Modelling the experimental and fundamental signal-to-noise in speckle contrast optical tomography

Anatomical Modeling and Optimization of Speckle Contrast Optical Tomography

“…In contrast, the previous results using an analytic model and a homogeneous medium do not vary for a single source-detector distance. 50,51,53 The normalized autocorrelation, 𝑔 1 (𝑟 𝑑 , 𝑟 𝑠 , 𝜏) , varies with each source-detector pair but was visualized (Fig. 2a) for representative measurements at each nearest-neighbor distance (first through sixth nearest neighbors), with the curves denoted as 𝑔 1 (𝑟, 𝜏) with source-detector distance 𝑟 = |𝑟 𝑠 − 𝑟 𝑑 |).…”

“…In contrast, the previous results using an analytic model and a homogeneous medium do not vary for a single source-detector distance. 50,51,53…”

“…To demonstrate the importance of noise in the overall performance of SCOT imaging, we incorporated system-specific noise models 50 and analyzed measurement SNR and imaging SNR (Fig. 4).…”

“…Existing models for SCOS and SCOT represent speckle contrast in terms of optical properties and flow within a medium and simulated measurements incorporate relevant noise sources. 8,50 Forward analytical models for SCOT measurements - with limited analytic, closed-form models 7,8,10 such as using a semi-infinite approximation for solving the diffusion equation - are well established. However, the analytic solutions cannot accommodate the complexity of real-world head anatomy.…”

“…We expect inverse solutions that include the noise variance 55 might improve the imaging SNR while allowing the improved localization associated with longer measurements. While our noise models were rigorously developed, 50 the results are specific to instrument details, and additional work remains to validate these models.…”