Fig. 10.

Significance testing of HbO₂^MSOT change due to gas challenge in RENCA-luc tumor-bearing mouse. Left: 4-pixel ROIs used. ROIs 1 and 2 are significant via the pixel-wise analysis for all filters, while ROI 3 is not significant for the SW filter (Table 1, Supplementary Fig. S2). Center: log₁₀(p) values over time for each filter in each ROI. Right: Binary classification of change as significant (s.) or not (n.s.), as well as logistic curve fit and indicated switching time. The SW filter is conspicuously noisy throughout time for all ROIs, but particularly so in ROI 3, causing an unreasonable fit of the switching time. The $\alpha$ filter denoises effectively and has a very sharp transition from non-significant to significant, but incurs a 0.5 min delay in the switching time. The $\alpha \beta$ filter denoises effectively while preserving the dynamic transition, giving a consistent switching time for all ROIs (see Table 1).