navigation using the different imaging modalities. This technique was especially helpful for precise guidance to surgical targets such as the intratumoral metabolic "hotspot" or the tumor margin. Two representative cases (Cases 4 and 11) of the 11 glioma patients are illustrated in Figs. 3 and 4.
(Enlarge Image)
Figure 3.
Case 4. Intraoperative navigation using the multimodality imaging protocol in a patient with a suspected LGG. A and B: During glioma resection, navigation with concurrent display of contour visualization of the tumor border (red contour), the metabolic hotspot using PET (color-coded region), and fiber tracking data with the corticospinal tract (green contour) was performed. C and D: Intraoperative navigation with continuous instrument tracking shows an excellent correlation of the brain surface and vessel anatomy (C) and the corresponding 3D brain model (D) that supports precise lesion localization and intraoperative orientation.
(Enlarge Image)
Figure 4.
Case 11. Intraoperative application of the multimodality imaging protocol in a patient with a glioma with nonsignificant contrast enhancement. A and B: Preoperative MR images revealing a left frontal hyperintense lesion on T2-weighted images (A) with patchy and faint contrast enhancement on contrast-enhanced T1-weighted sequences (B). C: During the glioma resection, navigation with multimodality imaging data including contour visualization of the tumor border (red contour), the metabolic PET hotspot (color-coded region), and the corticospinal tract (green contour) was conducted. D and E: Additionally, electromagnetic navigation with a 3D brain surface and vessel visualization model was performed (D) that showed a significant correlation with the brain surface and vessel anatomy (E). F–I: Histopathological examination of the tumor sample derived from the PET hotspot depicts WHO Grade IV glioma tissue (F) with a high proliferation index (G), whereas the tumor specimen from outside the metabolic hotspot shows only WHO Grade II glioma tissue with a low proliferation rate (I). H & E (F and H) and anti–Ki 67 (G and I), original magnification × 200. J: Postoperative T2-weighted MR image demonstrating complete resection of the lesion.
Results
In the present study, altogether 51 gliomas with nonsignificant contrast enhancement on MRI were included to define a standardized multimodality imaging protocol (n = 40) that was subsequently prospectively validated during advanced navigation-guided glioma surgery (n = 11).
Definition of Standardized Image Settings
Initially, we integrated the multiple imaging data obtained from the first 40 glioma patients into our planning station and retrospectively reviewed these data to define standardized settings of structural, metabolic, and fiber tract data and 3D brain visualization. We aimed at enabling an optimal concurrent visualization of all relevant multimodality image data in a single picture. Details of the standardized imaging settings for each modality are provided in Table 3 .
Structural Imaging. T1-Weighted Images With Contrast Medium, T2-Weighted, and FLAIR Sequences The planning station recognizes structural MR images and automatically selects the "grayscale" color map resulting in predefined level and width settings. Additional adjustments can be made depending on the intended visualization of the tumor and surrounding white and gray matter.
Contour Visualization. The planning station is capable of outlining only the contour of specific pathological or functional objects. For contour visualization of tumor borders, we propose selecting T2-weighted or FLAIR images to create a specific "object" by using the 3D model tool. Subsequently, the contour of this object is then superimposed on the reference study (T1-weighted contrast-enhanced MRI), enabling the concurrent visualization of the exact tumor border and all additional coregistered image data.
Metabolic Imaging. PET. As the PET scan is not automatically recognized by the planning station, the first step is to define it as "PET" in the image modality, which leads to the "rainbow" mode in the color map and thus enables color-coded visualization of the PET data (Fig. 1). To highlight only the intratumoral area with the highest metabolic activity corresponding to the PET hotspot, we propose to increase the level value to 25,000 ± 1000 and adjust the width value to 6000 ± 1000. Furthermore, for a simultaneous visualization of the PET hotspot with the structural images in the background, it was found to be optimal to decrease the transparency of the PET scan from 100% to 20%.
(Enlarge Image)
Figure 1.
Illustration of the standardized image settings in metabolic images using PET and MRS CSI. PET (upper) and CSI images (lower) on the planning station. A and B: The raw data of the PET scan and the specific CSI hybrid data set both in black and white (A) are converted into color-coded images (B). C: By specific adjustment of the level and width value, only the intratumoral PET and CSI hotspot can be visualized. D: Decreasing the transparency to 20% allows a concurrent visualization of structural images using contrast-enhanced T1-weighted images and the PET and CSI hotspot for navigation-guided tissue sampling.
CSI. This time, "SPECT" has to be selected within the image modality to allow a "rainbow" color-coded visualization of the CSI data (Fig. 1). The optimal adjustments for CSI data were defined as follows: the level value has to be increased to maximum and the width value should be adjusted to 150 ± 100 and the transparency to 20%.
Fiber Tracking. After image fusion, the specific objects created on the planning station were automatically extracted out of the fiber tract data by using the 3D model and import models tool. The tracts were visualized as 3D fiber bundles within the 3D brain surface model or as 2D fiber contours (contour visualization) superimposed on the reference image.
3D Brain Visualization. An anatomical 3D model of the brain can be created using the 3D model tool of the planning station. After selection of the T1-weighted images without contrast medium the lower threshold has to be increased by stages until only the brain tissue is visible and the surrounding bone, fat, and skin disappears (lower threshold 50–300 and upper threshold 1200). To obtain a detailed 3D brain model with detailed gyral and sulcal structures, we prefer the "basic" mode in the color map within the 3D model display. In contrast, for 3D brain vessel visualization we apply T1-weighted images with contrast medium that enable 3D brain models with visualization of superficial vascular structures. For an optimal visualization of the 3D vascular pattern either the "basic" or "shaded" mode in the color map can be selected. A concurrent visualization of both 3D brain models is possible.
Analysis of Impact of the Different Imaging Modalities
Second, the expected benefit of the various image modalities was retrospectively analyzed in each of the 40 cases by 2 experienced neurosurgeons (S.W. and G.W.) to define specific indications for the different imaging modalities. For each imaging modality, we applied the standardized imaging settings described above. A detailed overview of these indications is provided in Table 4, and the multimodality image data are illustrated in Fig. 2. 1) Impact of T2-weighted or FLAIR images for delineation of the tumor borders. The application of T2-weighted or FLAIR images in the context of the contour visualization function was considered useful in all cases (n = 36; 90% of cases) that are treated by resection for precise definition of the tumor margin. 2) Impact of metabolic imaging (PET or CSI) for identification of an intratumoral hotspot. The application of metabolic imaging using PET or CSI was considered useful in all cases; PET was capable of identifying an intratumoral hotspot (T/N ratio ≥ 1.2) in 33 patients. CSI was able to detect a CSI hotspot in all cases of a negative PET (T/N ratio < 1.2; n = 7). 3) Impact of fiber tracking for visualization of relevant white matter tracts adjacent to gliomas. Forty percent of the gliomas (n = 16) were localized close to eloquent cortical or subcortical structures (motor or speech area). Consequently, fiber tracking was considered useful in these patients, especially to limit glioma resection and thus preserve important functional structures. 4) Impact of 3D brain visualization. 3D brain surface visualization was considered useful for precise localization of all gliomas (n = 20 patients; 50% of cases) situated at the brain convexity. In contrast, only minor benefit of 3D brain surface visualization was expected in cases of deep-seated tumors and repeated surgeries.
(Enlarge Image)
Figure 2.
Illustration of the multimodality imaging data on the planning station that were obtained in a patient with a suspected LGG. A: Structural images consisting of contrast-enhanced T1-weighted images (reference image for navigation, upper) and FLAIR sequence (lower) reveal a lesion in the left frontal premotor region. B: The tumor volume was segmented using FLAIR images and is superimposed as red contour on the reference image. C: Furthermore, the color-coded PET "hotspot" is added. D: In the next step, fiber tracking data consisting of the corticospinal tract (green contour) and the arcuate fascicle (yellow contour) are included. E: Finally, the 3D brain surface model (upper) based on T1-weighted images without contrast medium administration clearly shows the exact topographic localization of the tumor (red) in the middle frontal gyrus. The 3D brain vessel model (lower) based on additional contrast-enhanced T1-weighted images enables visualization of the vascular brain surface anatomy important for approach planning and intraoperative orientation.
The 3D brain vessel visualization was deemed beneficial for approach planning and intraoperative orientation in 30 patients (75% of cases). To this end, especially the identification of bridging veins, vein of Labbé, and specific cortical vessels that serve as anatomical landmarks was considered useful.
Intraoperative Application of the Standardized Multimodality Imaging Protocol
The standardized imaging settings together with the specific indications for each imaging modality defined our new multimodality imaging protocol for gliomas with nonsignificant contrast enhancement (Table 3 and Table 4). This newly established protocol including structural images, metabolic data, fiber tracking, and/or 3D surface visualization was then prospectively applied during navigationguided surgery in 11 glioma patients (Table 2). Intraoperative monitoring with either cortical/subcortical stimulation (n = 4) or awake surgery (n = 3) was additionally applied to preserve neurological function if indicated. According to the appraisal of the neurosurgeons performing the surgery, this protocol was feasible and was rated surgically relevant in all cases. By using this protocol all surgically important imaging modalities were concurrently visualized in each case. In all gliomas that were considered to be resectable according to preoperative imaging (n = 5), a complete tumor resection was achieved by using contour visualization with T2-weighted or FLAIR images. Moreover, all tumor samples from the metabolic hotspot (PET or CSI) of each WHO Grade III or IV glioma (n = 5) revealed malignant tissue according to the histopathological WHO criteria. Additionally, the use of fiber tracking was a powerful tool to intraoperatively define brain areas adjacent to the glioma in which relevant white matter tracts are expected. The application of intraoperative monitoring such as cortical/subcortical stimulation or awake surgery provided the neurosurgeon with additional "real-time" information that is unaffected by brain shift regarding the integrity of the language/motor fiber tracts during tumor removal. A permanent postoperative neurological deficit was not observed in any of these 11 patients (unchanged neurological findings in 7 cases; transient neurological deficit in 4 cases), whereby DTI (10 cases) and/or intraoperative monitoring (7 cases) were applied during surgery in the vast majority of patients. Finally, the 3D brain surface and vessel models showed a significant intraoperative topographical correlation with gyral anatomy and superficial vessels in all cases in which this imaging technique was considered to be useful. Further details are provided in Table 2.
The additional application of electromagnetic instrument tracking enabled continuous navigation guidance during glioma surgery. For this purpose, a flexible "stylet" integrated in the suction device allowed "real-time" navigation using the different imaging modalities. This technique was especially helpful for precise guidance to surgical targets such as the intratumoral metabolic "hotspot" or the tumor margin. Two representative cases (Cases 4 and 11) of the 11 glioma patients are illustrated in Figs. 3 and 4.
(Enlarge Image)
Figure 3.
Case 4. Intraoperative navigation using the multimodality imaging protocol in a patient with a suspected LGG. A and B: During glioma resection, navigation with concurrent display of contour visualization of the tumor border (red contour), the metabolic hotspot using PET (color-coded region), and fiber tracking data with the corticospinal tract (green contour) was performed. C and D: Intraoperative navigation with continuous instrument tracking shows an excellent correlation of the brain surface and vessel anatomy (C) and the corresponding 3D brain model (D) that supports precise lesion localization and intraoperative orientation.
(Enlarge Image)
Figure 4.
Case 11. Intraoperative application of the multimodality imaging protocol in a patient with a glioma with nonsignificant contrast enhancement. A and B: Preoperative MR images revealing a left frontal hyperintense lesion on T2-weighted images (A) with patchy and faint contrast enhancement on contrast-enhanced T1-weighted sequences (B). C: During the glioma resection, navigation with multimodality imaging data including contour visualization of the tumor border (red contour), the metabolic PET hotspot (color-coded region), and the corticospinal tract (green contour) was conducted. D and E: Additionally, electromagnetic navigation with a 3D brain surface and vessel visualization model was performed (D) that showed a significant correlation with the brain surface and vessel anatomy (E). F–I: Histopathological examination of the tumor sample derived from the PET hotspot depicts WHO Grade IV glioma tissue (F) with a high proliferation index (G), whereas the tumor specimen from outside the metabolic hotspot shows only WHO Grade II glioma tissue with a low proliferation rate (I). H & E (F and H) and anti–Ki 67 (G and I), original magnification × 200. J: Postoperative T2-weighted MR image demonstrating complete resection of the lesion.
Source...