Relationship between NIR laser power and the human forehead tissue backscattering image features

Abstract

A novel markerless near infrared (NIR) laser-based head tracking system was recently proposed to resolve patient's head motion problem during cranial radiotherapy. Although previous research showed that we could track the patient's head position with the sub-millimetre range accuracy, the tracking performance strongly relied on the accuracy of the tissue thickness estimation. We noticed the factors that inuence the ROI features extracted from the backscattering images were the NIR laser power fluctuation and inconsistency. Therefore, the propose of this paper was to investigate the relationship between these parameters and determine a laser power independent feature transformation. We set up our head tracking system to project the pulsed NIR laser beam onto a single point on subject's forehead and observed the changes of the 5-ROI feature values on the different laser power level. The scatter plots between each ROI feature values and the laser power showed distinctive straight lines with similar slopes while applying linear regression to each scatter plot indicated that the slope of each ROI feature was also in the same range. According to the results, we could transform the data by subtracting the feature value of each ROI from their average slope value and the laser power. This new feature is laser power noise tolerance and could be used to enhance the tissue thickness estimation accuracy.