Error analysis for filtered back projection reconstructions in Besov spaces

Abstract

Filtered back projection (FBP) methods are the most widely used reconstruction algorithms in computerized tomography (CT). The ill-posedness of this inverse problem allows only an approximate reconstruction for given noisy data. Studying the resulting reconstruction error has been a most active field of research in the 1990s and has recently been revived in terms of optimal filter design and estimating the FBP approximation errors in general Sobolev spaces. However, the choice of Sobolev spaces is suboptimal for characterizing typical CT reconstructions. A widely used model are sums of characteristic functions, which are better modelled in terms of Besov spaces B q α , p ( R 2 ) . In particular B 1 α , 1 ( R 2 ) with α ≈ 1 is a preferred model in image analysis for describing natural images. In case of noisy Radon data the total FBP reconstruction error ‖ f − f L δ ‖ ⩽ ‖ f − f L ‖ + ‖ f L − f L δ ‖ splits into an approximation error and a data error, where L serves as regularization parameter. In this paper, we study the approximation error of FBP reconstructions for target functions f ∈ L 1 ( R 2 ) ∩ B q α , p ( R 2 ) with positive α ∉ N and 1 ⩽ p, q ⩽ ∞. We prove that the L p -norm of the inherent FBP approximation error f − f L can be bounded above by ‖ f − f L ‖ L p ( R 2 ) ⩽ c α , q , W L − α | f | B q α , p ( R 2 ) under suitable assumptions on the utilized low-pass filter’s window function W. This then extends by classical methods to estimates for the total reconstruction error.

Research on Data Optimization of CT System for Patients with Severe Pneumonia Based on Filtered Back Projection and Internet of Things

“…The access network is a fusion of the communication network and the Internet composed of various wired and wireless nodes and fixed and mobile gateways and can be accessed through the Internet, the broadcasting network, and the communication network. Different from classic information systems, IoT requires large-scale data access and transmission capabilities [31][32][33]. The role of the middleware layer is to shield heterogeneity, achieve two-way interaction, and data preprocessing.…”

Research on Biomechanical Simulation and Simulation of Badminton Splitting and Hanging Action Based on Edge Computing

“…At present, sports biomechanics diagnosis technology is integrated with multiple disciplines, such as the development of wearable devices, visual analysis, and smart monitoring based on big data, which can achieve precise and personalized diagnosis for different competition types and different sports people [26].…”

The Modulo Radon Transform: Theory, Algorithms and Applications

“…In this paper, we discuss the resulting inverse problem and show that accurate reconstruction of the underlying tomogram is feasible under the assumption that the output of the Radon transform is band-limited or low-pass filtered. This finite bandwidth assumption is O g m j S I S S x 5 B q F q P P M d h x f R x 7 a O Z Q p W e V 8 4 d U u e O q / S 8 q J 9 F x 1 / M p T i U T Y M 5 c f b k q s o F I Q l K l i Y + A 1 O h i r H W M I E v B L Q g 0 8 e G Z m z Z 0 E 3 d a 2 v T x m e O y K C 3 I + N J j W u b E K u K v L E m 4 h t j m M 5 y w W H M 8 W R J P G O 4 T T 8 H M 5 w 1 K g 4 o 6 n f 2 N z c 1 N s o / e A F 3 i h c / J 5 Q U 1 N f P / t D 5 N I K U v S N 0 c 9 X u L I v K i u s R f L + C v G / x w A T 9 s 8 N 0 F f L f B x w V e H e h S b l z 9 P m p X q z / X 9 6 T f a L v g D q Z l n t l p m D 0 s I B x 0 l 9 p r q G d 4 y n M 2 z x p i X y V X B O a Z 7 F 8 R Z a 7 m u T a E c S k 6 R E r G D c 5 l 0 t 2 Q I O 1 2 d l X X v 1 + q T c N z i M B k J b S c F u R 5 x 2…”

“…and, thus, existing error estimates [2,1,3,5] for the FBP approximation error f −f Ω carry over to the US-FBP error f − f λ Ω . For illustration, we apply [2, Theorem 5.5] and [1, Theorem 3] to obtain error estimates in Sobolev spaces of fractional order α > 0, given by…”