A new approach to quantifying lung damage after stereotactic body radiation therapy

Figures & data

Table I. Summary of differences between planning and follow-up scans.

	Planning scan	Diagnostic scan
Timing	Pre-radiotherapy	≥3 months post-radiotherapy
Breath control	Free breathing, with binning by respiratory phase (4-dimensional)	At inspiratory breath hold
Machine settings	140kVp and 100–110 mAs	120kVp and 100 mAs
Slice thickness	2.5 mm	2.5–5 mm
Intravenous contrast	None	Yes, unless contra-indicated
Positioning	Supine, arms above head on flat table	Supine, arms above head on rounded table

Figure 1. Examples of radiological pneumonitis scored as mild, moderate, or severe.

Figure 2. A. Representative example of using deformable registration to account for breathing. B. A unique situation illustrating the limits of deformable registration. The left-sided pleural effusion has resolved, but pneumonitis occurs in the same region. The deformed lung volume (outlined in white) does not include the region of pneumonitis.

Figure 3. A. Changes in lung Hounsfield unit density with breathing for four patients. Mean values are plotted for each phase of the breathing cycle. Standard deviations range from 0.001 to 0.04 (not shown). B. Lung density values as measured on the end-expiratory phase of the 4D-CT scan, after it was deformed to match each of the other phases of breathing, indicating that the HU density values did not change appreciably when the scans were deformed.

Table II. Accuracy of rigid registration vs. deformable registration.*

Structure	Rigid (mm)	Deformable (mm)	p-value
Apex right diaphragm	15.9 ± 2.6	4.0 ± 1.1	0.001
Apex left diaphragm	12.4 ± 2.9	3.5 ± 1.1	0.009
Sternal notch	10.5 ± 1.7	8.0 ± 3.1	0.37
Right brachiocephalic artery origin	8.9 ± 1.3	5.9 ± 1.2	0.047
T5 spinous process	8.7 ± 0.7	4.7 ± 0.9	0.001
Orifice of right upper lobe bronchus	7.9 ± 1.5	3.2 ± 0.4	0.011
Orifice of left lower lobe bronchus	7.4 ± 1.5	3.4 ± 0.6	0.032
Carina	6.6 ± 1.1	3.4 ± 0.4	0.023
Apex right lung	4.6 ± 1.3	3.1 ± 0.8	0.36
Apex left lung	3.9 ± 1.0	4.2 ± 2.1	0.88

*For ten patients, ten landmark structures were contoured on an end-inspiratory planning scan and transferred to a 3-month follow-up diagnostic scan using either rigid or deformable registration. The contoured landmark on the diagnostic scan was considered the ‘gold standard’, and mean (±SE) 3D displacement error of the centre of mass was calculated. Note that scans were done in different positions, and deformable registration was not fully optimized, to avoid overfitting differences between tumor and pneumonitis.

Figure 4. Changes in ipsilateral whole lung density (left) for all 18 patients (left) did not correlate with degree of pneumonitis (Spearman's r = 0.30; p = 0.22) whereas changes in lung density in the peri-tumoral region (right) showed strong correlation with radiological pneumonitis (Spearman's r = 0.75; p < 0.001). Mean values for each category are denoted by open circles, and error bars represent one standard error (SE). HU: Hounsfield units.

Figure 5. Pre- and post-treatment scans, showing radiological pneumonitis near the original tumor. Left: Deformed end-inspiratory phase of pre-treatment 4D-CT scan, with a right-sided tumor evident. The ipsilateral lung (yellow) and 3-cm expansion around the tumor (orange) are contoured. Right: Follow-up scan three months later, showing good resolution of the tumor, with a new area of radiological pneumonitis in the peri-tumoral region.