Font Size: a A A

Memory Optimization Oriented Back Projection In Cone Beam CT Algorithm Research And Hardware Realization

Posted on:2015-08-02Degree:MasterType:Thesis
Country:ChinaCandidate:Q LiFull Text:PDF
GTID:2308330479489920Subject:Microelectronics and Solid State Electronics
Abstract/Summary:PDF Full Text Request
Computed Tomography is an image reconstruction technique using the projection data received by X-rays which penetrate the object. The modern medical CT is mainly used in medical diagnostic field. The current most widely used algorithms are inexact. And Katsevich algorithm is an accurate CT reconstruction algorithm, which belongs to a filtered back projection algorithm in one. Because of this algorithm is accurate reconstruction, so the quality of reconstruction image using Katsevich algorithm is higher. Since the algorithm is very complex and computationally intensive, it is not yet widely used in the field of medical CT requiring fast data processing speed to generate a reconstructed image.With the image resolution and reconstruction speed requirements increase, limited FPGA resources, in particular its storage resources are no longer able to meet the requirements. This paper presents a storage optimization approach for the problem arises. After storage optimization, computed without doing it all before back-projection operator temporary data, an effective solution to the problem of back-projection hardware resource bottlenecks encountered in the implementation process.Firstly, cone beam CT reconstruction algorithms were introduced, focusing on the back projection step. Back-projection in cone beam CT image reconstruction is an important step and its execution time spent in the proportion of the entire reconstruction algorithm execution time is up to 50%. Accelerate the process of reducing the overall back-projection reconstruction of time plays a vital role. For a long time, due to the back-projection calculation involves more complex, hardware implementations have some difficulty. This article is divided reasonable back-projection circuit modules, detailing the structure of each module and signal connections between them. Computing core uses 32 floating-point IP core structures and pipelined structure. Execution process of circuit is smooth. The circuit has high computing efficiency and a significant acceleration effect for back-projection.
Keywords/Search Tags:CT, Katsevich algorithm, memory optimization, FPGA, back-projection
PDF Full Text Request
Related items