Specific aim #2: Study strategies to increase the feasibility of intraoperative implementation of the atlas-based method. They extracted the surface of the cortex and lateral ventricles from pre- and intraoperative MR images. The elastic strains of the solid phase and the pressure gradients of the fluid phase are coupled together.
As for the brain motion compensation framework, the approach starts with the segmentation of the object of interest, i.e. The need for anatomical accuracy must be balanced with computational efficiency, and for this reason the structures chosen for segmentation. brain, tumor and dural septum. -automatic segmentation: Falx cerebri and tentorium cerebelli are two important substructures of the dural septum.
Falks were manually segmented by drawing on a midsagittal slice of the brain. The segmentation steps are described in the diagram below. a) and (b) show a selection of three points used to cut a plane in the mesh, (c) shows a cut plane (with these three points) overlaid with mesh and falx, (d) shows a cut plane segmented into an approximate structure in of the tentorium shape, and (e) shows the segmented plane with the finite surface of the tentorium created by transforming the plane in (d) using the thin plate gluing algorithm. With the framework components mentioned above, Figure 6 shows a schematic of the entire process, from image acquisition to model update in the operating room.
The atlas-based method of brain displacement correction shifts the majority of the computational burden preoperatively.
Intraoperative Brain Shift Compensation: Accounting for Dural Septa
Results are reported to reflect the difference in the magnitude of the deformation occurring during surgery, as opposed to the pre- and post-operative measurements used in Dumpura et al. The dural septa (not shown) are included in the model by assigning them a slip boundary condition. The Optimization Toolbox MATLAB® (Mathworks Inc) implementation of the Lagrange multiplier method was used to solve this linear optimization problem.
The mean displacement correction for the mesh without the inclusion of the dural septa is 68±17%. While the mean difference in percent shear correction in the remaining six patient cases for the mesh with and without the dural septa was 5±3%, the difference in shear correction for Patient 1 was 28%. The hyperintense regions in the center of the image represent the sinus enclosed in the falx cerebri.
For the mesh without septa, there is considerable movement in the position of the fold, whereas the centerline remains stable for the mesh with septa. To examine the difference in subsurface displacement caused by the introduction of dural septa into the model, color-coded vector differences in the displacement predictions of the tumor boundaries for the two models are shown in Figure 16. 2, 3], where an atlas of solutions was used to compensate for the inherent uncertainty in the operating room.
No statistical difference was observed in the overall surface displacement correction when comparing the results for the mesh with and without the dural septa. However, when the dural septa are accounted for in the model, the subsurface deformations are in greater agreement with the observations made in previous literature. The inclusion of dural septa in the model has a greater impact on the predicted results if the observed displacement was greater.
Another interesting difference between postoperative and intraoperative data is the contributions of different mechanisms in the atlas. In addition, the magnitude of the regression coefficients was relatively smaller in the gravity atlas. The method corrects an average of 75% of brain movement caused by various factors in the operating room.
Sensitivity analysis and automation for intraoperative implementation of the atlas-based method for brain shift correction
The method for segmenting the brain and dural septa was a manual and laborious process. The remainder of the tetrahedral mesh construction continues with the automated algorithm using [59, 70] as described in [73]. Each of the key orientations was systematically eliminated from the atlas and used as ground truth.
The occurrence of errors in brain tissue segmentation was based on the visual evaluation of the quality of the overlay between the segmentation mask and the brain tissue in the patient's MRI. The results of the automated segmentation of the falx and tentorium were evaluated both quantitatively and qualitatively. By visual assessment of the dural septa, the automated segmentation algorithm provided acceptable results for modeling purposes.
Figures 24 (a) – (e) show the surface of the finite element mesh and the dural septa – the falx and tentorium, created using the automatic segmentation algorithm. The biggest difference in terms of distances is in the anterior region of the false. The overlay of the falx and the tentorium on the MRI images is also shown for the same patient in Figure 24 (g) and Figure 24 (h), respectively.
The blue lines are the results of the automatic segmentation and the red contour are the results of manual segmentation. The overlay images also show the least overlap between the two segmentation methods in the anterior area of the false. The mean error and the standard deviation for each of the different size atlases (Figure 21) are shown in the figure below.
Specifically, an automatic segmentation method for the cerebrum and dural septa was evaluated and the results of a sensitivity analysis to determine the constitution of the deformation atlas were presented. The biggest advantage of automatic segmentation for both the cerebrum and the dural septum is the reduction in time. The number of head orientations is the variable that contributes the largest number of solutions to the size of the atlas.
Integrating retraction modeling into an atlas-based framework for brain shift prediction
It is challenging to accurately locate the retractor before surgery, and the retractor is often displaced during the procedure. The location of the retractor-tissue interface can be located on the mesh intraoperatively by digitizing the retractor with a tracked tool tip. The side in contact with the puller is assigned a fixed displacement along the direction normal to the puller plane and free to move in tangential directions.
The performance of the model would depend on the accuracy of tracking the location of the retractor in the OR. The location of the retractor in the mesh was obtained from the third scan (Figure 36(c)) and used to split the nodes along that plane. For the five phantom data sets, this is reported at the surface points (which were used to constrain the least squares error solution) and subsurface points, using the gravity atlas and the superimposed retraction atlas in the following figure.
This is reflected in the percentage of retraction solutions selected by the superimposed atlas for the first two data sets in Figure 39. Past published works using the atlas-based method do not take into account some intraoperative forces such as resection and retraction [ 3, 73]. The work presented in this paper presents a method to integrate traction modeling into the atlas-based framework.
As shown in Figure 40, the magnitude of the gravitational displacement is smaller in the first two cases compared to the other cases. The location of the retractor in the phantom was determined by localizing the retractor plane and the mesh was divided along the corresponding vertices in the undeformed state. An atlas-based framework for calculating brain displacement accounts for uncertainties in the intraoperative environment by pre-calculating deformations through various perturbation boundary conditions and applied forces.
In the experiment, in addition to the CT data, the location of the embedded markers was also recorded using optical tracking that is usually available during surgery. The errors listed are a combination of image marker localization error and registration error. The shift correction results obtained using the trace data are shown in the figure below.
Preliminary work towards computational tumor growth model of a space occupying lesion for estimation of stresses associated with resection
The prognosis depends on the patient's age at diagnosis and the histological tumor type. The second term represents the proliferative component shown as a function of cell concentration. In the third implementation, the mechanical balance was driven by displacements of the tumor front.
At the inner boundary, the displacement is given by the difference in the position of the tumor front at that time and the threshold size. Model 3' consists only of the portion of the tumor cell domain that excludes the threshold tumor size. However, even when taking into account a faster onset, the rate of invasion is slower for Model 2. The stresses and thus the diffusion coefficients depend on the.
Initially there are larger concentration gradients in the inner part of the domain, but towards the outside the gradient is very small. As the model progresses in time, the cell concentration in the inner part of the domain levels off due to logistic growth. In the results sections, part of the domain was truncated, with the results of Model 2 presented in Figures 49-51.
Since the terms σrθ and σθr are not in equation (7.8), the remaining terms can be used to solve the stress equations in terms of strain. An example of a vessel map calculated between two images is shown in Figure 55 below. We hypothesize that the expansion of the matrix M and vector usparse would better condition the inverse problem.
The interpolation errors would be much larger on the right side of the image due to the lack of data in that area. Biomechanical modeling of the human head for physically based non-rigid image registration”, IEEE Transactions on Medical Imaging, vol. Registration of 3-D intraoperative brain MR images using a finite element biomechanical model”, IEEE Transactions on Medical Imaging, vol.
Dawant, "The adaptive Baser algorithm for intensity-based non-rigid image registration," Medical Imaging, IEEE Transactions on, vol. Mollica, "The role of stress in the growth of a multicellular spheroid," Journal of Mathematical Biology, vol.
