TY - JOUR

T1 - Validation of kinetic modeling of [15O]H2O PET using an image derived input function on hybrid PET/MRI

AU - Vestergaard, Mark B.

AU - Calvo, Oriol P.

AU - Hansen, Adam E.

AU - Rosenbaum, Sverre

AU - Larsson, Henrik B.W.

AU - Henriksen, Otto M.

AU - Law, Ian

N1 - Publisher Copyright: © 2021

PY - 2021

Y1 - 2021

N2 - In present study we aimed to validate the use of image-derived input functions (IDIF) in the kinetic modeling of cerebral blood flow (CBF) measured by [15O]H2O PET by comparing with the accepted reference standard arterial input function (AIF). Additional comparisons were made to mean cohort AIF and CBF values acquired by methodologically independent phase-contrast mapping (PCM) MRI. Using hybrid PET/MRI an IDIF was generated by measuring the radiotracer concentration in the internal carotid arteries and correcting for partial volume effects using the intravascular volume measured from MRI-angiograms. Seven patients with carotid steno-occlusive disease and twelve healthy controls were examined at rest, after administration of acetazolamide, and, in the control group, during hyperventilation. Agreement between the techniques was examined by linear regression and Bland-Altman analysis. Global CBF values modeled using IDIF correlated with values from AIF across perfusion states in both patients (p<10−6, R2=0.82, 95% limits of agreement (LoA)=[-11.3–9.9] ml/100 g/min) and controls (p<10−6, R2=0.87, 95% LoA=[-17.1–13.7] ml/100 g/min). The reproducibility of gCBF using IDIF was identical to AIF (15.8%). Values from IDIF and AIF had equally good correlation to measurements by PCM MRI, R2=0.86 and R2=0.84, (p<10−6), respectively. Mean cohort AIF performed substantially worse than individual IDIFs (p<10−6, R2=0.63, LoA=[-12.8–25.3] ml/100 g/min). In the patient group, use of IDIF provided similar reactivity maps compared to AIF. In conclusion, global CBF values modeled using IDIF correlated with values modeled by AIF and similar perfusion deficits could be established in a patient group.

AB - In present study we aimed to validate the use of image-derived input functions (IDIF) in the kinetic modeling of cerebral blood flow (CBF) measured by [15O]H2O PET by comparing with the accepted reference standard arterial input function (AIF). Additional comparisons were made to mean cohort AIF and CBF values acquired by methodologically independent phase-contrast mapping (PCM) MRI. Using hybrid PET/MRI an IDIF was generated by measuring the radiotracer concentration in the internal carotid arteries and correcting for partial volume effects using the intravascular volume measured from MRI-angiograms. Seven patients with carotid steno-occlusive disease and twelve healthy controls were examined at rest, after administration of acetazolamide, and, in the control group, during hyperventilation. Agreement between the techniques was examined by linear regression and Bland-Altman analysis. Global CBF values modeled using IDIF correlated with values from AIF across perfusion states in both patients (p<10−6, R2=0.82, 95% limits of agreement (LoA)=[-11.3–9.9] ml/100 g/min) and controls (p<10−6, R2=0.87, 95% LoA=[-17.1–13.7] ml/100 g/min). The reproducibility of gCBF using IDIF was identical to AIF (15.8%). Values from IDIF and AIF had equally good correlation to measurements by PCM MRI, R2=0.86 and R2=0.84, (p<10−6), respectively. Mean cohort AIF performed substantially worse than individual IDIFs (p<10−6, R2=0.63, LoA=[-12.8–25.3] ml/100 g/min). In the patient group, use of IDIF provided similar reactivity maps compared to AIF. In conclusion, global CBF values modeled using IDIF correlated with values modeled by AIF and similar perfusion deficits could be established in a patient group.

KW - Cerebral blood flow

KW - Image derived input function

KW - Perfusion

KW - PET/MRI

KW - [O]HO PET

U2 - 10.1016/j.neuroimage.2021.117950

DO - 10.1016/j.neuroimage.2021.117950

M3 - Journal article

C2 - 33716159

AN - SCOPUS:85102854242

VL - 233

JO - NeuroImage

JF - NeuroImage

SN - 1053-8119

M1 - 117950

ER -