Comparing apples with oranges: on the difficulty of determining PET-comparable absolute myocardial blood flow using dynamic CT

Background/Introduction With the arrival of CT scanners able to scan the full heart in one heartbeat, myocardial perfusion imaging became feasible. However, when comparing PET- to CT-based myocardial blood flow (MBFct), it becomes apparent that the values are on a different scale with different units. MBFct is usually presented in ml/100ml/min, while PET uses ml/g/min. This complicates interpretation of MBFct when trying to apply a PET-based methodology. Purpose To identify and present methods to determine PET-comparable absolute MBF values using dynamic CT. Methods We explored two approaches to convert MBFct into PET-comparable MBF values. We applied these methods to a list of theoretical but realistic MBFct values ranging from 10-300 and to a clinical case. First, we applied a density-based factor of 105 gram of myocardial tissue per 100 ml blood to obtain MBFct values with the same unit. Secondly, as suggested by Ichihara, we considered (density corrected) MBFct values as K1, the inflow of contrast from capillaries to tissue. We converted K1 into MBF using a Renkin-Crone model by applying a correction based on three pre-established sets of alpha and beta values (Takafuji, Møller and Kikuchi). Converted MBF values were compared to an established PET-based stress MBF threshold of 2.0 ml/g/min to identify ischemia for clinical interpretation. Results The density correction with or without the Takafuji correction resulted in similar MBF values that seem rather low over the full range of initial MBFct values. The methods by Møller and Kikuchi provided similar corrected MBF around input values MBFct 100 ml/100ml/min. However, below and above this level corrected MBF values diverged from each other. For MBFct >150 ml/100ml/min, the Kikuchi correction led to unrealistic values (>6 ml/g/min). Results are summarized in Table 1. In a patient with a stenotic left anterior descending artery, stress MBFct was obtained using commercially available software for all 17 segments. Visually, segments 8, 13 and 14 demonstrated lower perfusion on a bullseye plot and were quantified with MBFct values of 78, 119 and 82 ml/100ml/min, respectively. MBFct values around 80 ml/100mL/min were corrected to 0.76, 0.77, 1.07 and 1.66 ml/g/min using the density and Renkin-Crone corrections. Despite the variation all these corrected MBF values are indicative of potential ischemia. However, MBFct of 119 ml/100mL/min was corrected to values ranging from 1.14 to 2.71 ml/g/min, possibly resulting in different clinical interpretations. Conclusion Simple correction of MBFct with a density factor (105 gr/100ml) does not seem sufficient in bringing it into the realm of absolute MBF as established by PET. The three suggested Renkin-Crone based corrections may lead to more realistic values but still show large differences in corrected MBF values. Further exploration is needed to establish a reliable approach to convert CT-based flow measurements into PET-comparable MBF.