Nuclear medicine image registration by spatially noncoherent interferometry

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Optical Metrology

Background:

• Accurate image registration is crucial for longitudinal studies, particularly in brain SPECT imaging.
• Current methods often rely on SPECT slices, which can be less precise for coregistering subsequent acquisitions.
• Patient head contour data offers an alternative reference frame for improved registration accuracy.

Purpose of the Study:

• To introduce a novel technique for acquiring high-resolution patient head surface data.
• To enable coregistration of brain SPECT studies using the patient's facial contour.
• To integrate a miniature range finder (conoscope) with a gamma camera gantry.

Main Methods:

• Utilized conoscopic holography, an interferometry technique, to measure absolute distances to the skin surface.
• Mounted a conoscope on the gamma camera gantry, employing a scanning system for facial imaging.
• Calibrated the system to align conoscopic measurements and SPECT slices within the same coordinate frame.

Main Results:

• Achieved system calibration with a spatial error (SE) of less than 2 mm, smaller than SPECT pixel size.
• Biometric measurements of a brain phantom showed accuracy within 3%-5% of actual values.
• Successfully registered brain phantom and volunteer brain SPECT images using facial mask data.

Conclusions:

• The conoscope system provides real-time head surface data at low cost.
• This surface-based registration method is independent of the dataset of interest.
• Offers advantages in calibration robustness and sensor compactness compared to other optical methods.