*2.3. Protein Binding of Radioactivity*

The plasma protein binding results are shown in Figure 4. Overall, protein binding seemed to vary relatively little over the time of scanning (the late measurement in pig no. 10 deviates, but may also have the largest uncertainty due to the low level of radioactivity remaining at the late time point). A constant fraction of protein binding reflects an equilibrium between the protein-bound and non-bound tracer, allowing the plasma to be treated as a single compartment. Partly for this reason, and partly because we lack protein data for the rest of the pigs, we did not include protein binding in further analysis.

**Figure 4.** Plasma protein binding of [68Ga]Ga-DOTA-Siglec-9 in the pigs. Percentage of binding as a function of minutes after tracer injection. For pig no. 10, the data are a summary of previously published results [22].

#### *2.4. Analysis of the Parent Tracer Fraction*

The parent tracer fraction (fraction of radioactivity originating from [68Ga]Ga-DOTA-Siglec-9) rapidly decreased during the investigated time. A sample fraction curve for the parent tracer is shown in Figure 5. All fraction curves are shown in Supplementary Figure S1.

**Figure 5.** Fraction of radioactivity from intact [68Ga]Ga-DOTA-Siglec-9 (the parent tracer fraction) in arterial plasma as a function of time. Data points and curve fits (Equation (2)) are from pig no. 9 as a representative example.

#### *2.5. VOIs and Model Fits*

In the 9 scans, a total of 24 VOI pairs were drawn: 10 VOI pairs were drawn in bone and 14 VOI pairs were drawn in soft tissue positions (counting the VOIs in pig no. 26 twice, as this pig was scanned twice). In pig no. 24, no infectious lesions were identified. In several pigs, bone lesions in the small pedal bones had to be excluded from VOI drawing and analysis, as the resolution of the PET scans would result in a partial volume effect that would be too high for the results to be robust. The studied lesions are summarized in Tables 2 and 3. The individual VOI volumes (cm3) can be found in Supplementary Table S1.


**Table 2.** Investigated osteomyelitis lesions, right hind limb.



\* Pig no. 26 was scanned twice; between the two scans, 5 mg of unlabelled ("cold") DOTA-Siglec-9 peptide was injected into the animal. † See also Figure 3A,B.

For each VOI, the PET data were fitted with the models shown in Figure 6. Using pig no. 9 as an example, the model fits based on uncorrected and corrected input function are shown in Figures 7 and 8, respectively. All model fits are shown in Supplementary Figures S2–S9.

**Figure 7.** Model fits for pig no. 9 with the *uncorrected* input function. Infection data are from osteomyelitis lesions in the distal femur and proximal tibia and from a soft tissue infection at the calcaneus, all in the animal's right hind limb. Control data are from anatomically corresponding positions in the noninfected left hind limb. All data were modelled with each of the three kinetic models shown in Figure 6.

**Figure 8.** Model fits for pig no. 9 with the *corrected* input function. Same PET data as in Figure 7, different input function. The plots are very similar to those in Figure 7, but not identical; visual differences from Figure 7 are most notable for the early part of 1TCM.

Many of the Patlak plots showed signs of non-linearity and/or horizontal fits, indicating that uptake was reversible rather than irreversible. The Patlak plots for pig no. 9 are shown in Figure 9, and all Patlak plots are shown in Supplementary Figures S2–S9.

Plotting the fitted parameters from the corrected vs. uncorrected input function (plots not shown) revealed that the main difference was a highly elevated ratio *k*3/*k*<sup>4</sup> if the corrected input function was used. A high value of *k*3/*k*<sup>4</sup> indicates that the tracer leaves the second compartment only slowly, i.e., a nearly irreversible uptake. Accordingly, the Patlak plots based on the corrected input function were closer to linearity than the Patlak plots based on the uncorrected input function (cf. Figure 9 and Supplementary Figures S2–S9).

**Figure 9.** Patlak plots for pig no. 9. **Left**: Based on the uncorrected input function (data as in Figure 7). **Right**: Based on the corrected input function (data as in Figure 8). After the early points, irreversible uptake will be characterized by a linear Patlak plot with a positive slope, while nonlinearity or horizontal slope are signs of reversible uptake.
