Quantitative assessment of noise and standard uptake values by different iterative reconstruction

Standard Uptake Values (SUV) are semiquantitative measurements of radioactivity concentrations for tumors diagnosis and response monitoring. The aim of this study was to assess the effect on SUVs by varying different subsets and iterations in OSEM Iterative Reconstruction.

Image acquisitions were performed on PET-CT Scanner having integrated PET with BGO crystals and 16 slice CT. Images of different 20 tumors were acquired along with Torso phantom having refillable hot and cold spheres.

CT acquisitions were done at 80 mA, 140 kVp at 3.75 mm slice thickness.

PET Emission images were acquired for 3 minutes per AFOV using 2D Mode. The reconstruction of PET images was performed using Iterative reconstruction OSEM.

The iterations were varied as 1, 2, 3, 4, 5, 7, 10, 20, 30, 40 and 50 whereas subsets were varied as 1, 3, 5, 10, 15, 30, 42, and 70. On each constructed transaxial images 4 cm ROI was drawn at same location for all subsets and iteration. The SUVs normalized to body weight were quantified and analyzed. SUV were calculated by formula, SUV = mean pixel value (Bq/cc, decay corrected to start of scan) *(patient weight [kg]) * 1000/injected dose (Bq).

Noise was assessed as percentage Root Mean square % rms = (ó/m)*100

Noise increased with increasing number of subsets and iterations. This increasing was statistically significant (p<10⁻⁴).

At arbitrary fixed number (3) of iterations, the percentage variations in maximum SUVs between 1&3, 3&5, 5&10, 10&15, 15&30, 30&42, 42&70 subsets were 97.5%, 4.5%,1%, 2%, 1.6%, 1.2%, 2%, 6% respectively.

At arbitrary fixed number (30) of subsets, the percentage variations in maximum SUVs between 1&2, 2&3, 3&4, 4&5, 5&7, 7 &10, 10&20, 20&30, 30&40, 40&50 iterations were 0.33%, 1.0%, 2.1%, 0.65%, 0.32%, 1.6%, 3.1%, 1.3%, 0.76%, 0.3% respectively.

Our work demonstrates the additivity of subset and iteration in OSEM reconstruction over noise. By increasing the subsets the max. SUV first increases to peak value then decreases. This peak value depends on no. of iterations considered. But by increasing the iterations the max. SUV keeps on increasing.

The quality of images degrades rapidly as iteration increases but increases slowly as subsets are increased. Hence SUV must be assessed carefully based on the iterations and subsets used in OSEM reconstruction algorithm.