(Monday, 7/15/2019) 9:30 AM - 10:00 AM
Room: Exhibit Hall | Forum 6
Purpose: Live electronic portal imaging (EPID) based assessment of the inspiration level for Deep Inspiration Breath Hold (DIBH) treatment of breast cancer patients holds the potential for accurate treatment delivery without additional imaging dose and using existing imaging hardware. This work increases the robustness of our current lung depth (LD) based image analysis method and expands it to cases with little or no lung in the image. The distance from the skin to the field edge is investigated as a reliable breath hold indicator.
Methods: 643 existing EPID cine images from 13 patients were re-analyzed. In addition to the midfield line, traditionally used to determine LD, intensity profiles at the superior and inferior quartile of each image were extracted. Derivatives of the profiles were used to identify relevant landmarks. Distances from the posterior field edge to intersections of each of the lines with the chest wall and with the air-skin interface were calculated and analysed.
Results: The amplitude of the peak in the first derivative of the line intensity profile at the skin-air interface was generally about 3 times larger than that at the lung-chest wall interface. Skin to posterior field edge distance data could be extracted from all images for all three locations. LD was available for all images at midline, but at the superior and inferior lines only in 80% and 76% of images, respectively. Analysis along the midline showed that the mean standard deviation of the skin to posterior field edge distance for image from one fraction (~50 images) was lower (0.07 mm) than that for the LD (0.12 mm).
Conclusion: Distance of the skin-air interface to the posterior field edge is a more robust and universal measure than LD for breath hold assessment with MV cine EPID images.
Funding Support, Disclosures, and Conflict of Interest: This work has been supported by National Health and Medical Research Council (NHMRC) grant 1147533 of the Australian Government.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
