Based on the characteristic of the PET-CT multimodal image series, a novel image registration and fusion method is proposed, in which the cubic spline interpolation method is applied to realize the interpolation of PET-CT image series, then registration is carried out by using mutual information algorithm and finally the improved principal component analysis method is used for the fusion of PET-CT multimodal images to enhance the visual effect of PET image, thus satisfied registration and fusion results are obtained. The cubic spline interpolation method is used for reconstruction to restore the missed information between image slices, which can compensate for the shortage of previous registration methods, improve the accuracy of the registration, and make the fused multimodal images more similar to the real image. Finally, the cubic spline interpolation method has been successfully applied in developing 3D-CRT (3D Conformal Radiation Therapy) system.
Algorithms ; Artificial Intelligence ; Diagnostic Imaging ; methods ; Humans ; Image Enhancement ; instrumentation ; methods ; Image Interpretation, Computer-Assisted ; instrumentation ; methods ; Imaging, Three-Dimensional ; instrumentation ; methods ; Pattern Recognition, Automated ; methods ; Positron-Emission Tomography ; instrumentation ; methods ; Radiotherapy Planning, Computer-Assisted ; instrumentation ; methods ; Tomography, X-Ray Computed ; instrumentation ; methods

In this paper, BP neural network is used to segment whole body bone SPECT image so that the lesion area can be recognized automatically. For the uncertain characteristics of SPECT images, it is hard to achieve good segmentation result if only the BP neural network is employed. Therefore, the segmentation process is divided into three steps: first, the optimal gray threshold segmentation method is employed for preprocessing, then BP neural network is used to roughly identify the lesions, and finally template match method and symmetry-removing program are adopted to delete the wrongly recognized areas.
Algorithms ; Bone and Bones ; diagnostic imaging ; Humans ; Image Interpretation, Computer-Assisted ; methods ; Neural Networks (Computer) ; Pattern Recognition, Automated ; methods ; Tomography, Emission-Computed, Single-Photon ; Whole Body Imaging