A good correlation (R2 = 0.76) was observed between the post-mortem TTC and the TPD developed by Slomka et al. . However, the study shows that the novel proposed method provided a higher AIF-TTC correlation (R2 = 0.94) with a linear fit passing through the axis origin. In addition, normal database of myocardial perfusion for mice is no longer required. As a result, AIF is independent of imperfections in the heart reorientation. The suppression of normal databases including different strains and ages in the infarct size assessment process should allow a more precise quantification of the infarct size changes with time. This is an important issue in the field of studies evaluating the impact of regenerative therapies such as stem cells in preclinical models of ischemic heart disease.
The novel method requires only the generation of a 20-segment polar map which is commonly proposed in commercial softwares. Afterwards, the dimensions of the segment contours are measured in arbitrary units on the long axis slices (Figure 1). This can be done by pasting the displayed segment contours into imaging softwares such as paint and PowerPoint, or by simply measuring the contours on a paper copy using a ruler. The segments’ relative level is then rescaled by an s-shaped function taking into account all the effects affecting the final spatial resolution and finally weighted with the segment volume before being summed (Equation 5).
The model optimisation was performed only on 11 mice all bearing LCA infarction geometry, which represents some limitation of the validation. However, the fact that the optimal s-shaped curve experimentally found was very close to the theoretical predictions (Figures 5 and 6), and the fact that the correlation between the AIF and the total infarcted fraction was excellent (Figure 8, R2 = 0.94) for a wide infarction geometries modelling (i.e. various background levels, various defect sizes, various defect positions versus the breathing and versus the polar map segmentation) strongly supports the validity of the method. Note that the major effects, i.e. system resolution, breathing and surrounding background, are independent of the infarction geometry. The curve pivoting induced by the system resolution and breathing corresponds to a reduction in sensitivity, i.e. the infarcted fraction range (0, 1) in a segment is mapped into the reduced segment relative level range (0.5, 0.8). When the resolution is too altered, a loss of information occurs, i.e. segments with different infarcted fraction can end up to the same relative level (dashed line in Figures 5,6,7). The use of sestamibi leads to an overestimation of infarction area since segments with low perfusion are not always totally necrotic. However, this additional effect altering the effective spatial resolution is accounted for by the fine tuning of the theoretical curve to the experimental data.
Other groups have demonstrated the capability to quantify an infarct size in rats [9–11] and in mice [5, 7]. In mice, Wu et al.  used a pinhole SPECT system with a spatial resolution of 1.1 mm and Tc-99m-sestamibi as perfusion tracer and the validation was performed against autoradiography. Stegger et al.  used a small animal-dedicated PET camera with an effective spatial resolution of 0.7 mm and F-18-FDG. More recently, Wollenweber et al.  used a clinical pinhole SPECT system with a spatial resolution of 1.9 mm and Tc-99m-sestamibi. Wu et al.  found a good correlation between infarct size (IS) measured with pinhole SPECT and autoradiography (r = 0.83; y = 0.85x + 2.8). However, they observed an underestimation of ISs relative to autoradiography for small infarcts sizes. This underestimation was attributed to partial volume effect and to physiological motion blurring (heart beats and breathing). Wollenweber et al.  observed a systematic underestimation of the infarct size by the pinhole SPECT compared to histology for infarct extension ranging from 15% to 45%). Our results displayed a better correlation (r = 0.97), similar to that obtained by Stegger et al.  using the state-of-the-art, high-resolution PET camera (r = 0.98). These better results are probably at least partially explained by the higher spatial resolution of the Linoview SPECT system (0.6 mm) compared to those of the pinhole SPECT system used by Wu et al. (1.1 mm) or Wollenweber et al. (1.9 mm). Moreover, the Bland-Altman analysis revealed no systematic bias in the wide range of values obtained (Figure 10).
In these previously published studies, perfusion defects were quantified by using specially created softwares based on the generation of circumferential profiles [5, 10] or none [8, 9]. The novel proposed method only requires a polar map which was obtained in the present study using QPS, a widely available software routinely used in daily clinical practice. This software has been used by Constantinesco et al.  who obtained with a multiple pinhole SPECT system an accurate quantification of left ventricular parameters (perfusion and volumes) in normal mice. Unfortunately, in this study however, left ventricular perfusion evaluation was performed only in normal mice.
In our study, the mean activity injected was 173 ± 27 MBq of Tc-99m-sestamibi. This activity is at least two times less than that reported in most studies [5–7]. Tc-99m sestamibi is cleared through the hepatobiliary system, and the gallbladder uptake appears as a focus of radioactivity near the heart in the projection images. This often shadows the apex of the heart in the reconstructed images. To avoid such problem, an intraperitoneal injection of 0.1 μg/kg cholecystokinin (CCK) was performed 2 h after injection of Tc-99m-sestamibi. CCK induces gallbladder contraction, dispersing the activity away from the heart into the small intestine . Tc-99m-tetrofosmin or better Tc-99m-N-DBODC5 has a more rapid liver clearance than Tc-99m-sestamibi [21, 22]. The use of those tracers instead of Tc-99m-sestamibi should potentially induce less activity in the gallbladder allowing to skip CCK administration. This issue needs to be further studied.