White paper

Improve Image Quality with Anatomically Guided Bayesian Reconstruction for Bone SPECT/CT

Hermia Advanced Bone SPECT Reconstruction

Bone SPECT/CT has a new clarity with Bayesian reconstruction in Hermia. Increase lesion contrast and decrease noise in your clinical bone SPECT/CT routine today with no additional hardware requirements; Hermia advanced reconstruction works with current generation scanners from all manufacturers.

With easy-to-use software enabling precise determination of reconstruction parameters, we are proud to offer several Bayesian reconstruction algorithms to optimise your bone SPECT/CT images, including two algorithms leveraging the information from the CT to guide the SPECT reconstruction, namely AMAP-Smooth and AMAP-RDP. AMAP-RDP showed superior performance to all other tested reconstruction algorithms in the study by Kangasmaa et. al. (1)

SPECT/CT image showing L5 vertebral metastasis from primary breast cancer. Conventional OSEM reconstruction and CT Guided RDP reconstruction (AMAP-R).

What is it?

  • Using prior information about how the final image is expected to look, the reconstruction algorithm is guided towards a clearer, less noisy solution
  • Anatomical prior information from the CT can be included to smooth only in uniform tissue areas and avoid smoothing over edges

What is the benefit?

  • Higher contrast lesions for bone SPECT/CT
  • Less noisy images
  • Better quantitative accuracy

What do the experts say?

In the study performed by the team at Vaasa and Lahti Central Hospitals, in Finland1, artificial lesions were added to the SPECT data to allow determination of absolute errors in SUV. They found average relative error in mean SUV for the artificial lesion study for OSEM, RDP, AMAP-S, and AMAP-R was − 53%, − 35%, −15%, and − 10%, when the CT study had matching lesions, see graph below.

The authors concluded, ‘‘The Bayesian methods with anatomical prior, especially the relative difference prior-based method (AMAP-R), outperformed OSEM and reconstruction without anatomical prior in terms of quantitative accuracy.”

1. Kangasmaa et al. EJNMMI Physics (2021) 8:2)