SYSTEMS AND METHODS FOR ACCELERATED ONLINE ADAPTIVE RADIATION THERAPY
Systems and methods for accelerated online adaptive radiation therapy (“ART”) are described. The improvements to online ART are generally provided based on the use of textural analysis and machine learning algorithms implemented with a hardware processor and a memory. The described systems and methods enable more efficient and accurate online adaptive replanning (“OLAR”), which can also be implemented in clinically acceptable timeframes. For example, OLAR can be reduced from taking 10-30 minutes down to 5-10 minutes.
1 . A method for assessing a quality of a radiation treatment plan, the method comprising:
(a) accessing with a computer system, a first radiation treatment plan for a subject, wherein the first radiation treatment plan includes a first three-dimensional dose map;
(b) accessing with the computer system, a second radiation treatment plan for the subject, wherein the second radiation treatment plan includes a second three-dimensional dose map;
(c) generating at least one spatial dose feature based on the first three-dimensional dose map and the second three-dimensional dose map; and
(d) selecting as a higher quality radiation treatment plan one of the first radiation treatment plan or the second radiation treatment plan based on an analysis of the at least one spatial dose feature.
2 . The method as recited in claim 1 , wherein the at least one spatial dose feature comprises a dose level size zone matrix (DLSZM).
3 . The method as recited in claim 2 , wherein the at least one spatial dose feature comprises a large cold spot feature computed from the DLSZM.
4 . The method as recited in claim 1 , wherein the at least one spatial dose feature comprises a dose level co-occurrence matrix (DLCM).
5 . The method as recited in claim 4 , wherein the at least one spatial dose feature comprises a hot contrast feature computed from the DLCM.
6 . The method as recited in claim 4 , wherein the at least one spatial dose feature comprises an inverse covariance feature computed from the DLCM.
7 . The method as recited in claim 1 , wherein the at least one spatial dose feature comprises a mutual objective function distance.
8 . The method as recited in claim 1 , wherein the at least one spatial dose feature comprises a neighbor level-dose difference matrix.
9 . The method as recited in claim 1 , wherein the at least one spatial dose feature comprises a coarseness feature computed from at least one of the first three-dimensional dose map and the second three-dimensional dose map.
10 . The method as recited in claim 1 , wherein the at least one spatial dose feature comprises a busyness feature computed from at least one of the first three-dimensional dose map and the second three-dimensional dose map.
11 . The method as recited in claim 1 , wherein the at least one spatial dose feature comprises a plurality of spatial dose features, and the higher quality radiation treatment plan is selected from the first radiation treatment plan and the second radiation treatment plan based on a general quality index value computed from plurality of spatial dose features.
12 . The method as recited in claim 11 , wherein the general quality index (GQI) is computed as:
GQI
=
α
1
H
C
1
→
2
+
α
2
I
C
1
→
2
+
α
3
L
C
S
+
α
4
C
o
+
α
5
B
u
+
α
6
M
O
F
D
1
→
2
;
wherein HC 1→2 is a hot contrast between the first radiation treatment plan and the second radiation treatment plan; IC 1→2 is an inverse covariance between the first radiation treatment plan and the second radiation treatment plan; LCS is a large cold spot feature; Co is a coarseness feature; Bu is a busyness feature; MOFD 1→2 is a mutual objective function distance between the first radiation treatment plan and the second radiation treatment plan; and α i for i=1, . . . , 6 are coefficients.
13 . The method as recited in claim 12 , wherein the coefficients are determined from ground truth data.
14 . The method as recited in claim 11 , wherein the higher quality radiation treatment plan is selected from the first radiation treatment plan and the second radiation treatment plan having a lower general quality index value.
15 . The method as recited in claim 1 , wherein the higher quality radiation treatment plan is selected based also on an analysis of a dose volume parameter computed from at least one of the first radiation treatment plan and the second radiation treatment plan.
16 . The method as recited in claim 15 , wherein the dose volume parameter is a dose volume histogram.