Partial volume correction of the microPET blood input function using ensemble learning independent component analysis.

Medical images usually suffer from a partial volume effect (PVE), which may degrade the accuracy of any quantitative information extracted from the images. Our aim was to recreate accurate radioactivity concentration and time-activity curves (TACs) by microPET R4 quantification using ensemble learning independent component analysis (EL-ICA). We designed a digital cardiac phantom for this simulation and in order to evaluate the ability of EL-ICA to correct the PVE, the simulated images were convoluted using a Gaussian function (FWHM = 1-4 mm). The robustness of the proposed method towards noise was investigated by adding statistical noise (SNR = 2-16). During further evaluation, another set of cardiac phantoms were generated from the reconstructed images, and Poisson noise at different levels was added to the sinogram. In real experiments, four rat microPET images and a number of arterial blood samples were obtained; these were used to estimate the metabolic rate of FDG (MR(FDG)). Input functions estimated using the FastICA method were used for comparison. The results showed that EL-ICA could correct PVE in both the simulated and real cases. After correcting for the PVE, the errors for MR(FDG), when estimated by the EL-ICA method, were smaller than those when TACs were directly derived from the PET images and when the FastICA approach was used.

Quantification method in [18F]fluorodeoxyglucose brain positron emission tomography using independent component analysis.

OBJECTIVE: To extract accurate image-derived input functions from dynamic brain positron emission tomography images (DBPIs) using independent component analysis (ICA). METHODS: A modified linear model with haematocrit correction was used to improve the accuracy of input functions estimated by independent component analysis and to reduce the error of quantitative analysis. Two types of material were examined: (1) a simulated dynamic phantom with a three-compartment, four-parameter model; (2) clinical 2-h DBPIs with a standard plasma sampling procedure. The input function was extracted from DBPIs using independent component analysis. The modified linear model with haematocrit correction was used to obtain the independent component analysis-estimated input function (Iica). For comparison, the input function derived from the last three blood samples (Iest) was used. The image-derived input functions (Iica and Iest) were compared with the input function from blood sampling (Itp). The mean percentage error of the metabolic rate of [F]-2-fluoro-2-deoxy-D-glucose (MRFDG) was calculated for both Iica and Iest against that of Itp. RESULTS: In simulated studies, the mean percentage errors of MRFDG between true simulated and estimated values of Iest and Iica were 8.2% and 4.2%, respectively. In clinical studies, six clinical cases were collected. The mean percentage errors and standard deviations of MRFDG with Iest and Iica were 12.6+/-7.5% and 7.7+/-3.3%, respectively. CONCLUSIONS: We have proposed a technique for estimating image-derived input functions using independent component analysis without blood sampling. The results of our method were highly correlated with those from standard blood sampling, and more accurate than those of other methods proposed previously.