The Design and Characterization of a Multiple Pinhole Collimator and Its Applications in Biopharmaceutics

Abstract

A 4-pinhole collimator was designed to image small objects and its characteristics were studied. The field of view formed by this collimator was a conically-tipped pyramid with a minimum object distance of 3 cm. The line-spread function at 8 cm from the collimator had a full-width-half-maximum of 8.5 mm and the depth resolution was 2 cm at 7.5 cm from the collimator. The geometric distortion of images caused by multi-pinhole collimation was also investigated. It was concluded that tablet disintegration can only be discerned from visual images under favorable conditions such as short source-to-collimator distances and parallelism of the planes of the object and the collimator.

Software was written to perform image reconstruction from the projected image. Only one plane was to be reconstructed as a representation of the object. The object distance was calculated at the same time. The enhancement techniques of smoothing, median filtering and their combinations were studied and it was found that the method of smoothing the reconstructed image followed by median filtering produced the best visual image and elevated the signal total counts and signal-to-noise ratio.

The designed collimator was used to image the disintgration of tablets in an artifical disintegrating medium which was a mixture of ethylene glycol and water. The disintegration state of a tablet was correlated to the signal total counts obtained from the series of images taken over time and were reconstructed and enhanced by smoothing and median filtering. The disintegration process was found to be monoexponential, and a disintegration constant and half-life were defined. The tablet was found to have a half-life of 45 minutes in the stationary disintegrating medium. The half-life decreased when the medium was allowed to flow around the tablet. The natural log of the disintegration constant and the linear flow rate were found to have a direct proportionality relationship. Experiments were also performed to test the efficacy of the correction for the loss of signal total counts due to distance using the calculated distance from the reconstruction algorithm. It was found that the correction worked adequately for distance variations within a range of about 8 cm. Overcorrection is likely for greater distance ranges.