[¹⁸F]-flutemetamol across acquisition timeframes: clinical interpretability combining visual read and Centiloids

Quantification is now in the prescribing information of amyloid-PET tracers allowing its use to support visual read. As Centiloid (CL) scaling is increasingly used for decision-making, its robustness to scanning parameters is essential. We aim to examine the impact of timing on the interpretability of [¹⁸F]-flutemetamol PET scans, combining visual read and quantification.

We analysed two datasets: (1) real-world static scans from the IDEAS study (N = 499), and (2) 'dynamic' data reconstructed into pseudo-static frames in the same individual at varying start times (N = 89). Visual reads were performed by local and expert readers. Linear models were used to assess the impact of acquisition start time and duration on CL quantification. Measurement variability was compared to test-retest repeatability.

Visual interpretation was robust across start times, with no significant impact on the ability to classify scans as amyloid positive/negative. Some amyloid-negative scans acquired earlier (60-80 min start-time) exhibited lower grey-white matter contrast, which could reduce expert reader confidence. Quantitative analysis showed that CL values increased with later acquisition times, especially for amyloid-positive scans. In the intermediate CL range (10-30), differences between acquisitions starting at 60 vs 90 min were within expected test-retest variability (<6CL). Group-level adjustment factors can partially mitigate timing effects, but substantial individual variability remains.

With the presently available data, the individual-level effect of acquisition start time on CL cannot be reliably and accurately estimated from static imaging. Combined visual and quantitative assessment of [¹⁸F]-flutemetamol PET scans across the recommended timeframe supports reliable clinical interpretation. Consistency in acquisition parameters is recommended for repeat scans.