IMRT - Conventional Techniques Flashcards
What is IMRT?
A technique where the beam is altered from its flat or wedged profile. This produces dose distributions which conform to irregular target shapes
Why is IMRT used?
Greater dose conformality Greater sparing of normal tissue (lower toxicity/complications) Potential for dose escalation Easy to create non-uniform distributions Potential efficiency savings Can block OAR Can shape to concavities
How can varying the MLCs deliver the desired dose?
Each leaf pair gives a 1D profile which is stacked to give a 2D fluence
For step and shoot, split the profile into a series of static fields to be delivered
Why is the leaf sweep technique used?
Saves time
Less chance of leaves colliding
Why can we not increase the number of segments to infinity to deliver a perfect distribution?
This is less efficient as the beam is suspended more
For very small segments the beam will not be stable for ~2s and the dose/MU can very up to 2% initially
How many control points are associated with each segment?
2 - 1 sets the shape, 1 increments the MU
How are dynamic MLC moves used?
Can create field gradients by sweeping a jaw across the beam to change the beam profile.
Intensity profiles are converted into leaf moves in the TPS - need to consider constraints on leaf velocity, adjacent profiles, treatment time
What is thee dosimetric accuracy most dependent on?
Positional accuracy of collimators
What are the problems that IMRT presents to the linac?
Small segments
Low dose segments - start up characteristics affect low MU segments, MU linearity, flatness, and energy
MLC accuracy
Why is inverse planning used for IMRT?
It is too complex for a person
How is inverse planning performed?
Planner specifies dose distribution through dose limits and volume constraint to PTV and OARs
TPS calculates beam profiles, iterates to reduce cost/objective function
Planner assesses plan and alters dose limits/constraints if needed
What is a cost function?
A physical way of summarising the merit of a plan in a single figure to drive the plan optimisation
What is the simple cost function based on the weighted least squares minimisation algorithm?
C = sum(penalty function.(calculated dose - prescribed dose)^2 This = 0 if all the plan constraints are met
What are the common stopping criterion for the TPS?
Stop after x iterations
Stop when the difference in the cost function between successive plans is a specified percentage
What is the difference between an objective and a constraint?
An objective is something you would like, weighted against other objectives for importance
A constraint is something which must be achieved for the plan to be acceptable
What parameters are used for the optimisation of target volumes and OARs?
Target - min dose, max dose, dose gradient/uniformity parameter
OAR - max dose, dose volume controls
What are the optimisation algorithms on offer?
Analytic techniques - inverse CT
Deterministic iterative techniques - gradient methods - most common in comercial TPS - fine if not optimising gantry too
Stochastic iterative techniques - simulated annealing, genetic algorithms, neural networks - prevents optimisation getting stuck in local minima
How does the gradient descent method work?
Walks with a predefined step size to find the global minima. Needs good first guess and a simple objective function
How does simulated annealing work?
Step size starts large and reduces over time. Always accept a step reducing the cost function. But also accept steps which increase the cost function with probability p(i) = e^(-delta.OF/kb.T) to get out of local minima
How is beamlet optimisation achieved?
Field split into beamlets of discrete fluence, with the corresponding dose distribution computed
Beamlet weights are optimised to produce an optimal fluence map for each beam direction
Optimal fluence map translated into deliverable segments or leaf trajectory - leaf segmentation
The final deliverable dose distribution is slightly degraded
What is aperture based optimisation?
Segment shapes and weights optimised together
Takes limitations of MLCs into account
What is sequencing in IMRT planning?
Converting the theoretical/ideal fluence into deliverable segments
What is the compromise between the number of levels in the optimisation?
Fewer levels - greater efficiency, easier to verify, poorer plan
More levels - closer to optimisation dose distribution, less efficient
What are current optimisations based on an what may they move to in the future?
Currently - absorbed dose
Future - outcomes ie biological effectiveness
What considerations need to be made in inverse planning?
Use odd number of equidistant beams - no opposing beams
Direction less important as number of beams increases
Can come through OAR
Choice of energy doesn’t matter >10MV not used
What is the purpose of tuning volumes?
Drive optimisation to acceptable solution by having more control over dose distribution
How can overlapping ROIs be dealt with?
Set overlap priority
Create rinds to reduce the dose in that area
How is normalisation performed for IMRT?
Normalise to a volume
Can have large absorbed dose gradients in individual beams
Use level 2 reporting
What is a class solution?
A set of inverse planning parameters which can be used for patients with the same disease and staging to speed up planning. Can be a stand alone or a good starting point
What is included in a class solution?
Volumes needed
Beam numbers and orientation
Optimisation parameters
Sequencing parameters
What is the common method for making an IMRT plan?
Make a plan with good dose to the PTV then push on OARs