Methods for real-time image guided radiation therapy
Disclosed herein are systems and methods for guiding the delivery of therapeutic radiation using incomplete or partial images acquired during a treatment session. A partial image does not have enough information to determine the location of a target region due to, for example, poor or low contrast and/or low SNR. The radiation fluence calculation methods described herein do not require knowledge or calculation of the target location, and yet may help to provide real-time image guided radiation therapy using arbitrarily low SNR images.
1. A method for delivering a radiation fluence comprising:
acquiring a partial image x i of a target region;
calculating a radiation fluence to be delivered to the target region by multiplying a radiation-firing matrix P and the partial image x i , wherein the radiation-firing matrix P is calculated based on a previously-acquired image X of the target region; and
delivering the calculated radiation fluence to the target region with a therapeutic radiation source.
2. The method of claim 1 , wherein the radiation-firing matrix P is a diagonal matrix of scalars s diag(s), such that P·x i =s⊙x i , where ⊙ is a point-wise product operation.
3. The method of claim 1 , wherein calculating the radiation fluence comprises convolving the radiation-firing matrix P with the partial image x i .
4. The method of claim 1 , wherein the radiation-firing matrix P is a Toeplitz matrix toep(f), such that P·x i =f*x i , where * is a convolution operation and f is a fluence to be delivered to the target region according to a treatment plan.
5. The method of claim 1 , wherein a signal-to-noise ratio (SNR) of the partial image x i is less than the SNR of the image X.
6. The method of claim 1 , wherein the partial image x i is acquired using a tomographic imaging system.
7. The method of claim 1 , wherein the partial image x i comprises a reconstruction from a set of positron annihilation emission paths, X-ray projections, or sub-samples in k-space from a MRI imaging pulse sequence.
8. The method of claim 1 , wherein the image X is a motion de-blurred image.
9. The method of claim 1 , further comprising:
selecting a projection angle α;
calculating a projected fluence f α of the calculated fluence at a radiation beam firing angle α and storing the projected fluence f α in a controller memory.
10. The method of claim 9 , wherein the therapeutic radiation source is in a radiation therapy system, and wherein delivering the calculated radiation fluence comprises:
moving the therapeutic radiation source to the radiation beam firing angle α, the radiation therapy system further comprising a multi-leaf collimator having an array of leaves disposed in a beam path of the therapeutic radiation source;
segmenting the projected fluence f α to collimator leaf position instructions and storing the leaf position instructions in controller memory;
adjusting the position of each of the collimator leaves according to the collimator leaf position instructions; and
emitting a radiation beam from the therapeutic radiation source.
11. The method of claim 10 , further comprising continuously repeating the projected fluence f α calculation and delivering radiation to the target region after each fluence calculation, until a desired fluence is applied.
12. The method of claim 11 , wherein calculating the projected fluence f α comprises:
selecting a projection angle α;
calculating a projection x i,α of the partial image x i at the radiation beam firing angle α and storing the projected partial image x i,α in controller memory; and
multiplying a per-angle radiation-firing matrix P α with the projected partial image x i,α (P α x i,α ), wherein the per-angle radiation-firing matrix P α comprises a set of radiation-firing matrices P i,α for each projection angle α.
13. The method of claim 12 , wherein the per-angle radiation-firing matrix P α is a Toeplitz matrix toep(p α ) that implements a convolution operation P α *x i .
14. The method of claim 12 , wherein the per-angle radiation-firing matrix P α is a diagonal matrix diag(p α ) that implements a pointwise multiplication operation p α ·x i .
15. The method of claim 12 , wherein the partial image x i is generated from a set of positron annihilation emission paths, X-ray projections, or sub-samples in k-space from a MRI imaging pulse sequence.
16. The method of claim 12 , wherein the signal-to-noise ratio (SNR) of the partial image x i is less than the SNR of the image X.
17. The method of claim 12 , wherein multiplying the radiation-firing matrix P and the partial image x i , selecting the projection angle α, calculating the projected fluence f α , moving the therapeutic radiation source, segmenting the projected fluence f α to collimator leaf position instructions, adjusting the position of each of the collimator leaves, and emitting the radiation beam occur within a specified time period after acquiring the partial image x i .
18. The method of claim 17 , wherein the specified time period is less than about 10 seconds.
19. The method of claim 18 , wherein the specified time period is less than about 5 seconds.
20. The method of claim 19 , wherein the specified time period is less than about 1 second.
21. The method of claim 10 , wherein the therapeutic radiation source is mounted on a rotatable gantry configured to rotate at a speed of about 20 RPM or more.
22. The method of claim 1 , further comprising applying a linear contrast filter to the partial image x i .
23. The method of claim 22 , wherein any negative values of the filtered partial image x i are added to a subsequent filtered partial image x i+1 .
24. The method of claim 1 , further comprising calculating a real-time delivered dose estimate.
25. The method of claim 1 , further comprising applying a spatial filter to the partial image x.
26. The method of claim 1 , wherein the previously-acquired image X is a first image and A is a known dose calculation matrix, and the method further comprises calculating an updated radiation-firing matrix P prescan by acquiring a second image X prescan and iterating through matrix values for P prescan such that the following conditions are met:
A·P·X≈A·P prescan ·X prescan.
27. The method of claim 26 , further comprising calculating a dose matrix D prescan based on X prescan , calculating a difference value between D prescan with a dose matrix D that has been calculated based on image X, and if the difference value exceeds a pre-selected threshold, generating a notification that the pre-selected threshold has been exceeded.
28. The method of claim 17 , wherein the specified time interval is about an hour.
29. The method of claim 1 , further comprising preprocessing a plurality of partial images x i with a transform T, wherein the image X is a sum of the preprocessed plurality of partial images x i :
X=ΣT ( x i ).
30. The method of claim 29 , wherein the transform T is a linear transform.
31. The method of claim 29 , wherein the transform T is a non-linear transform.