Three-dimensional (3-D) positron emission tomography (PET) imaging of inhaled radiolabeled aerosol in the nasal or pulmonary regions provides an in vivo measurement of drug distribution using the drug itself as the tracer. Repeated or dynamic PET scans over the time after inhalation provides us with further information about the fate of the deposited drug. These quantitative measurements are sufficient to describe the performance of a drug or device and they are obtained in a noninvasive fashion, which cannot be achieved by using any other methods. Using this PET-imaging paradigm, we conducted a sequence of drug studies to evaluate the performance of aerosol drugs and delivery devices; to compare the performance of similar drugs from different manufacturers; to assess the similarity between different formulations and propellants for the same drug; to appraise delivery devices such as spacers and nebulizers, etc. This paper reviews only the imaging and data analysis techniques developed for the above-mentioned studies that include multi-modality image registration, region definition and region-based data analysis, and nonregion-based data analysis. We separated the techniques into nasal and pulmonary studies because of the uniqueness of each group. Specific drugs or devices are not identified and no result about drug performance is given because the imaging and data analysis methodology, which is the focus of this paper, applies to all these studies regardless of the drugs or their delivery devices. The quantitative data are used as the scientific basis for evaluation although we also developed visualization techniques to enhance the results drawn from the data.

Image processing procedures are usually carried out in the spatial domain where the images are acquired and presented/utilized. The linear nature of the Fourier transform allows only those operations that are linear to be mapped into the frequency domain. In contrast, nonlinear operations and manipulations cannot be realized directly in the frequency domain. One of these nonlinear operations is thresholding. When operating in the spatial domain to segment image contents into object and background, thresholding is simple and efficient. However, it has no obvious representation in the frequency domain and cannot be carried out there in a straightforward fashion. In this paper, a means to relax the rigid linear limitation of the Fourier transform was investigated. A novel approach was established to achieve spatial thresholding using only frequency domain operations. The spatial grayscale or scalar data set (two-dimensional (2-D) image or three-dimensional (3-D) volume) was expanded into a binary volume in hyperspace having one more dimension than the original data set. The extended dimension is the gray level of the original data. Manipulating only on that dimension produces the effect of thresholding.

Combined rendering of volume data and geometric objects (graphics objects) is important in a number of medical applications. Efficient generation of such rendering is required to support user interactions at a level that are useful in clinical applications. In this study, we describe the manipulation of graphics objects with volume data utilizing a widely used graphics API standard, OpenGL.>