Hybrid radiosity-SP3 equation based bioluminescence tomography reconstruction for turbid medium with low- and non-scattering regions

Abstract

To provide an ideal solution for a specific problem of gastric cancer detection in which low-scattering regions simultaneously existed with both the non- and high-scattering regions, a novel hybrid radiosity-SP3 equation based reconstruction algorithm for bioluminescence tomography was proposed in this paper. In the algorithm, the third-order simplified spherical harmonics approximation (SP3) was combined with the radiosity equation to describe the bioluminescent light propagation in tissues, which provided acceptable accuracy for the turbid medium with both low- and non-scattering regions. The performance of the algorithm was evaluated with digital mouse based simulations and a gastric cancer-bearing mouse based in situ experiment. Primary results demonstrated the feasibility and superiority of the proposed algorithm for the turbid medium with low- and non-scattering regions.

Optical tomography of turbid media containing diffusive and non-diffusive inclusions: An error modeling approach

“…To reduce this cost, several hybrid methods have been proposed to study the light propagation through biological tissues, combining the diffusion theory with other models. 51 For example, DA combined with radiosity, 11,52 the last one describing light propagation in low-scattering regions, has been tested on simulations, for geometries with localized inclusions. Firbank et al 19 also presented a study for layered geometry, simulating brain tissue, finding good agreement of theoretical approaches with MC simulations and experiments, concluding that the effect on clear layers must not be obvious when calculating fluence in objects.…”

“…In a similar way, a hybrid model combining SP N with the radiosity theory has been proposed and tested on numerical phantoms. 11,53 In contrast to the radiositydiffusion model, this one presents an improvement with respect to its application to a whole body due to its accuracy when there coexist regions with high-, low-, and non-scattering characteristics. A different approach is presented by Coelho et al, where a multi-scale method for solving the RTE is introduced, coupling the RTE with the DA.…”

“…The accuracy on these hybrid methods relies on a priori information that can be obtained from other imaging techniques or anatomical information. 11,10 As an alternative to the approaches summarized above, it is possible to make a correction for compensating the weakness of the model employed for the reconstruction, i.e., it is possible to compensate for the errors of using the DA instead of using the RTE in the low-or non-scattering regions. The approximation error method has been implemented for treating diffusive volumes with low-or non-scattering regions included, 35,47 and has been performed only on numerical phantoms.…”

Bayesian sparse‐based reconstruction in bioluminescence tomography improves localization accuracy and reduces computational time

“…However, reconstructing the source distribution is extremely challenging because of how ill‐posed and ill‐conditioned the inverse problem is. In the last decade, extensive efforts have been made to develop image reconstruction algorithms for BLT, and many algorithms have been proposed, including using high‐order light propagation models , designing an effective regularization term , integrating a priori information provided by structural/functional imaging modalities or using multispectral measurements .…”

Performance evaluation of the simplified spherical harmonics approximation for cone-beam X-ray luminescence computed tomography imaging

“…[19][20][21] Further, some hybrid models based on the SP N have also been proposed to exhibit the applicability in optical imaging. 22,23 However, SP N models do not converge to the exact RTE approximation when N ! 1.…”