search

e0404 / matRad

An open source multi-modality radiation treatment planning sytem developed by e0404 @ DKFZ
http://www.matRad.org
Other
222 stars 176 forks source link

dose influence matrix unit #257

Closed pinardursun closed 6 years ago

pinardursun commented 6 years ago

Hi, what is unit of dose influence matrix? Gy per MU or Gy per 100 MU. Thanks

markbangert commented 6 years ago

There is no established normalization of photon fluence in matRad at the moment. However you can do so very fast yourself. Just simulate a rectangular field (e.g. 10x10cm²) and use some reference value e.g. in 10cm depth to renormalize your weight vector accordingly...

pinardursun commented 6 years ago

Thank you for quick reply. But, I am sorry, I could not get the idea. Do I need to do this simulation using phantom etc. Or, can I directly do using the d_ij matrices that matRad provide me and my ct scans? I am not from physics area, and I just want to extract data to use in an VMAT optimization algorithm.. Pınar

markbangert commented 6 years ago

Hi Pinar,

the normalization to MU is usually done by relating the dose at a point within a water phantom to a specific fluence

E.g. you can say that you have 100MU when the dose at a point in 10cm depth on the central ray equals 1Gy when irradiated with a 10x10cm² field.

If you now want to have some MU calibration you have to run such a simulation (i.e. 10x10cm² field) and determine the fluence weight that will give you 1Gy in 10cm depth. And then you have the linear conversion factor that you need or in other words you can compute how many Gy you have at what weight.

Cheers / Mark

On Mon, Mar 19, 2018 at 3:53 PM, pinardursun notifications@github.com wrote:

Thank you for quick reply. But, I am sorry, I could not get the idea. Do I need to do this simulation using phantom etc. Or, can I directly do using the d_ij matrices that matRad provide me and my ct scans? I am not from physics area, and I just want to extract data to use in an VMAT optimization algorithm.. Pınar

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/e0404/matRad/issues/257#issuecomment-374240240, or mute the thread https://github.com/notifications/unsubscribe-auth/AJr8fXSi5TJXpDy4ND2vovYjInUDiEoqks5tf8ZkgaJpZM4SwJBm .

pinardursun commented 6 years ago

Hi Mark, Thank you so much for explanation. I am working on this simulation that you have advises to me.

I use "BOXPHANTOM" data set and obtained dij matrices for gantry angle 0. Then, I determine the voxel in 10 cm depth on the central ray. I sum all dij values corresponding 10x10cm² field and the sum was about 0.92. Let me ask my first simple question: Can I assume that 92 MU radiation is delivered at gantry angle 0? Thus, can I divide my all dij matrice by 92 to obtain Gy/MU values?

My other question is that do I need to do this simulation for all gantry angles? Also, do I need to do this simulation for all other data sets? Or 1/92 is my only linear conversion factor for MatRad?

Kind regards, Pınar

markbangert commented 6 years ago

Hi Pinar,

if you use the Boxphantom and a beam at 0° then your reference conditions relate to a 75mm x 75mm field. Note that your normalization relates to this field size; if you would simulate a smaller(bigger) field you would have less(more) scattering contributions on the central ray.

If you now have a dose of 0.92 in the very voxel you want to consider at unit fluence then I would do the following:

At w = 1/0.92 you have one Gy in your very voxel under consideration, so I would say that you introduce a normalization of 100MU corresponding to 1/0.92 in fluence weight.

Now you can recompute all weights to MU with this scaling! You do not have to recompute everything for every gantry angle. Your normalization globally corresponds to 100MU (i.e. 1/0.92) delivering 1 Gy in 10cm depth on the central ray for a 75mm^2 field.

I hope this solves your problem. Let me know if we can close this issue.

On Thu, Mar 22, 2018 at 11:43 AM, pinardursun notifications@github.com wrote:

Hi Mark, Thank you so much for explanation. I am working on this simulation that you have advises to me.

I use "BOXPHANTOM" data set and obtained dij matrices for gantry angle 0. Then, I determine the voxel in 10 cm depth on the central ray. I sum all dij values corresponding 10x10cm² field and the sum was about 0.92. Let me ask my first simple question: Can I assume that 92 MU radiation is delivered at gantry angle 0? Thus, can I divide my all dij matrice by 92 to obtain Gy/MU values?

My other question is that do I need to do this simulation for all gantry angles? Also, do I need to do this simulation for all other data sets? Or 1/92 is my only linear conversion factor for MatRad?

Kind regards, Pınar

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/e0404/matRad/issues/257#issuecomment-375259339, or mute the thread https://github.com/notifications/unsubscribe-auth/AJr8ffLJ9VeTSbBx5Y27NpnVJQvkEXFLks5tg4BagaJpZM4SwJBm .

pinardursun commented 6 years ago

Hi Mark,

I have already changed the BODY structure from OAR to target in BOXPHANTOM and calculate dij matrice. Thus the size of the field was 275mm^2. But I only consider the 100mm^2 field around central ray and get the fluence which is around 0.92.

Thank you again for your help. I think now I have reasonable values and the problem is solved.

Kind regards, Pınar