| Article Access Statistics | | | Viewed | 5758 | | | Printed | 164 | | | Emailed | 0 | | | PDF Downloaded | 97 | | | Comments | [Add] | | | Cited by others | 1 | | |
|
Click on image for details.
|
|
 |
| NI FEATURE: THE QUEST - COMMENTARY |
|
|
|
| Year : 2017 | Volume
: 65
| Issue : 2 | Page : 333-340 |
Three-dimensional visualization of intracranial tumors with cortical surface and vasculature from routine MR sequences
Zafar Neyaz1, Rajendra V Phadke1, Vivek Singh1, Chaitanya Godbole2
1 Department of Radiodiagnosis, Sanjay Gandhi Post Graduate Institute of Medical Sciences, Lucknow, Uttar Pradesh, India
2 Department of Neurosurgery, Sanjay Gandhi Post Graduate Institute of Medical Sciences, Lucknow, Uttar Pradesh, India
| Date of Web Publication | 10-Mar-2017 |
Correspondence Address:
Dr. Zafar Neyaz
Department of Radiodiagnosis, Sanjay Gandhi Post Graduate Institute of Medical Sciences, Rae Bareilly Road, Lucknow - 226 014, Uttar Pradesh
India
 Source of Support: None, Conflict of Interest: None  | Check |
DOI: 10.4103/neuroindia.NI_1167_16
The simultaneous three-dimensional (3D) visualization of intracranial tumors, brain structures, skull, and vessels is desired by neurosurgeons to create a clear mental picture of the anatomical orientation of the surgical field prior to the surgical intervention. Different anatomical and pathological components are usually visualized separately on different magnetic resonance (MR) sequences; however, during surgery, they are tackled simultaneously. Another problem is that most present day MR workstations enable review of two-dimensional (2D) slices only with limited postprocessing options. With recent software developments, a simultaneous 3D visualization simulating the real surgical field is possible using commercial or open source softwares. The authors have reviewed the important concepts and described a technique of interactive 3D visualization from routine 3D T1-weighted, MR angiography, and MR venography sequences using open source FSL (Functional MRI of the brain software library) and BrainSuite softwares.
Keywords: 3D image, brain surface, image processing, intracranial tumor, magnetic resonance imaging (MRI), neurosurgical planning
Key Messages:
Simultaneous interactive 3D visualization of intracranial tumors and their relationship with the brain surface and the vessels in their vicinity is highly desired by surgeons. It helps in better planning of the surgical approach and anticipation of complications.
How to cite this article:
Neyaz Z, Phadke RV, Singh V, Godbole C. Three-dimensional visualization of intracranial tumors with cortical surface and vasculature from routine MR sequences. Neurol India 2017;65:333-40 |
How to cite this URL:
Neyaz Z, Phadke RV, Singh V, Godbole C. Three-dimensional visualization of intracranial tumors with cortical surface and vasculature from routine MR sequences. Neurol India [serial online] 2017 [cited 2023 Jun 9];65:333-40. Available from: https://www.neurologyindia.com/text.asp?2017/65/2/333/201822 |
Magnetic resonance imaging (MRI) plays a major role in neurosurgical planning. Although two-dimensional (2D) MRI images contain all the required information for surgical planning, such as tumor morphology, as well as the relationship of the tumor to the brain parenchyma, and the vasculature in its close vicinity, neurosurgeons wish to have a good three dimensional orientation of the tumor and adjacent vital structures prior to surgery. Most MR workstations enable a review of 2D slices with limited postprocessing options such as multiplanar reconstruction (MPR), maximum intensity projection (MIP), and volume rendering (VR). Although the MIP and VR images allow the three-dimensional (3D) visualization of MR angiography (MRA) or MR venography (MRV) data by highlighting the vascular structures, the details of the tumor and brain surface are suppressed [Figure 1]. Another limitation of these techniques is that only one MR sequence can be visualized at a time. Different tissue types and tumor-related changes are usually visualized separately on different specially designed MR sequences; however, during surgery they are tackled simultaneously. Hence, a simultaneous interactive visualization of the various pathological and anatomical components (tumor, brain surface, skull, arteries, and veins) is desired by surgeons to create a clear 3D mental picture.
Limited work has been done on 3D visualization based on MRI images for neurosurgical planning.[1],[2],[3],[4],[5],[6],[7],[8] Kikinis et al., used a computer-based image processing software (Sun Microsystems) for preoperative surgical planning in 1996.[1] They processed the MRI images in 14 cases of intra- and extra-axial tumors and a vascular malformation. The 3D reconstructions offered significant advantages over cross-sectional tomographic images, especially in defining the relationship of the tumor to the eloquent cortex, gray matter nuclei, white matter tract, and blood vessels. Later, the same software was used for 3D reconstruction of blood vessels in the presence of cerebrovascular diseases.[2] Mert et al., performed a 3D brain surface visualization with and without incorporating the vessels using a commercially available neurosurgical planning software (StealthViz, Medtronic).[3] They found that the 3D brain visualization was clinically reliable and improved intraoperative orientation. Harput et al., have used OsiriX software for the 3D reconstruction of neocortical supratentorial lesions and compared it with true surgical views.[5] They concluded that reconstruction of the cortical surface was easy and helped the surgeons in creating a clear mental picture of the tumor location. Intraaxial tumors located under the cortical surface may be difficult to identify intraoperatively, and identification of sulcal landmarks may help in their localization. | Figure 1: MR venography (MRV) and MR angiography (MRA) images along with multi-planar reconstruction (MPR) and maximum intensity projection (MIP) reconstruction of a patient with a meningioma. MIP images provide clear delineation of the vascular structures; however, details of the tumor and brain surface are lacking
Click here to view |
| » Concepts | |  |
The important concepts and innovations which have made 3D visualization of MRI images a reality are discussed in brief. The four important computerized methods used are registration, segmentation, surface generation, and visualization.[9],[10],[11] Image registration implies alignment of two or more images taken at different times or from different viewpoints.[9],[10] Even studies done on different modalities on separate occasions (i.e., MRI, ultrasound, CT scan) can be spatially aligned using registration methods. Registration helps in motion correction between acquisitions of two MR sequences in a single patient [Figure 2]. | Figure 2: (a) Inaccurate superimposition of cerebral veins with the cortical surface. (b) Proper superimposition of cerebral veins over the cortical surface due to image registration
Click here to view |
Image segmentation results in the classification of MRI images into different tissue types [i.e., gray matter, white matter, cerebrospinal fluid (CSF)] or delineation of specific anatomical structures by drawing region boundaries [Figure 3]. Image segmentation is essential for the 3D visualization of a specific tissue or structure without overlap of an undesired tissue such as scalp. Manual segmentation can be done by a human operator to label a specific brain region or tissue. However, manual processing is time consuming and labor intensive. Thus, automatic or semiautomatic computer-aided image processing is necessary. A high contrast difference between the different tissue types on MR imaging facilitates segmentation, including edge detection.[11] The results of segmentation may be saved as a new image file containing voxels of a specific structure/tissue or as a binary mask. A binary mask file stores information by assigning a value of 1 for the voxels of interest and 0 for rest of the voxels. The masks can be applied to extract or remove the particular voxels from a MR image later [Figure 4]. Reconstruction of a 3D shape can be done by generating a computational surface mesh around the boundary of a defined tissue or structure.[12] These meshes are processed into a 3D surface for visualization [Figure 5]. The binary masks are very useful for generating surfaces. | Figure 3: (a) Finding a boundary between brain and skull using edge detection. (b) Classification of different types of brain tissues based on signal intensity. (c) Segmentation and labelling of different brain structures. (d) Detection of inner boundary of the cerebral cortex. (e) Image volume containing gray matter voxels. (f) Image volume containing white matter voxels
Click here to view |
 | Figure 4: Binary mask of the whole brain. This mask can be used to extract brain by removing the skull and scalp. The mask could be inverted to remove the brain as welloperating
Click here to view |
 | Figure 5: Generation of surface mesh around the mask. (a) Green outline mask is seen around cerebral arteries. (b) Generation of arterial surface from the mask. (c) 3D surface of the circle of Willis. (d and e) Magnified images showing the surface mesh
Click here to view |
| » Technique | | |
At present, only few commercial softwares are available for 3D visualization and neurosurgical planning and some are very expensive. However, there are many open source, freely available MRI processing softwares, which can be used. One such frequently used open source software is Slicer 3D. Images shown in this article were processed on FSL and BrainSuite software packages for the generation and visualization of surfaces of the tumor, brain, vessels, scalp, and skull [Figure 6]. As the BrainSuite software is designed to generate cortical surface only, a combination of automated and manual processing was used to accomplish the task. | Figure 6: Generation of 3D surfaces of tumor, cerebral cortex, brainstem, vessels, scalp, and skull from 3D T1-weighted, MRA, MRV, and postcontrast 3D T1-weighted sequences
Click here to view |
All cases presented in this article were performed on a GE Signa HDxt 3 tesla MR scanner (GE Healthcare, Milwaukee, WI) using head-neck spine (HNS) coil. DICOM images of the following four MR sequences were used for processing. The precontrast 3D T1-weighted gradient sequence (BRAVO) was acquired using the following parameters: sagittal plane, slice thickness 1.3 mm, interslice gap 0 mm, in-plane resolution 288 × 256, TI 400 ms, bandwidth 31.25, field of view (FOV) 24 cm, flip angle 10°, approximately 150 slice. The plane of sequence was aligned with the patient's head to obtain images in the true sagittal plane. Noncontrast 3D time of flight (TOF) MR angiography sequence was acquired with the following imaging protocol: axial plane, TR 18.0 ms, flip angle 15°, bandwidth 31.25, slice thickness 1.2 mm, overlap locs 13, locs per slab 48, in-plane resolution 320 × 224, FOV 24 cm, phase FOV 0.75. Four slabs were taken to cover the entire skull. For contrast-enhanced 3D TOF MRV acquisition, a gadolinium-based contrast (omniscan, GE healthcare, 0.1 mmoL/kg of body weight) was manually injected into an upper limb vein. MRV was performed using the following parameters: sagittal plane, TE 1.5 ms, flip angle 35°, bandwidth 62.5, FOV 27 cm, phase FOV 0.8, slice thickness 1.3 mm, locs 140, in-plane resolution 320 × 320, NEX 1. A postcontrast 3D T1-weighted gradient sequence (BRAVO) was repeated with the same parameters.
Two laptop computers with the following softwares were used for data processing [Figure 7]. (a) Dell Inspiron N5010 (Intel® core (TM) i3- M380 CPU@ 2.53 GHz, 4.00 GB RAM, 32/64-bit operating sytem) with Ubuntu 16.04.1 LTS Xenial Xerus operating system having dcm2nii converter (https://www.nitrc.org/projects/dcm2nii/) and fsl-5.0-complete package (https://www.fmrib.ox.ac.uk/fsl,">http://neuro.debian.net/pkgs/fsl-5.0-complete.html">https://www.fmrib.ox.ac.uk/fsl, http://neuro.debian.net/pkgs/fsl-5.0-complete.html). (b) Dell Inspiron 11, 3000 series (Intel® core (TM) i3-4030U CPU@ 1.90 GHz, 4.00 GB RAM, 64-bit operating sytem, x64-based processor) having Windows 10 operating system with BrainSuite 15b version installed (www.brainsuite.org).
FSL-FLIRT linear registration tool was used for registration of MRA, MRV, and postcontrast 3D T1-weighted images with the precontrast 3D T1-weighted images.[9] Registration was performed using Rigid Body (6 parameter model). Apart from showing the arteries, TOF-MRA sequence also nicely delineates the inner and outer table of skull bones. This unique feature makes it an ideal sequence for generating the inner and outer skull surface masks. Four TOF-MRA slabs were purposely acquired to produce a complete skull mask. These masks easily remove the scalp including the skull from MRA and MRV images, revealing the intracranial vasculature without any superimposition. FSL-BET brain extraction tool was used to produce the masks.[13],[14] The fractional intensity threshold was maintained at 0.1–0.3 (default 0.5), and the BET tool was run with “Run bet2 and then betsurf to get additional skull and scalp surfaces” option. The co-registered NIfTI files and masks generated by FSL-BET were copied into the Dell Inspiron 11 PC for further processing with the BrainSuite software.
For arterial surface, the co-registered MRA NIfTI file was opened into the BrainSuite, and the inner skull mask was applied to remove scalp tissue and skull [Figure 8]. A value lower than vessels and higher than brain parenchyma was used as the threshold to outline arteries. A similar method was used to generate the venous surface [Figure 9]. To create a skull surface, the MRA file was used [Figure 10]. Threshold settings were used to outline skull bones. A portion of the skull mask was removed (virtual craniotomy) to simulate the intraoperative appearance. To generate a tumor surface, usually the postcontrast T1-weighted file was used [Figure 11]. However, nonenhancing tumors were marked on precontrast T1-weighted images. The precontrast 3D T1-weighted data was loaded in the BrainSuite software, and cortical surfaces were generated using Cortical Surface Extraction Sequence [Figure 12].[15],[16],[17] For brainstem and cerebellum surface, the precontrast 3D T1 data was loaded in the BrainSuite. Cerebral hemispheres were removed using the cerebrum mask leaving behind the brainstem and cerebellum [Figure 13]. | Figure 8: Generation of arterial surface. (a) Inner skull mask (green line) applied on MRA NIfTI file to remove superimposition of the scalp and skull. (b) Creation of mask around arteries by setting appropriate threshold value. (c) Generation of arterial surface
Click here to view |
 | Figure 9: Generation of venous surface. (a) Inner skull mask (green line) applied on MRV NIfTI file to remove superimposition of the scalp and skull. (b) Creation of mask around veins by setting the appropriate threshold value. (c) Generation of venous surface
Click here to view |
 | Figure 10: Generation of skull surface. (a) An inverted mask applied on MRA image using threshold value of approximately 100 to highlight only low signal structures. (b) The resulting mask converted into a label volume and an outer skin mask applied to remove everything outside the skull. (c) 3D skull surface with sutures. (d) Visualization of inner surface of the skull. (e) Skull mask file obtained from FSL-BET tool opened as a 3D image file. (f) Generation of a smooth 3D skull surface
Click here to view |
 | Figure 11: Generation of tumor surface by “region growing method.” (a) Small portion is marked inside the tumor using mask tool. (b) Dilation of the mask using conditional dilation. (c) Manual editing to remove adjacent vessels from tumor mask. (d) 3D tumor surface
Click here to view |
 | Figure 12: Generation of cortical surface. (a) Pre-contrast 3D T1-weighted image is loaded and tumor portion removed (star) using tumor mask. Whole brain mask applied to remove everything outside the brain. (b) Application of cerebrum mask to remove the brainstem and cerebellum. (c) Detection of inner cortical surface. Generation of pial (cortical) surface is done by depositing layers over the inner cortical surface. (d) Smoothed 3D cortical surface
Click here to view |
 | Figure 13: Generation of the brainstem and cerebellum surface. (a) Whole brain mask used to remove the skull and scalp. (b) Cerebrum mask used to remove bilateral cerebral hemispheres. (c) Setting up appropriate threshold to remove CSF. (d) 3D surface of brainstem and cerebellum
Click here to view |
| » Visualization of Surfaces | | |
Once all the desired surfaces were generated, they were copied in a separate folder. A T1-weighted dataset was opened in BrainSuite to review the surfaces. Surface display toolbox was opened and all the surfaces were loaded one by one. To view the complete surface properly, the “show slices” icon was unclicked on “3D display properties.” Different surfaces were given different colors. To show or hide any surface, “show” icon on the surface delineation tool bar was checked. Translucency option was used to make a structure transparent. The surfaces were visualized by rotating and zooming. Desired snapshots were saved as.png extension.
| » Tips and Tricks | | |
Working on the segmentation software is similar to working on Adobe Photoshop. Concept of the mask, layer, threshold, and editing are similar in both. The most important processing tool is the mask. Whenever a region or structure is extracted, its mask should always be saved for future use. For example, a tumor mask could be used to remove the tumor region from venography data if the tumor is showing marked enhancement. Otherwise, enhancing tumor tissue will be displayed as a venous structure. Another important concept is region growing by seed placement, which is especially useful for delineating the vessels and enhancing tumors. Sometimes edges defined by software using default settings are imperfect and adjustment of segmentation parameters is needed. Even after these adjustments, slight manual editing is required to obtain good results. In the setting of brain edema, the signal intensity of white matter resembles the gray matter on T1-weighted images. In such cases, the software could not differentiate between gray matter and white matter and manual intervention was needed. Surfaces could also be generated for CSF, white matter, and various deep nuclei [Figure 14]. Automatic surface labeling with standard anatomical Atlas More Detailses also allow for the identification of different brain gyri. Once surfaces are generated, they can be saved for future review. The axial, coronal, and sagittal reconstructions can be simultaneously visualized with surface images. Review of 3D surfaces is also possible in operation theaters as these softwares can be installed in small laptops. The volume of tumor and brain can be estimated. | Figure 14: (a) 3D surface of white matter. (b) 3D surfaces of CSF, basal ganglia, and thalami. (c) Automatic labeling of cerebral gyri using a standard anatomical atlas
Click here to view |
| » Neurosurgeon's Perspective and Illustrative Cases | | |
From a neurosurgeon's point of view, this kind of software utility makes operative planning easy, especially for young neurosurgeons. A more accurate scalp incision and craniotomy may be performed by visualizing the tumor through skull by making the skull transparent on imaging. Though skull surface generated from MRI images is of inferior quality compared with those produced from CT data, it is still sufficient to plan the incision and craniotomy and is obtained without subjecting the patient to any radiation.
Case 1
A 38-year-old lady presented with a middle-third falcine meningioma extending on the right side more than the left side [Figure 15]. With the help of 3D reconstruction, the exact relationship of the tumor was delineated with respect to the pre-and postcentral gyrus, as well as the veins draining into the superior sagittal sinus. The venous structures and opposite cerebral cortex behind the tumor could also be visualized by making tumor transparent, which are structures that are, otherwise, hidden during surgery. | Figure 15: A 38-year-old female with middle third falcine meningioma. (a) Visualization of cortical surface with superior sagittal sinus and cortical veins. Tumor is hidden by cerebral hemispheres. (b) Visualization of opposite cortex and arterial anatomy through the tumor. (c) Relationship of the tumor with arterial and venous structures. (d) Coronal post contrast T1-weighted image
Click here to view |
Case 2
A 41-year-old male patient presented with a right insular lesion, which was proven later to be a diffuse astrocytoma WHO grade II [Figure 16]. The most important surgical step in this case was identification of the long middle cerebral artery perforators supplying the centrum semiovale and also the lateral lenticulostriate branches supplying the basal ganglia and internal capsule. The exact location of the arteries could be appreciated preoperatively. The next step of medial dissection up to the basal ganglia is very illusive as the nutmeg intraoperative appearance is difficult to appreciate even by the trained surgeon. The right basal ganglia surface was manually generated in this case, which was helpful during the actual surgical dissection. | Figure 16: A 41-year-old male patient with right insular mass. (a) Axial T1-weighted image showing the large hypointense mass. (b) Visualization of cortical surface with veins. The tumor is hidden by the cerebral hemisphere. (c and d) The tumor is encasing the right middle cerebral artery and its branches. The location of the right basal ganglia (green) is seen in relation to the tumor
Click here to view |
Case 3
A 38-year-old male patient presented with a left medial wing sphenoid meningioma [Figure 17]. The major concern was orientation of the major vessels such as the internal cerebral, anterior cerebral, and middle cerebral arteries. The appreciation of vessel shift on 3D imaging becomes difficult, especially after positioning the patient in a nonanatomic position during surgery. The reconstructed images were rotated into the patient's position, which helped in a better understanding of the orientation of vessels. The close relationship of the left superficial middle cerebral vein to the tumor was also anticipated, which, however, could not be preserved during surgery. | Figure 17: A 38-year-old male patient with left medial wing sphenoid meningioma. (a) Postcontrast axial T1-weighted image showing a large enhancing mass. (b) Visualization of the cortical surface with venous anatomy. (c and d) Relationship of the tumor with cerebral arteries
Click here to view |
Case 4
A 30-year-old-female patient presented with a left frontal glioma with left caudate nucleus extension [Figure 18]. As the gliomas are infiltrative in nature, the aim of surgery was maximum safe dissection. Although this was a nonsurfacing lesion located at a depth of 1cm, the extent of tumor could be seen in relation to the gyri by making cerebral cortex transparent. The 3D images helped in deciding the cortisectomy site at the right middle frontal gyrus. The surrounding eloquent cortex could be appreciated well, and the terminal portions of lesion that had to be left out of the surgical decompression were better planned preoperatively, especially in the region of the caudate nucleus. | Figure 18: A 30-year-old female patient with a left frontal glioma. (a) Postcontrast axial T1-weighted image showing a solid-cystic tumor. The cystic component is hyperintense. (b) Visualization of the cortical surface with venous anatomy. (c and d) Relationship of the tumor (yellow, enhancing tumor; brown, cystic component) with arteries and the caudate nucleus (green)
Click here to view |
| » Shortcomings of Presented Method | | |
There are many limitations of the method described. The poor delineation of small cortical arteries and the inability to delineate cranial nerves with the present method needs to be addressed. To enhance the segmentation and visualization of vessels, use of a computational filter has been suggested by Hsu et al.[6] The other limitations are that there are many steps involved in the successful rendering of the final image and the processing time is long. The processing of images takes approximately 3–4 hours per case; however, the actual time spent during manual processing is 30–45 minutes, with the rest of the time spent on data processing. Distortion of the brain anatomy due to mass effect from a large tumor and perilesional edema may result in poor quality surface generation [Figure 19]. The presence of motion artifacts and poor quality images makes brain extraction difficult and may lead to an inaccurate mask production [Figure 20]. | Figure 19: (a) A large tumor causing mass effect and distortion of brain anatomy. (b) Poor generation of the cortical surface over the mass
Click here to view |
 | Figure 20: (a) Curvilinear hypointensity noted over the cortical surface on T1-weighted image due to equipment malfunctioning. (b) Poor generation of bilateral cortical surfaces
Click here to view |
| » Possibilites for Improvement and Future Scope | | |
The collaboration with computer software engineers is needed to develop an automatic image processing pipeline, thereby reducing the number of steps and user supervision. Hsu et al., have developed an automatic image processing pipeline for the surface reconstruction of cerebral vasculature, including gray and white matter, CSF, skull, and scalp using conventional MR sequences.[6] However, they have processed MR data of two healthy individuals only. Necessary hardware modifications are also needed for a faster image processing. Optimization of MRI sequences for producing a high contrast between different structures will also improve surface quality. In addition to the visualization of anatomical structure, relationship of the tumor with white matter tracts and eloquent cortices is also very important. Some progress has already been made in this direction. Interactive diffusion tensor tractography visualization for neurosurgical planning has been developed by Golby et al.[7] The DTI and fMRI data were used to create and visualize white matter tracts in 3D in relation to the surgical lesion. Non-rigid alignment of the preoperative MRI, fMRI, and DT-MRI with intraoperative MRI has been done by Archip N et al., for an enhanced visualization and for navigation in image-guided Neurosurgery.[8] Integration of the 3D data of various structures into the neuronavigation system will make surgeries easy thus enabling the surgeon to be more confident during the operative dissection.[18],[19] Another potential application is the 3D printing of segmented structures, which is an emerging technique for neurosurgical training and preoperative planning.[20]
| » Conclusion | | |
For surgical planning of intracranial tumors, identification of adjacent vascular structures and brain surface landmarks is extremely important. The simultaneous 3D visualization of tumor, brain surface, vasculature, and skull from routine MR sequences is possible using commercial or open source softwares. Further optimization and integration of various tools into a single fully-automated software is needed to make the process of 3D visualization fast and user friendly.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
| » References | | |
| 1. | Kikinis R, Gleason PL, Moriarty TM, Moore MR, Alexander E 3 rd, Stieg PE, et al. Computer-assisted interactive three-dimensional planning for neurosurgical procedures. Neurosurgery 1996;38:640-9.  |
| 2. | Nakajima S, Atsumi H, Bhalerao AH, Jolesz FA, Kikinis R, Yoshimine T, et al. Computer-assisted surgical planning for cerebrovascular neurosurgery. Neurosurgery 1997;41:403-9. |
| 3. | Mert A, Buehler K, Sutherland GR, Tomanek B, Widhalm G, Kasprian G, et al. Brain tumor surgery with 3-dimensional surface navigation. Neurosurgery 2012;71 (2 Suppl Operative):ons286-94. |
| 4. | Kin T, Nakatomi H, Shojima M, Tanaka M, Ino K, Mori H, et al. A new strategic neurosurgical planning tool for brainstem cavernous malformations using interactive computer graphics with multimodal fusion images. J Neurosurg 2012;117:78-88. |
| 5. | Harput MV, Gonzalez-Lopez P, Türe U. Three-dimensional reconstruction of the topographical cerebral surface anatomy for presurgical planning with free OsiriX Software. Neurosurgery 2014;10(Suppl 3):426-35. |
| 6. | Hsu CY, Schneller B, Ghaffari M, Alaraj A, Linninger A. Medical image processing for fully integrated subject specific whole brain mesh generation. Technologies 2015;3:126-41. |
| 7. | Golby AJ, Kindlmann G, Norton I, Yarmarkovich A, Pieper S, Kikinis R. Interactive diffusion tensor tractography visualization for neurosurgical planning. Neurosurgery 2011;68:496-505. |
| 8. | Archip N, Clatz O, Whalen S, Kacher D, Fedorov A, Kot A, et al. Non-rigid alignment of pre-operative MRI, fMRI, and DT-MRI with intra-operative MRI for enhanced visualization and navigation in image-guided neurosurgery. Neuroimage 2007;35:609-24. |
| 9. | Jenkinson M, Bannister P, Brady M, Smith S. Improved optimization for the robust and accurate linear registration and motion correction of brain images. Neuroimage 2002;17:825-41. |
| 10. | Despotović I, Goossens B, Philips W. MRI segmentation of the human brain: Challenges, methods, and applications. Comput Math Methods Med 2015;2015:450341. |
| 11. | Dubey RB, Hanmandlu M, Gupta SK, Gupta SK. The brain MR image segmentation techniques and use of diagnostic packages. Acad Radiol 2010;17:658-71. |
| 12. | Peiró J, Formaggia L, Gazzola M, Radaelli A, Rigamonti V. Shape reconstruction from medical images and quality mesh generation via implicit surfaces. Int J Number Meth Fl 2007;53:1339-60. |
| 13. | Smith SM. Fast robust automated brain extraction. Hum Brain Mapp 2002;17:143-55. |
| 14. | Jenkinson M, Pechaud M, Smith S. BET2: MR-based estimation of brain, skull and scalp surfaces. In Eleventh Annual Meeting of the Organization for Human Brain Mapping, 2005. |
| 15. | Shattuck DW, Leahy RM. BrainSuite: An automated cortical surface identification tool. Med Image Anal 2002;6:129-42. |
| 16. | Woods RP, Grafton ST, Holmes CJ, Cherry SR, Mazziotta JC. Automated image registration: I. General methods and intrasubject, intramodality validation. J Comput Assist Tomogr 1998;22:139-52. |
| 17. | Woods RP, Grafton ST, Watson JDG, Sicotte NL, Mazziotta JC. Automated image registration: II. Intersubject validation of linear and nonlinear models. J Comput Assist Tomogr 1998;22:153-65. |
| 18. | Mohindra N, Neyaz Z. Magnetic resonance sequences: Practical neurological applications. Neurol India 2015;63:241-9. [ PUBMED] [Full text] |
| 19. | Manglore S, Bharath RD, Panda R, George L, Thamodharan A, Gupta AK. Utility of resting fMRI and connectivity in patients with brain tumor. Neurol India 2013;61:144-51. [ PUBMED] [Full text] |
| 20. | Anderson TI. Efficient brain MRI segmentation for 3D printing applications. In Conference on Visual Information Engineering; 2008. |
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7], [Figure 8], [Figure 9], [Figure 10], [Figure 11], [Figure 12], [Figure 13], [Figure 14], [Figure 15], [Figure 16], [Figure 17], [Figure 18], [Figure 19], [Figure 20]
| This article has been cited by | | 1 |
3D Reconstruction Methods Purporting 3D Visualization and Volume Estimation of Brain Tumors |
|
| Aju D., Sushitha Susan Joseph | | International Journal of e-Collaboration. 2022; 18(1): 1 | | [Pubmed] | [DOI] | |
|
|
|
|
