InitRech 2015/2016, sujet 9 : Différence entre versions

De Wiki d'activités IMA
(Synthesis)
(Applications:)
 
(5 révisions intermédiaires par le même utilisateur non affichées)
Ligne 20 : Ligne 20 :
 
The experimental result remain from a test on a patient-specific 3D model and shown many things:
 
The experimental result remain from a test on a patient-specific 3D model and shown many things:
  
-The computational time is compatible with a clinical use
+
*The computational time is compatible with a clinical use
  
-Even if the trajectory found in dynamical model are more “dangerous” it's still acceptable
+
*Even if the trajectory found in dynamical model are more “dangerous” it's still acceptable
  
-In case of brain shift, most of safest zones for insertion don't overlap those of no shift simulation
+
*In case of brain shift, most of safest zones for insertion don't overlap those of no shift simulation
  
In conclusion the brain shift prediction are pretty accurate by removing many point who becomes dangerous, and even if the system can be improved, the result are already here.
+
In conclusion the brain shift prediction are pretty accurate by removing many points who becomes dangerous, and even if the system can be improved, the result are already here.
  
 
=Contribution:=
 
=Contribution:=
Ligne 32 : Ligne 32 :
 
As said before, this contribution remain in two mains parts: model and simulation of the brain and trajectory planning for the needle.
 
As said before, this contribution remain in two mains parts: model and simulation of the brain and trajectory planning for the needle.
 
So first, to define the best model of the brain itself many hypothesis are set in order to work with the appropriate tools and stick with the reality.  
 
So first, to define the best model of the brain itself many hypothesis are set in order to work with the appropriate tools and stick with the reality.  
The main points are: the brain is subject of the law of continuum mechanics, so numerical method describe it as finite set of element to solve motion equation. And the other one is the fact that the process take place at a very low velocity so this is a quasi-static problem with a multiple state at equilibrium. This one is solved by using first order linearization at each step. Among those global state hypothesis, they can define relation and parameters through law of biological behaviour like Hook's law.
+
The main points are: the brain is subject of the law of continuum mechanics, so numerical method describe it as finite set of element to solve motion equation. For their problem they chose a group P1 of Lagrange tethaedral elements.
 +
 
 +
And the other one is the fact that the process take place at a very low velocity so this is a quasi-static problem with a multiple state at equilibrium.Thus the equation to solve is f(x)=0. This one is solved by using first order linearization at each step, which give the following equation: f (x+dx) = f (x)+ K(x)dx where K(x) = ∂f / ∂x depends on the nodes position. Among those global state hypothesis, they can define relation and parameters through law of biological behaviour like Hook's law.
  
 
The equilibrium equation is solved with Newton-Raphson algorithm.
 
The equilibrium equation is solved with Newton-Raphson algorithm.
 
The trajectory planning itself is solve as an optimization problem with hard and soft constraint. To do this, surgical rules are formalized as geometrics constraints divided in the two categories quoted above.
 
The trajectory planning itself is solve as an optimization problem with hard and soft constraint. To do this, surgical rules are formalized as geometrics constraints divided in the two categories quoted above.
 
The difference between static model and dynamic one is represented by the intensity of shift, so the best trajectory must pass the constraint test whatever the intensity of the shift between a minimum and a maximum.
 
The difference between static model and dynamic one is represented by the intensity of shift, so the best trajectory must pass the constraint test whatever the intensity of the shift between a minimum and a maximum.
 
 
  
 
=Applications:=
 
=Applications:=
  
 
The main application of this proceed is obviously to increase the security when performing deep brain stimulation for cure many diseases (Parkinson, dystonia) or tremors from car crash accident for instance as it has a very low computational time. Therefore clinical validation are still required.
 
The main application of this proceed is obviously to increase the security when performing deep brain stimulation for cure many diseases (Parkinson, dystonia) or tremors from car crash accident for instance as it has a very low computational time. Therefore clinical validation are still required.
 +
 
But we can pursue the research to increase to increase the quality of the result and even maybe develop a robot who may be able make the surgery with real-time processing between his sensors and the algorithm.
 
But we can pursue the research to increase to increase the quality of the result and even maybe develop a robot who may be able make the surgery with real-time processing between his sensors and the algorithm.
  
Furthermore, the difference between the algorithm information and those bring by the sensor may allow to improve it and get a better understand of the brain behaviour. Maybe some other surgery procedure  may face to the same type of behaviour like those of the vertebral spine which need a lot of accuracy.
+
Furthermore, the difference between the algorithm information and those bring by the sensor may allow to improve it and get a better comprehension of the brain behaviour. Thus we could improve our knowledge about the brain, and use the brain shift at our advantage in other procedure.
 +
 
 +
Maybe some other surgery procedure  may face to the same type of behaviour like those of the vertebral spine which need a lot of accuracy.
 +
 
 +
On the other way, it should be possible to apply those simulation in other case. Indeed, this algorithm have been developed with law of biological model but other parameters may allow to simulate the same type of pressure losses in other case such as surgery procedure on nervous system, where blood leak may cause shift of other organs.  
  
On the other way, it should be possible to apply those simulation in other case. Indeed, this algorithm have been developed with law of biological model but other parameters may allow to simulate the same type of pressure losses in other case such as sounding balloon composed of two balloon and simulate the effect of hole in the first one and consequences on the second one.
+
But other domains may be interested, like aerospace, with inflatable space habits such as the recent one test on ISS: BEAM. Indeed, the cerebrospinal fluid may be replaced by other fluids in the model. So when you have an inflatable device with two layers separated by a fluid, we may be able to understand who the second one would react in case of leak caused by a hole in the first one.
Or in an inflatable space habits such as the recent one test on ISS: BEAM.  
 
  
In conclusion, once this system will be approved by clinical validation, it may quickly improve decease the rate of failure and by the way, side effect of those type of procedure.
+
In conclusion, once this system will be approved by clinical validation, it may quickly improve safeness of Deep Brain Stimulation, deceasing the rate of failure and by the way, side effect of those type of procedure.
 
And it should be even more true for urgent procedures where doctors cannot spend time to analyse the patient's brain.
 
And it should be even more true for urgent procedures where doctors cannot spend time to analyse the patient's brain.
 +
Moreover they should be other domains which could be interested in.

Version actuelle datée du 19 juin 2016 à 20:41

Synthesis

Brain stimulation in neurosurgery procedure involve the use of deeply implanted electrodes into the brain. The target depends of the procedure but many rules exists to ensure that the trajectory of the electrodes will ensure the result as well as the patient's safety.

However a problem remain unsolved in those procedures: the shift of brain tissues after the skull drilling. This one is caused by the lose of pressure in brain environment as the cerebrospinal fluid in which bathe the brain leak from the drill's hole.

First, the article develop how deep brain stimulation and brain shift are treated nowadays.

Currently, there are many models available from the most simple: mass-spring-damper model, not very accurate but low computational time. To the most used FEM, which can be set with a lot of parameters. So based on those models, monitoring of the patient and set of rules, surgeons can plan the best way to implant the electrodes.

Therefore this process remains quite long and difficult for this one, that's why the contributors developed a an automatic planning system including simulations.

Then the article develop the main contribution of searchers.

The solution remains in the use of models and simulations of the brain which mean a study of: the brain deformations, the cerebrospinal fluid model and the deformation constraints, which occurs by contact with the endocranium. Then, those studies are coupled with a trajectory planning in two phases : static environment and dynamic environment. Once the dynamic environment is generated, they compute the feasible insertion zone, and then the optimized trajectory for a given point.

Finally, it shows how their solution can solve a part of the brain shift problem by the experimentation. The experimental result remain from a test on a patient-specific 3D model and shown many things:

  • The computational time is compatible with a clinical use
  • Even if the trajectory found in dynamical model are more “dangerous” it's still acceptable
  • In case of brain shift, most of safest zones for insertion don't overlap those of no shift simulation

In conclusion the brain shift prediction are pretty accurate by removing many points who becomes dangerous, and even if the system can be improved, the result are already here.

Contribution:

As said before, this contribution remain in two mains parts: model and simulation of the brain and trajectory planning for the needle. So first, to define the best model of the brain itself many hypothesis are set in order to work with the appropriate tools and stick with the reality. The main points are: the brain is subject of the law of continuum mechanics, so numerical method describe it as finite set of element to solve motion equation. For their problem they chose a group P1 of Lagrange tethaedral elements.

And the other one is the fact that the process take place at a very low velocity so this is a quasi-static problem with a multiple state at equilibrium.Thus the equation to solve is f(x)=0. This one is solved by using first order linearization at each step, which give the following equation: f (x+dx) = f (x)+ K(x)dx where K(x) = ∂f / ∂x depends on the nodes position. Among those global state hypothesis, they can define relation and parameters through law of biological behaviour like Hook's law.

The equilibrium equation is solved with Newton-Raphson algorithm. The trajectory planning itself is solve as an optimization problem with hard and soft constraint. To do this, surgical rules are formalized as geometrics constraints divided in the two categories quoted above. The difference between static model and dynamic one is represented by the intensity of shift, so the best trajectory must pass the constraint test whatever the intensity of the shift between a minimum and a maximum.

Applications:

The main application of this proceed is obviously to increase the security when performing deep brain stimulation for cure many diseases (Parkinson, dystonia) or tremors from car crash accident for instance as it has a very low computational time. Therefore clinical validation are still required.

But we can pursue the research to increase to increase the quality of the result and even maybe develop a robot who may be able make the surgery with real-time processing between his sensors and the algorithm.

Furthermore, the difference between the algorithm information and those bring by the sensor may allow to improve it and get a better comprehension of the brain behaviour. Thus we could improve our knowledge about the brain, and use the brain shift at our advantage in other procedure.

Maybe some other surgery procedure may face to the same type of behaviour like those of the vertebral spine which need a lot of accuracy.

On the other way, it should be possible to apply those simulation in other case. Indeed, this algorithm have been developed with law of biological model but other parameters may allow to simulate the same type of pressure losses in other case such as surgery procedure on nervous system, where blood leak may cause shift of other organs.

But other domains may be interested, like aerospace, with inflatable space habits such as the recent one test on ISS: BEAM. Indeed, the cerebrospinal fluid may be replaced by other fluids in the model. So when you have an inflatable device with two layers separated by a fluid, we may be able to understand who the second one would react in case of leak caused by a hole in the first one.

In conclusion, once this system will be approved by clinical validation, it may quickly improve safeness of Deep Brain Stimulation, deceasing the rate of failure and by the way, side effect of those type of procedure. And it should be even more true for urgent procedures where doctors cannot spend time to analyse the patient's brain. Moreover they should be other domains which could be interested in.