The basal ganglia matching tools package for striatal uptake semi-quantification: description and validation

Abstract

Purpose

To design a novel algorithm (BasGan) for automatic segmentation of striatal ¹²³I-FP-CIT SPECT.

Methods

The BasGan algorithm is based on a high-definition, three-dimensional (3D) striatal template, derived from Talairach’s atlas. A blurred template, obtained by convolving the former with a 3D Gaussian kernel (FWHM = 10 mm), approximates striatal activity distribution. The algorithm performs translations and scale transformation on the bicommissural aligned image to set the striatal templates with standard size in an appropriate initial position. An optimization protocol automatically performs fine adjustments in the positioning of blurred templates to best match the radioactive counts, and locates an occipital ROI for background evaluation. Partial volume effect correction is included in the process of uptake computation of caudate, putamen and background. Experimental validation was carried out by means of six acquisitions of an anthropomorphic striatal phantom. The BasGan software was applied to a first set of patients with Parkinson’s disease (PD) versus patients affected by essential tremor.

Results

A highly significant correlation was achieved between true binding potential and measured ¹²³I activity from the phantom. ¹²³I-FP-CIT uptake was significantly lower in all basal ganglia in the PD group versus controls with both BasGan and a conventional ROI method used for comparison, but particularly with the former. Correlations with the motor UPDRS score were far more significant with the BasGan.

Conclusion

The novel BasGan algorithm automatically performs the 3D segmentation of striata. Because co-registered MRI is not needed, it can be used by all nuclear medicine departments, since it is freely available on the Web.

Appendix: PVE correction

Activity concentration is considered to be uniform within each compartment. We suppose that PVEs do not produce appreciable cross-talk effects between the left and the right side in activity evaluation and, therefore, that each side can be processed independently.

Relative to one side (e.g. right side), we denote by $\widetilde{C}$, $\widetilde{P}$ and B the unknown activity concentrations in the caudate nucleus, putamen and background compartments, respectively. Activity concentration B of the background can be easily evaluated by the ratio of all counts contained in the background ROI over the number of voxels of the ROI itself, since the background ROI is not appreciably affected by PVEs. The value of B is the same for both sides. The problem consists in finding the values of $\widetilde{C}$ and $\widetilde{P}$ relative to non-overlapping compartments on the basis of the information derived from a PVE-affected activity map, where the counts contained in a voxel may contain contributions from different compartments.

We denote by c _i , p _i and b _i (i = 1,..., N), respectively, the low-resolution templates of the overlapping compartments caudate nucleus, putamen and background as projected into the image space at the optimal positioning (indeed, index i explores all grabbed voxels in the input activity map, i.e. all voxels targeted by the projection). Moreover, we point out that the background compartment b _i (i = 1,..., N) must not be confused with the background ROI used to estimate B. After projecting the templates c _i and p _i into image space, the background compartment b _i is obtained as the complement to unity of the sum c _i +p _i over the range of all grabbed voxels. Thus the following relation holds:

$$ c_{i} + p_{i} + b_{i} = 1\;{\left( {i = 1, \ldots ,N} \right)} $$

(3)

which can be given the probabilistic interpretation that the grabbed voxel i may contain counts from caudate compartment with probability c _i , from putaminal compartment with probability p _i and from background with probability b _i . Thus, the PVE-affected data v _i (i.e. the grabbed voxels of the activity image) contain contributions from the three compartments according to the following relation:

$$\upsilon _{i} = \widetilde{{C\,}}c_{i} + \widetilde{P}\,p_{i} + B\,b_{i} .$$

(4)

Thanks to the hypothesis that the background receptors are uniformly spread all over the brain, including the striatum, one is interested in the activity concentrations C and P given by:

$$C = \widetilde{C} - B\,;\,P = \widetilde{P} - B.$$

(5)

Finally, the set of equations connecting the experimental data v _i with the unknowns C and P are:

$$\upsilon _{i} - B = C\,c_{i} + P\;p_{i} \;{\left( {i = 1, \ldots ,N} \right)},$$

(6)

whose solution can be estimated by solving the following least-squares problem:

$${\sum\limits_{i = 1}^N {{\left( {\upsilon _{i} - B - C\,c_{i} - P\;p_{i} } \right)}^{2} = {\text{minimum}}} }.$$

(7)

In this way one obtains PVE-corrected estimates for the activity concentrations C and P, from which the ratios C/B and P/B can be derived, respectively denoted as the caudate and putaminal values.