Audiovisual (AV) biofeedback improves anatomical position management in breath-hold (b) Taeho Kim, Sean Pollock, Danny Lee and Paul Keall Radiation Physics Laboratory, Sydney Medical School,The University of Sydney 1
Rationale: Motion Artifacts in 4D CT with irregular motion [1]Yamamoto et al., Int. J. Radiation Oncology Biol. Phys. 72, 1250 (2008) 2
Variation in Anatomical Position without Guidance Breath-hold technique for respiratory motion-compensation: Reducing respiratory motion artifacts during imaging, conventional treatments and stereotactic treatments. Selected patients with the breath-hold: However, anatomical position is variable during breath-hold practice without guidance. [2]George et al., Int. J. Radiation Oncology Biol. Phys. 65, 924 (2006) 3
Audiovisual (AV) Biofeedback Audiovisual biofeedback by Venkat et al. [3] : Guide patients for regular breathing. Respiratory motion control in radiotherapy and in imaging such as CT/PET imaging [4]. [3]Venkat et al., Phys. Med. Biol. 53, N197 (2008). [4]Yang, et al., MP 39, 1046 (2010). 4
AV Biofeedback combined with breath-hold Hypothesis: AV biofeedback improves anatomical position management in breath-hold. AV biofeedback study: 14 healthy human subjects in 24 breath-hold MRI studies: no prior history of lung disease. 10 subjects: two MRI imaging sessions & 4 subjects: one MRI imaging session. MR images acquired with free breathing (FB) and AV biofeedback conditions. MRI and Respiratory motion monitoring systems: 3T whole-body GE MRI with cardiac array coil. Real-time respiratory motion monitoring systems: MR bellows belt and Real-time position management (RPM). 5
Method: AV Biofeedback Study Setup (a) AV biofeedback setup in MRI RPM: Real-time Position Management (Varian) (b) Visual display with a waveguide 6
Method: AV Biofeedback Study Setup (a) AV biofeedback setup in MRI RPM: Real-time Position Management (Varian) (b) Visual display with a waveguide 7
Breath-hold Target Position In FB, the breath-hold target position was determined by the radiographer: Based on previous respiratory signal obtained from the MR bellows belt. Respiratory signal self-scaled by the system. (b) RPM_AV biofeedback AV biofeedback with breath-hold target position: Based on subject s average respiratory signal from the RPM system (10 respiratory cycles). 3 breath-hold target positions indicated using the red line. Non-scaled respiratory signal. 8
Result: Anatomical position management in breath-hold MRI (a) (b) (c) Abdominal breath-hold position variability: At the inhalation position (a), 142% ± 68% with FB 87% ± 15% with AV (p-value < 0.0001). At the 50% inhalation position (b), 80% ± 48% with FB 48% ± 7% with AV (p-value < 0.0001). At the exhalation position (c), 26% ± 47% with FB 8% ± 8% with AV (p-value = 0.014). 9
Reduction in Image Variation with AV Biofeedback Two sequential breath-hold images at inhalation. Image difference with scale bars shown. 10
Improving anatomical position management in MRI Sequential MR images at inhalation: 19 of 24 studies showed higher correlation than R=0.95 using AV biofeedback. 11 of 24 studies with free breathing. At 50% inhalation, 18 of 21 studies using AV biofeedback. 16 of 21 studies with free breathing. At exhalation, 16 of 21 studies with free breathing and with AV biofeedback. Suggesting considerable anatomical position reproducibility with both breathing methods. 11
Conclusion The study demonstrated the significant improvement of anatomical position reproducibility using AV biofeedback. At exhalation and 50% inhalation, considerable anatomical position reproducibility with both breathing methods: However, the achieved abdominal positions were not close to the planned position with free breathing. This system provides clinically applicable abdominal position management in breathhold medical imaging to increase image consistency and reduce motion artifacts. 12
Acknowledgements Dr. Ricky O'Brien Dr. Jamie Sherman Dr. Elaine Ryan University of Sydney Mike Graf (GE Healthcare) Brain and Mind Research Institute (BMRI) This work was supported by Sydney Medical School New Staff/Early Career Researcher Scheme grant, NIH/NCI R01CA93626 and an NHMRC Australia Fellowship. 13