PURPOSE: Breathing management can reduce breath-to-breath (intrafraction) and
day-by-day (interfraction) variability in breathing motion while utilizing the
respiratory motion of internal and external surrogates for respiratory guidance.
Audiovisual (AV) biofeedback, an interactive personalized breathing motion
management system, has been developed to improve reproducibility of intra- and
interfraction breathing motion. However, the assumption of the correlation of
respiratory motion between surrogates and tumors is not always verified during
medical imaging and radiation treatment. Therefore, the aim of the study was to
test the hypothesis that the correlation of respiratory motion between surrogates
and tumors is the same under free breathing without guidance (FB) and with AV
biofeedback guidance for voluntary motion management.
METHODS: For 13 lung cancer patients receiving radiotherapy, 2D coronal and
sagittal cine-MR images were acquired across two MRI sessions (pre- and
mid-treatment) with two breathing conditions: (a) FB and (b) AV biofeedback,
totaling 88 patient measurements. Simultaneously, the external respiratory motion
of the abdomen was measured. The internal respiratory motion of the diaphragm and
lung tumor was retrospectively measured from 2D coronal and sagittal cine-MR
images. The correlation of respiratory motion between surrogates and tumors was
calculated using Pearson's correlation coefficient for: (a) abdomen to tumor
(abdomen-tumor) and (b) diaphragm to tumor (diaphragm-tumor). The correlations
were compared between FB and AV biofeedback using several metrics: abdomen-tumor
and diaphragm-tumor correlations with/without ≥5 mm tumor motion range and
with/without adjusting for phase shifts between the signals.
RESULTS: Compared to FB, AV biofeedback improved abdomen-tumor correlation by 11%
(p = 0.12) from 0.53 to 0.59 and diaphragm-tumor correlation by 13% (p = 0.02)
from 0.55 to 0.62. Compared to FB, AV biofeedback improved abdomen-tumor
correlation by 17% (p = 0.01) and diaphragm-tumor correlation by 15% (p < 0.01)
while correcting 0.3 s (p = 0.54) and 0.2 s (p = 0.19) phase shifts,
respectively. In addition, AV biofeedback with ≥5 mm tumor motion range, compared
to FB improved abdomen-tumor correlation by 14% (p = 0.18) and diaphragm-tumor
correlation by 17% (p = 0.01). The highest abdomen-tumor and diaphragm-tumor
correlations were found using ≥5 mm tumor motion range and phase shifts,
resulting in a 12% improvement in AV biofeedback.
CONCLUSIONS: Our results demonstrated that AV biofeedback improves the
correlation of respiratory motion between surrogates and the tumor. This suggests
a need for AV biofeedback for respiratory guidance utilizing respiratory
surrogates during image-guided and MRI-guided radiotherapy in thoracic regions.