Rapidly converging 3D iterative deconvolution method for image improvement from different clinical modalities

Purpose: To demonstrate an application of rapidly converging, 3D iterative deconvolution method [1] with a novel resolution subsets-based (Richardson-Lucy like algorithm [2-3] with multiple resolution levels) approach RSEMD that operates on DICOM images to improve the image resolution and contrast. Materials and Methods: The RSEMD method was tested on phantoms, pre-clinical and clinical imaging data from different imaging scanners [4-8] (PET, CT, MRI, X-RAY, PEM, DBT, MBI). This method was applied to images previously reconstructed with conventional software to determine improvements in image resolution, SNR and CNR. The blurred clinical image is iterated against different resolution kernels to maximize SNR. Results: In the entire phantom, pre-clinical and clinical studies the improved images proved to have higher resolution, contrast and lower noise as compared with images reconstructed by conventional FBP, MLEM, EMSM, EMBD or OSEM software. Conclusions: The proposed RSEMD method can be applied to clinical images to further improve image quality in order to aid in the diagnosis of cancer at the earliest stages.