Evaluation of image quality and clinical acceptance of HD-Bone, a novel multimodal SPECT/CT reconstruction method for Tc99m diphosphonate imaging

Objectives A new true-hybrid multimodal SPECT reconstruction method called High-Definition Bone (Siemens HD-Bone) has been developed. The aim of this pilot study was to assess image quality and clinical acceptance of HD-Bone among experienced readers compared to Flash3D (F3D), which is in clinical use.

Methods All data were acquired on Symbia SPECT/CT systems in dual mode (clinical and NM listmode VA63A) setting under approved IRBs from 4 sites. SPECT images of the chest were reconstructed with F3D and the prototype of HD-Bone. Both methods included 3D resolution recovery, attenuation and scatter corrections. In addition, HD-Bone incorporated CT information to delineate tissue boundaries during image reconstruction. The F3D and HD-Bone images were presented to ten experienced readers in random order. Both SPECT scans were displayed together with the corresponding whole body planar and MIP (Maximum Intensity Projection) images. The readers were asked to rank six image features on a scale of 1-5: noise and its diagnostic importance, resolution and its diagnostic importance, diagnostic utility and clinical acceptance. Ratings were averaged over the ten readers.

Results A Bland-Altman analysis of the scores resulted in an average improvement for HD-Bone of 0.8 compared to Flash3D (ΔMedian(M)=0.75, p<0.0001, Wilcoxon). HD-Bone images were ranked equivalent to or higher than Flash3D images in >85% of the cases. The strongest improvement was observed with image resolution (ΔM=2.2, p=0.004(W)) and its diagnostic importance (ΔM=0.7, p=0.004(W)) followed by diagnostic utility and image acceptance.

Conclusions Visual evaluation by 10 readers showed that HD-Bone improved image resolution, noise, diagnostic utility and image acceptance in clinical practice.