Feasibility of EDose for pre-treatment dose verification
A γ value of 1.0 or less indicates that a particular point falls within the criteria of 3% dose difference and 3mm DTA and, therefore, is a passing point. The γ passing rates obtained from the EDose, Seven29, and 3Dmap were compared with the TPS calculations. The average (avg.), minimum (min.), maximum (max.), and standard deviation (SD) of the γ passing rates and γmean of 50 selected cases for the criteria 3%/3mm, 3%/2mm, and 2%/2mm were shown in Table 1.
All plans’ average γ passing rates of the 3%/3mm criterion was > 98% for EDose, Seven29, and 3Dmap. The corresponding average γ passing rates when using criteria of 3%/2mm and 2%/2mm were > 95% and 90%. So, all plans were acceptable according to AAPM Task Group 119 that the planar dose distributions were assessed using gamma criterion of 3%/3 mm. Additionally, The average γmean of the three dose verification methods were similar for the 3%/3mm (0.35, 0.38, 0.35), 3%/2mm (0.41, 0.48, 0.43) and 2%/2mm criteria (0.50, 0.56, 0.55).
The SD of γ passing rates of the 3%/3mm criterion was 1.69%, 1.61%, 1.37% for EDose, Seven29, 3Dmap severally, and the corresponding the 3%/2mm criterion was 2.89%, 3.04%, 2.17%. Therefore, the narrow range of SD showed the fine reproducibility and stability of the response obtained using the EDose system, and consistent with the other two methods.
Table 1
Overview of avg., min., max., and SD value of γ passing rates and γmean obtained using EDose, Seven29 and 3Dmap for 50 cases.
|
3%/3mm
|
|
3%/2mm
|
|
2%/2mm
|
EDose
|
Seven29
|
3Dmap
|
|
EDose
|
Seven29
|
3Dmap
|
|
EDose
|
Seven29
|
3Dmap
|
γ passing rate
|
avg. [%]
|
98.12
|
98.35
|
98.53
|
|
95.53
|
96.53
|
95.41
|
|
91.59
|
94.69
|
90.90
|
min. [%]
|
93.32
|
93.60
|
93.98
|
|
88.61
|
87.20
|
88.55
|
|
74.04
|
82.50
|
78.25
|
max. [%]
|
99.85
|
100.00
|
100.00
|
|
99.17
|
100.00
|
98.38
|
|
98.38
|
100.00
|
96.69
|
SD [%]
|
1.69
|
1.61
|
1.37
|
|
2.89
|
3.04
|
2.17
|
|
5.53
|
3.76
|
3.81
|
γmean
|
avg.
|
0.35
|
0.38
|
0.35
|
|
0.41
|
0.48
|
0.43
|
|
0.50
|
0.56
|
0.55
|
min.
|
0.25
|
0.27
|
0.23
|
|
0.29
|
0.29
|
0.30
|
|
0.35
|
0.40
|
0.38
|
max.
|
0.57
|
0.56
|
0.45
|
|
0.69
|
0.72
|
0.58
|
|
0.81
|
0.87
|
0.73
|
SD
|
0.07
|
0.07
|
0.05
|
|
0.08
|
0.09
|
0.06
|
|
0.10
|
0.10
|
0.08
|
Small differences were observed in γ passing rates and γmean when comparing the three dose verification methods (EDose, Seven, and 3Dmap) for all the selected cases. Obtained average γ passing rates were similar with 98.12%, 98.35%, 98.53% for EDose, Seven, 3Dmap respectively to 3%/3mm criterion, and corresponding 95.53%, 96.53%, 95.54–3%/2mm criterion. However, Seven 29 exhibited higher average γ passing rates (94.69%) compared to EDose (91.59%) and 3Dmap (90.90%) to 2%/2mm criterion.
The γ passing rates and γmean for the 5 kinds of disease obtained from EDose, Seven29, and 3Dmap were shown in Fig. 3 and Fig. 4. We could see the average γ passing rates of the larynx cancer were consistent among the three dose verification methods for three passing rates criteria (3%/3mm, 3%/2mm, 2%/2mm). For the 3%/3mm criterion, the average γ passing rates was acceptable in all the nasopharyngeal, glioblastoma, cerebral metastatic, parotid gland, and larynx cancer groups, with an average rates of (96.60 ± 2.17)%, (98.83 ± 1.77)%, (98.85 ± 1.05)%, (98.12 ± 1.24)%, (98.20 ± 1.18)% for EDose, (97.43 ± 1.37)%, (98.03 ± 1.48)%, (98.87 ± 1.33)%, (99.18 ± 1.35)%, (98.22 ± 2.07)% for Seven29, and (98.32 ± 1.14)%, (98.63 ± 1.24)%, (98.44 ± 1.85)%, (98.32 ± 1.37)%, (98.92 ± 1.32)% for 3Dmap, respectively.
The differences among the three methods gradually increased with increasing the criteria from 3%/3mm to 3%/2mm to 2%/2mm, shown in Fig. 5. The maximum difference (4.49%) of the γ passing rate was found in the nasopharyngeal cancer (NPC) group at the 2%/2mm criterion, while the differences of the other four groups were both < 3%, and the larynx cancer was only 0.88%. The corresponding differences at 3%/2mm, 3%/3mm criteria were < 2%, 1% separately and the larynx cancer was < 0.5%.
Fig. 4γ passing rates for the 5 kinds of disease obtained from EDose,Seven29 and 3Dmap measurements respectively. (a) 3%/3mm criteria; (b) 3%/2mm criteria; (c) 2%/2mm criteria; The columns represent the mean values of the γ passing rates, and the error bars are the correspondingSD.
Fig. 5 γmean for the 5 kinds of disease obtained from EDose,Seven29 and 3Dmap measurements respectively. (a) 3%/3mm criteria; (b) 3%/2mm criteria; (c) 2%/2mm criteria; The columns represent the mean values of the γmean and the error bars are the correspondingSD.
Action level
With a total of 287 clinical patients’ plans verified by EDose, the γ passing rates were summarized with various disease sites under the three criteria 3%/3mm, 3%/2mm, and 2%/2mm, illustrated in Table 2 and Fig. 6. We set action levels for our clinic using the gamma values calculated. Specifically, we established a set of action levels based on the institutional average values and standard deviations.
The percentage of points passing the 3%/3mm criterion was on average (98.11 ± 2.05)% for brain case, (97.34 ± 2.20)% for H&N case, (98.10 ± 2.97)% for the abdomen, (95.59 ± 2.59)% for pelvis case, (96.53 ± 2.33)% for the bone case, and (97.79 ± 1.23)% for other cases. The overall results were (97.46 ± 2.31)%. Thus, the corresponding action level was 93%, according to the definition described in TG-119 reports. The average percentage of points passing the criteria of 3%/2mm and 2%/2mm were (94.87 ± 3.78)% and (89.29 ± 7.49)%, well the corresponding action levels of 87% and 75%, respectively.
Table 2
γ passing rates of 287 cases with EDose method for 3%/3mm, 3%/2mm, and 2%/2mm criteria in different disease sites.
sites
|
number
of cases
|
3%/3mm [%]
|
|
3%/2mm [%]
|
|
2%/2mm [%]
|
avg.±SD
|
range
|
|
avg.±SD
|
range
|
|
avg.±SD
|
range
|
Brain
|
102
|
98.11 ± 2.05
|
92.45–99.96
|
|
95.8 ± 3.74
|
82.65–99.44
|
|
90.20 ± 8.14
|
68.09–98.80
|
H&N
|
127
|
97.34 ± 2.20
|
90.85–99.84
|
|
94.42 ± 3.69
|
83.80–99.11
|
|
89.15 ± 7.00
|
68.83–98.28
|
Abdomen
|
10
|
98.10 ± 2.97
|
91.16–99.92
|
|
96.52 ± 4.52
|
86.37–99.66
|
|
91.07 ± 10.37
|
71.64–99.15
|
Pelvis
|
19
|
95.59 ± 2.59
|
91.34–99.99
|
|
92.80 ± 3.83
|
86.19–99.93
|
|
84.82 ± 6.62
|
73.74–99.22
|
Bone
|
23
|
96.53 ± 2.33
|
91.82–99.74
|
|
94.06 ± 3.35
|
84.93–99.20
|
|
88.60 ± 5.33
|
76.56–97.70
|
Other
|
6
|
97.79 ± 1.23
|
96.08–99.06
|
|
95.61 ± 2.10
|
92.47–98.10
|
|
91.12 ± 5.68
|
80.75–95.70
|
Total
|
287
|
97.46 ± 2.31
|
90.85–99.99
|
|
94.87 ± 3.78
|
82.65–99.93
|
|
89.29 ± 7.49
|
68.09–99.22
|
3D online (in vivo) verification
28 randomly head and neck patients including 17 nasopharyngeal carcinomas, six laryngeal cancer, five oropharyngeal cancer) had been selected for conducting 3D in vivo dose verification at each treatment fraction. The average γ pass rate under the 3%/3 mm criterion for all 28 patients was obtained by comparing the 3D dose-reconstruction results from the EDose and the calculation results from the TPS, shown in Fig. 7. The results showed that the in vivo γ pass rate had a decreasing trend during the 33 treatment fractions. The γ passing rates means ± SD of the first fraction was (91.92 ± 3.31)% while the 33th fraction was (85.73 ± 8.75)%. The dotted line was the linear regression of the average γ pass rate and a linear coefficient of determination y = -0.1624x + 89.49, R2 = 0.1855.
The difference analysis of the DVH index for the first, the seventh, and the thirty-third were shown in Table3. The comparison results are expressed as the means ± SD for 28 head and neck patients. As illustrated in Table 3,
Compared to the TPS calculations, the average D98 of the PTVs obtained using EDose for the first treatment showed that the deviations were less than 3%, except the PGTVnd, which was − 3.72%. While the seventh treatment was more than 3% except the PGTVnx was − 2.63%, and the thirty-third treatment were all less than 3%. The ∆Dmean of the PGTVnx, PGTVnd, PCTV1, PCTV2 for the first treatment were (5.94 ± 2.62)%, (0.96 ± 2.04)%, (3.79 ± 2.39)%, (2.26 ± 2.24)%, the seventh treatment were (2.75 ± 3.58)%, (-0.30 ± 3.29)%, (0.72 ± 4.07)%, (-0.47 ± 3.65)% and the thirty-third treatment were (5.51 ± 7.41)%, (2.09 ± 7.54)%, (3.64 ± 6.34)%, (4.89 ± 9.43)%, respectively. The means ∆Dmax of the PTVs were large for these three treatment stages, all greater than 3%. The similar results were also obtained for organ at risk. The standard deviation (SD) indicated a higher value of the thirty-third for PTVs and organ at risk compared to the first treatment.
Two patient data examples with the maximum difference in passing rate were compared, which the γ image distributions in the transverse view and γ histogram were shown in Fig. 8. Figure (a) was a brain metastasis patient with a γ passing rate of 99.84%, γmean 0.281, SD 0.188, while figure (b) was a rectal cancer patient with a γ passing rate of 91.70%, γmean 0.487, SD 0.336. As shown in the γ image, the green represented the passing region, while the pink region represented the failure to pass. We could see that there was some normal tissue outside the planning target volume of rectal cancer were unacceptable results, and γ distribution of rectal cancer was higher and wider than brain metastases from the histogram.
Table 3
DVH comparison between EDose and TPS for the 1st, 17th and the 33rd
Evaluated Organs
|
PGTVnx
|
PGTVnd
|
PCTV1
|
PCTV2
|
Brain stem
|
Spinal cord
|
Parotid gland L
|
Parotid gland R
|
1st
|
∆D98 [%]
|
0.27 ± 5.70
|
-3.72 ± 4.10
|
-2.24 ± 7.16
|
-2.92 ± 4.11
|
|
|
|
|
∆Dmean [%]
|
5.94 ± 2.62
|
0.96 ± 2.04
|
3.79 ± 2.39
|
2.26 ± 2.24
|
|
|
|
|
∆Dmax [%]
|
7.85 ± 3.05
|
4.78 ± 2.50
|
7.24 ± 2.54
|
4.92 ± 4.70
|
9.08 ± 4.82
|
5.52 ± 4.09
|
|
|
∆D50 [%]
|
|
|
|
|
|
|
2.87 ± 4.98
|
2.79 ± 4.16
|
17th
|
∆D98 [%]
|
-2.63 ± 5.40
|
-5.79 ± 4.36
|
-6.13 ± 8.83
|
-6.25 ± 2.86
|
|
|
|
|
∆Dmean [%]
|
2.75 ± 3.58
|
-0.30 ± 3.29
|
0.72 ± 4.07
|
-0.47 ± 3.65
|
|
|
|
|
∆Dmax [%]
|
5.01 ± 3.36
|
3.24 ± 3.23
|
4.40 ± 3.37
|
3.24 ± 3.53
|
5.02 ± 7.74
|
1.29 ± 5.31
|
|
|
∆D50 [%]
|
|
|
|
|
|
|
-0.84 ± 6.76
|
0.50 ± 9.95
|
33th
|
∆D98 [%]
|
2.70 ± 8.28
|
-2.63 ± 8.57
|
-0.97 ± 9.75
|
1.18 ± 12.18
|
|
|
|
|
∆Dmean [%]
|
5.51 ± 7.41
|
2.09 ± 7.54
|
3.64 ± 6.34
|
4.89 ± 9.43
|
|
|
|
|
∆Dmax [%]
|
7.79 ± 6.94
|
5.92 ± 7.86
|
6.87 ± 6.00
|
5.31 ± 12.88
|
8.69 ± 7.14
|
6.07 ± 8.91
|
|
|
∆D50 [%]
|
|
|
|
|
|
|
6.62 ± 6.86
|
3.21 ± 15.22
|
To observe the difference between the 3D online and pre-treatment dose verification, a cervical lymph node metastases plan was adopted as an example, Fig. 9 showed γ image distributions of the 3D online (Fig. 9(a)) and pre-treatment (Fig. 9(b)) dose distributions in isocentric plane of three axes, where the passing rate of 3D online and pre-treatment were 97.06% and 95.60%, γmean 3D online and pre-treatment were 0.38 and 0.41. The sky blue represented the passing region, while the yellow area represented the failed region. As shown in the pictures, the result distributions were similar between the 3D online and pre-treatment, and the failed areas were mainly distributed in regions of the low dose and areas close to the skin.