Method for optimization of radiation therapy planning

• US 5,602,892 A
• Filed: 03/21/1996
• Issued: 02/11/1997
• Est. Priority Date: 03/21/1996
• Status: Expired due to Term

• Current Assignee: EC Engineering Consultants, LLC
• Original Assignee: Jorge Llacer
• Inventors: Llacer, Jorge
• Primary Examiner(s): Porta, David P.
• Assistant Examiner(s): Bruce, David Vernon
• Application Number:
  US08/619,305
• Time in Patent Office: 327 Days
• Field of Search: 364/413, 364/26, 378/65
• US Class Current: 378/65
• CPC Class Codes: A61N 5/1031 using a specific method of dose optimization
• CPC Class Codes: G16H 50/50 for simulation or modelling of medical disorders

Abstract

A method for optimization of radiation therapy planning based on a new Dynamically Penalized Likelihood (DLP) algorithm. The target function of the DLP algorithm contains likelihood terms and penalty terms connected to the likelihood terms by a number of dynamically updated penalty coefficients. The method results in a highly uniform dose to the tumor or radiosurgery volume, at the expense of some non-uniformity in the dose delivered to defined sensitive tissues.

Claims

1. An improved method of inverse radiation treatment planning enabling the production and display of a desirable radiation dose distribution from a predetermined number and arrangement of active beams, said method comprising the steps of:

a. calculating optimum fluences for said active beams by the process of maximization of a target function which includes a plurality of likelihood terms and a plurality of penalty terms linked by a plurality of penalization parameters, and

b. modifying said penalization parameters during said process of maximization according to a schedule whose characteristics change dynamically as said process of maximization proceeds.

2. The method of treatment planning according to claim 1, wherein said step of calculating optimum fluences includes the maximization of a target function comprising a plurality of likelihood terms of the form:

##EQU11## wherein j=index for each active beam,

a_j =fluence of a specific active beam,

i=index for each volume element (voxel) in the patient anatomy,

d_i =dose desired in a specific voxel of the treatment volume,

F_ij =matrix element that defines the dose delivered by active beam j to patient voxel i per unit fluence of the beam.

3. The method of treatment planning according to claim 1, wherein said step of calculating optimum fluences includes the maximization of a target function comprising a plurality of penalty terms which are functions of an expression comprising the terms ##EQU12## wherein s_i is the dose desired in voxel i of sensitive tissues.

4. The method of treatment planning according to claim 3, wherein said functions are quadratics of the form:

##EQU13## wherein β

_i is the penalization parameter for voxel i in the patient anatomy.

5. The method of treatment planning according to claim 1, wherein said step of calculating optimum fluences by the process of maximization of a target function is carried out by an iterative formula comprising the expression:

##EQU14## wherein;

the superscript (k) indicates the results of the k^th iteration, ##EQU15## D=region that includes all treatment volume voxels, S=region that includes all sensitive tissue voxels,

C is a normalization constant that is used to require that ##EQU16## after each iteration, and n is an exponent that can be used to accelerate convergence.

6. The method of treatment planning according to claim 1, wherein said step of modifying said penalization parameters includes the steps of:

a) initializing the values of said penalization parameters in a predetermined way at the beginning of said process of maximization, and

b) updating the values of said penalization parameters as a function of expressions comprising the terms ##EQU17## as said process of maximization proceeds.

7. The method of treatment planning according to claim 6 wherein said step of initializing the values of said penalization parameters is carried out by a formula comprising the expression:

##EQU18## wherein β

_i.sup.(initial) is the initial value of the penalization parameter for voxel i in the patient anatomy, K is a constant smaller than unit, ##EQU19## after the first iteration and N.sub.λ

is the number of values of λ

_j that are positive.

8. The method of treatment planning according to claim 6 wherein said step of updating the values of said penalization parameters is carried out by a formula comprising the expression:

##EQU20## wherein;

β

_i.sup.(new) is the updated value of the penalization parameter for voxel i in the patient'"'"'s anatomy, and

β

_i.sup.(old) is the value of the penalization parameter for voxel i in the patient'"'"'s anatomy before updating.

9. An apparatus for optimizing the process of inverse radiation treatment planning enabling the production and display of a desirable radiation dose distribution from a predetermined number and arrangement of active beams, involving computer techniques comprising:

a. means for calculating optimum fluences for said active beams by the process of maximization of a target function which includes a plurality of likelihood terms and a plurality of penalty terms linked by a plurality of penalization parameters, and

b. means for modifying said penalization parameters during said process of maximization according to a schedule whose characteristics change dynamically as said process of maximization proceeds.

10. The apparatus of claim 9, wherein said means for calculating optimum fluences includes means for the maximization of a target function comprising a plurality of likelihood terms of the form:

##EQU21## wherein j=index for each active beam,

a_j =fluence of a specific active beam,

i=index for each volume element (voxel) in the patient anatomy,

d_i =dose desired in a specific voxel of the treatment volume,

F_ij =matrix element that defines the dose delivered by active beam j to patient voxel i per unit fluence of the beam.

11. The means of claim 9, wherein said means for calculating optimum fluences includes means for the maximization of a target function comprising a plurality of penalty terms which are functions of an expression comprising the terms ##EQU22## wherein s_i is the dose desired in voxel i of sensitive tissues.

12. The means of claim 11, wherein said functions are quadratics of the form:

##EQU23## wherein β

_i is the penalization parameter for voxel i in the patient anatomy.

13. The means of claim 9, wherein said step of calculating optimum fluences by the process of maximization of a target function is carried out by an iterative formula comprising the expression:

##EQU24## wherein;

the superscript (k) indicates the results of the k^th iteration, ##EQU25## D=region that includes all treatment volume voxels, S=region that includes all sensitive tissue voxels,

C is a normalization constant that is used to require that ##EQU26## after each iteration, and n is an exponent that can be used to accelerate convergence.

14. The means of claim 9, wherein said step of modifying said penalization parameters includes the means for:

a) initializing the values of said penalization parameters in a predetermined way at the beginning of said process of maximization, and

b) updating the values of said penalization parameters as a function an expression comprising the terms F_ij a_j -s_i as said process of maximization proceeds.

15. The means of claim 14 wherein said step of initializing the values of said penalization parameters is carried out by a formula comprising the expression:

##EQU27## wherein β

_i.sup.(initial) is the initial value of the penalization parameter for voxel i in the patient anatomy, K is a constant smaller than unity, ##EQU28## after the first iteration and N.sub.λ

is the number of values of λ

_j that are positive.

16. The means of claim 14 wherein said step of updating the values of said penalization parameters is carried out by a formula comprising the expression:

##EQU29## wherein;

β

_i.sup.(new) is the updated value of the penalization parameter for voxel i in the patient'"'"'s anatomy, and

β

_i.sup.(old) is the value of the penalization parameter for voxel i in the patient'"'"'s anatomy before updating.