Use of relative Patlak plot Ki' images as an alternative to standard Patlak plot Ki images in clinical practice.

Background: The kinetic rate constant (Ki), derived from the Patlak slope, reflects F-FDG uptake and supports disease assessment. Standard Patlak Ki imaging requires prolonged dynamic acquisition and full arterial input function (IF), limiting clinical feasibility. The relative Patlak plot omits the early-phase IF and focuses on the linear phase of tracer kinetics using only late dynamic data and a partially sampled IF. This approach enables simplified parametric imaging without requiring full early-time data or population-based input functions (PBIFs). This study evaluated whether relative Ki′-images can substitute for standard Ki-images by assessing both quantitative agreement and visual appearance using a short axial field-of-view PET system.

Results: We analyzed data from 44 patients with lung or pancreatic disorders who underwent dynamic whole-body F-FDG PET/CT, followed by static SUV imaging. Standard Ki-images were generated using Patlak analysis from 7 to 60 min post-injection, based on full-time image-derived IF (0–60 min), with a scan duration of 53 min. Relative Ki′-images were reconstructed using three time schedules: (1) Ki′-49 min (t*=7 min post-injection; duration: 49 min), (2) Ki′-41 min (t*=15 min; duration: 41 min), and (3) Ki′-33 min (t*=24 min; duration: 33 min), each using partial IF (excluding data from injection to t*). Lesion conspicuity (score 0–3), SUV- and Ki/Ki′-parameters (SUVmax/mean, Ki-max/mean, Ki′-max/mean), and background noise were evaluated qualitatively (visual scale: 0–3) and quantitatively (standard deviation [SD] of liver Ki/Ki′) by two readers. Wilcoxon signed-rank test, Friedman’s test, McNemar’s test, and linear regression analyses were performed. Two readers assessed 35 positive and 9 negative scans (scores 1–3/0) for SUV, standard Ki, and relative Ki′-images. Correlations between standard Ki and Ki′ parameters were consistently strong (r²>0.97 for Ki-max and Ki-mean), with Ki′-49 min images showing the highest agreement (r²=0.98–0.99). Lesion conspicuity was comparable between standard Ki and Ki′-49 min/Ki′-41 min images, whereas Ki′-33 min images showed more downgrades ( < 0.05). Ki′-33 min also exhibited significantly higher background noise ( < 0.001), confirmed by liver SD.

Conclusions: Relative Ki′-images showed excellent correlation with standard Ki-images in quantitative analysis, indicating feasibility for reduced-scan protocols. Ki′-49 min images retained accuracy but required longer scans. Future work should assess alternatives such as PBIF or dual-window protocols for clinical streamlining.

Supplementary Information: The online version contains supplementary material available at 10.1186/s13550-025-01295-7.