Position dependence of recovery coefficients in 177Lu-SPECT/CT reconstructions - phantom simulations and measurements.

BACKGROUND: Although the importance of quantitative SPECT has increased tremendously due to newly developed therapeutic radiopharmaceuticals, there are still no accreditation programs to harmonize SPECT imaging. Work is currently underway to develop an accreditation for quantitative 177Lu SPECT/CT. The aim of this study is to verify whether the positioning of the spheres within the phantom has an influence on the recovery and thus needs to be considered in SPECT harmonization. In addition, the effects of these recovery coefficients on a potential partial volume correction as well as absorbed-dose estimates are investigated. METHODS: Using a low-dose CT of a SPECT/CT acquisition, a computerized version of the NEMA body phantom was created using a semi-automatic threshold-based method. Based on the mass-density map, the detector orbit, and the sphere centers, realistic SPECT acquisitions of all possible 720 sphere configurations of both the PET and the SPECT versions of the NEMA Body Phantom were generated using Monte Carlo simulations. SPECT reconstructions with different numbers of updates were performed without (CASToR) and with resolution modeling (STIR). Recovery coefficients were calculated for all permutations, reconstruction methods, and phantoms, and their dependence on the sphere positioning was investigated. Finally, the simulation-based findings were validated using SPECT/CT acquisitions of six different sphere configurations. RESULTS: Our analysis shows that sphere positioning has a significant impact on the recovery for both of the reconstruction methods and the phantom type. Although resolution modeling resulted in significantly higher recovery, the relative variation in recovery within the 720 permutations was even larger. When examining the extreme values of the recovery, reconstructions without resolution modeling were influenced primarily by the sphere position, while with resolution modeling the volume of the two adjacent spheres had a larger influence. The SPECT measurements confirmed these observations, and the recovery curves showed good overall agreement with the simulated data. CONCLUSION: Our study shows that sphere positioning has a significant impact on the recovery obtained in NEMA sphere phantom measurements and should therefore be considered in a future SPECT accreditation. Furthermore, the single-measurement method normally performed for PVC should be reconsidered to account for the position dependency.

A Deep-Learning-Based Partial-Volume Correction Method for Quantitative 177Lu SPECT/CT Imaging.

With the development of new radiopharmaceutical therapies, quantitative SPECT/CT has progressively emerged as a crucial tool for dosimetry. One major obstacle of SPECT is its poor resolution, which results in blurring of the activity distribution. Especially for small objects, this so-called partial-volume effect limits the accuracy of activity quantification. Numerous methods for partial-volume correction (PVC) have been proposed, but most methods have the disadvantage of assuming a spatially invariant resolution of the imaging system, which does not hold for SPECT. Furthermore, most methods require a segmentation based on anatomic information. Methods: We introduce DL-PVC, a methodology for PVC of 177Lu SPECT/CT imaging using deep learning (DL). Training was based on a dataset of 10,000 random activity distributions placed in extended cardiac-torso body phantoms. Realistic SPECT acquisitions were created using the SIMIND Monte Carlo simulation program. SPECT reconstructions without and with resolution modeling were performed using the CASToR and STIR reconstruction software, respectively. The pairs of ground-truth activity distributions and simulated SPECT images were used for training various U-Nets. Quantitative analysis of the performance of these U-Nets was based on metrics such as the structural similarity index measure or normalized root-mean-square error, but also on volume activity accuracy, a new metric that describes the fraction of voxels in which the determined activity concentration deviates from the true activity concentration by less than a certain margin. On the basis of this analysis, the optimal parameters for normalization, input size, and network architecture were identified. Results: Our simulation-based analysis revealed that DL-PVC (0.95/7.8%/35.8% for structural similarity index measure/normalized root-mean-square error/volume activity accuracy) outperforms SPECT without PVC (0.89/10.4%/12.1%) and after iterative Yang PVC (0.94/8.6%/15.1%). Additionally, we validated DL-PVC on 177Lu SPECT/CT measurements of 3-dimensionally printed phantoms of different geometries. Although DL-PVC showed activity recovery similar to that of the iterative Yang method, no segmentation was required. In addition, DL-PVC was able to correct other image artifacts such as Gibbs ringing, making it clearly superior at the voxel level. Conclusion: In this work, we demonstrate the added value of DL-PVC for quantitative 177Lu SPECT/CT. Our analysis validates the functionality of DL-PVC and paves the way for future deployment on clinical image data.

Effect of kilovoltage and quality reference mAs on CT-based attenuation correction in 177Lu SPECT/CT imaging: a phantom study.

INTRODUCTION: CT-based attenuation correction (CT-AC) plays a major role in accurate activity quantification by SPECT/CT imaging. However, the effect of kilovoltage peak (kVp) and quality-reference mAs (QRM) on the attenuation coefficient image (µ-map) and volume CT dose index (CTDIvol) have not yet been systematically evaluated. Therefore, the aim of this study was to fill this gap and investigate the influence of kVp and QRM on CT-AC in 177Lu SPECT/CT imaging. METHODS: Seventy low-dose CT acquisitions of an Electron Density Phantom (seventeen inserts of nine tissue-equivalent materials) were acquired using various kVp and QRM combinations on a Siemens Symbia Intevo Bold SPECT/CT system. Using manufacturer reconstruction software, 177Lu µ-maps were generated for each CT image, and three low-dose CT related aspects were examined. First, the µ-map-based attenuation values (µmeasured) were compared with theoretical values (µtheoretical). Second, changes in 177Lu activity expected due to changes in the µ-map were calculated using a modified Chang method. Third, the noise in the µ-map was assessed by measuring the coefficient of variation in a volume of interest in the homogeneous section of the Electron Density Phantom. Lastly, two phantoms were designed to simulate attenuation in four tissue-equivalent materials for two different source geometries (1-mL and 10-mL syringes). 177Lu SPECT/CT imaging was performed using three different reconstruction algorithms (xSPECT Quant, Flash3D, STIR), and the SPECT-based activities were compared against the nominal activities in the sources. RESULTS: The largest relative errors between µmeasured and µtheoretical were observed in the lung inhale insert (range: 18%-36%), while it remained below 6% for all other inserts. The resulting changes in 177Lu activity quantification were -3.5% in the lung inhale insert and less than -2.3% in all other inserts. Coefficient of variation and CTDIvol ranged from 0.3% and 3.6 mGy (130 kVp, 35 mAs) to 0.4% and 0.9 mGy (80 kVp, 20 mAs), respectively. The SPECT-based activity quantification using xSPECT Quant reconstructions outperformed all other reconstruction algorithms. CONCLUSION: This study shows that kVp and QRM values in low-dose CT imaging have a minimum effect on quantitative 177Lu SPECT/CT imaging, while the selection of low values of kVp and QRM reduce the CTDIvol.

Radiation-induced double-strand breaks by internal ex vivo irradiation of lymphocytes: Validation of a Monte Carlo simulation model using GATE and Geant4-DNA.

This study describes a method to validate a radiation transport model that quantifies the number of DNA double-strand breaks (DSB) produced in the lymphocyte nucleus by internal ex vivo irradiation of whole blood with the radionuclides 90Y, 99mTc, 123I, 131I, 177Lu, 223Ra, and 225Ac in a test vial using the GATE/Geant4 code at the macroscopic level and the Geant4-DNA code at the microscopic level. METHODS: The simulation at the macroscopic level reproduces an 8 mL cylindrical water-equivalent medium contained in a vial that mimics the geometry for internal ex vivo blood irradiation. The lymphocytes were simulated as spheres of 3.75 µm radius randomly distributed, with a concentration of 125 spheres/mL. A phase-space actor was attached to each sphere to register all the entering particles. The simulation at the microscopic level for each radionuclide was performed using the Geant4-DNA tool kit, which includes the clustering example centered on a density-based spatial clustering of applications with noise (DBSCAN) algorithm. The irradiation source was constructed by generating a single phase space from the sum of all phase spaces. The lymphocyte nucleus was defined as a water sphere of a 3.1 µm radius. The absorbed dose coefficients for lymphocyte nuclei (dLymph) were calculated and compared with macroscopic whole blood absorbed dose coefficients (dBlood). The DBSCAN algorithm was used to calculate the number of DSBs. Lastly, the number of DSB∙cell-1∙mGy-1 (simulation) was compared with the number of radiation-induced foci per cell and absorbed dose (RIF∙cell-1∙mGy-1) provided by experimental data for gamma and beta emitting radionuclides. For alpha emitters, dLymph and the number of α-tracks∙100 cell-1∙mGy-1 and DBSs∙µm-1 were calculated using experiment-based thresholds for the α-track lengths and DBSs/track values. The results were compared with the results of an ex vivo study with 223Ra. RESULTS: The dLymph values differed from the dBlood values by -1.0% (90Y), -5.2% (99mTc), -22.3% (123I), 0.35% (131I), 2.4% (177Lu), -5.6% (223Ra) and -6.1% (225Ac). The number of DSB∙cell-1∙mGy-1 for each radionuclide was 0.015 DSB∙cell-1∙mGy-1 (90Y), 0.012 DSB∙cell-1∙mGy-1 (99mTc), 0.014DSB∙cell-1∙mGy-1 (123I), 0.012 DSB∙cell-1∙mGy-1 (131I), and 0.016 DSB∙cell-1∙mGy-1 (177Lu). These values agree very well with experimental data. The number of α-tracks∙100 cells-1∙mGy-1 for 223Ra and 225Ac where 0.144 α-tracks∙100 cells-1∙mGy-1 and 0.151 α-tracks∙100 cells-1∙mGy-1, respectively. These values agree very well with experimental data. Moreover, the linear density of DSBs per micrometer α-track length were 11.13 ±â€¯0.04 DSB/µm and 10.86 ±â€¯0.06 DSB/µm for 223Ra and 225Ac, respectively. CONCLUSION: This study describes a model to simulate the DNA DSB damage in lymphocyte nuclei validated by experimental data obtained from internal ex vivo blood irradiation with radionuclides frequently used in diagnostic and therapeutic procedures in nuclear medicine.

GATE/Geant4-based dosimetry for ex vivo in solution irradiation of blood with radionuclides.

To establish a dose-response relationship between radiation-induced DNA damage and the corresponding absorbed doses in blood irradiated with radionuclides in solution under ex vivo conditions, the absorbed dose coefficient for 1 ml for 1 h internal ex vivo irradiation of peripheral blood (dBlood) must be determined. dBlood is specific for each radionuclide, and it depends on the irradiation geometry. Therefore, the aim of this study is to use the Monte Carlo radiation transport code GATE/Geant4 to calculate the mean absorbed dose rates for ex vivo irradiation of blood with several radionuclides used in Nuclear Medicine. METHODS: The Monte Carlo simulation reproduces the irradiation geometry of a blood sample of 7 ml mixed with 1 ml of a water equivalent radioactive solution in an 8 ml vial. The simulation was performed for ten different radionuclides: 18F, 68Ga, 90Y, 99mTc, 123I, 124I, 131I, 177Lu, 223Ra, and 225Ac. Two sets of simulations for each radionuclide were performed with 1x109 histories. The first set was simulated with a mass density of 1.0525 g/cm3 of the blood plus water mixture. The second set of simulations was performed with a mass density of 1 g/cm3 for comparison with previous studies. RESULTS: The values of dBlood for ten radionuclides were calculated. The values range from 10.23 mGy∙ml∙MBq-1 for 99mTc to 15632.02 mGy∙ml∙MBq-1 for 225Ac. The maximum relative change compared to previous studies was 13.0% for 124I. CONCLUSION: This study provides a comprehensive set of absorbed dose coefficients for 1 ml for 1 h internal ex vivo irradiation of peripheral blood in a special vial geometry and radionuclides typically used in Nuclear Medicine. Furthermore, the method proposed by this work can be easily adapted to a variety of internal irradiation conditions and serve as a reference for future studies.

Analysis of a deep learning-based method for generation of SPECT projections based on a large Monte Carlo simulated dataset.

BACKGROUND: In recent years, a lot of effort has been put in the enhancement of medical imaging using artificial intelligence. However, limited patient data in combination with the unavailability of a ground truth often pose a challenge to a systematic validation of such methodologies. The goal of this work was to investigate a recently proposed method for an artificial intelligence-based generation of synthetic SPECT projections, for acceleration of the image acquisition process based on a large dataset of realistic SPECT simulations. METHODS: A database of 10,000 SPECT projection datasets of heterogeneous activity distributions of randomly placed random shapes was simulated for a clinical SPECT/CT system using the SIMIND Monte Carlo program. Synthetic projections at fixed angular increments from a set of input projections at evenly distributed angles were generated by different u-shaped convolutional neural networks (u-nets). These u-nets differed in noise realization used for the training data, number of input projections, projection angle increment, and number of training/validation datasets. Synthetic projections were generated for 500 test projection datasets for each u-net, and a quantitative analysis was performed using statistical hypothesis tests based on structural similarity index measure and normalized root-mean-squared error. Additional simulations with varying detector orbits were performed on a subset of the dataset to study the effect of the detector orbit on the performance of the methodology. For verification of the results, the u-nets were applied to Jaszczak and NEMA physical phantom data obtained on a clinical SPECT/CT system. RESULTS: No statistically significant differences were observed between u-nets trained with different noise realizations. In contrast, a statistically significant deterioration was found for training with a small subset (400 datasets) of the 10,000 simulated projection datasets in comparison with using a large subset (9500 datasets) for training. A good agreement between synthetic (i.e., u-net generated) and simulated projections before adding noise demonstrates a denoising effect. Finally, the physical phantom measurements show that our findings also apply for projections measured on a clinical SPECT/CT system. CONCLUSION: Our study shows the large potential of u-nets for accelerating SPECT/CT imaging. In addition, our analysis numerically reveals a denoising effect when generating synthetic projections with a u-net. Clinically interesting, the methodology has proven robust against camera orbit deviations in a clinically realistic range. Lastly, we found that a small number of training samples (e.g., ~ 400 datasets) may not be sufficient for reliable generalization of the u-net.

Quantification of the volume fraction of fat, water and bone mineral in spongiosa for red marrow dosimetry in molecular radiotherapy by using a dual-energy (SPECT/)CT.

A patient-specific absorbed dose calculation for red marrow dosimetry requires quantifying patient-specific volume fractions of the red marrow, yellow marrow, and trabecular bone in the spongiosa of several skeletal sites. This quantification allows selecting appropriate S values calculated from the parameterized radiation transport models for bone and bone marrow dosimetry. Currently, no comprehensive, individualized, and non-invasive procedure is available for quantifying the volume fractions of red marrow, yellow marrow, and trabecular bone in the spongiosa. This study aims to provide a new quantitative method based on dual-energy computed tomography to fill this gap in red marrow dosimetry using a (SPECT/)CT system. METHODS: First, a method for parametrizing the photon attenuation coefficients relative to water was implemented. Next, a method to calculate the effective atomic number (Zeff) and effective mass density (ρeff) using dual-energy CT (DECT) was employed. Lastly, two- and three-material decomposition using a dual-energy quantitative CT method (DEQCT) was performed in an anthropomorphic spine phantom and two bone samples of a boar, respectively. The measurements of Zeff and ρeff were compared with the syngo.CT DE Rho/Z tool (Siemens Healthineers). Furthermore, the DEQCT method implemented in this study (DEQCT-I) was compared with a second DEQCT method based on the use of external material standards (DEQCT-II). DEQCT-II was used as reference method for calculating relative errors. RESULTS: The two-material decomposition in the anthropomorphic spine phantom presented a maximum relative error of -10% for the bone mineral density quantification. Furthermore, Zeff and ρeff calculated by DEQCT-I differed from syngo.CT DE Rho/Z tool by less than 4.4% and 1.9%, respectively. The three-material decomposition in the two bone samples showed a maximum relative error of 21%, -17%, and 15% for the quantification of the volume fractions of fat, water, and bone mineral equivalent materials. Lastly, Zeff and ρeff calculated by DEQCT-I differed from syngo.CT DE Rho/Z tool by less than 8.2% and 7.0%, respectively. CONCLUSION: This study shows that quantifying the volume fraction of fat, water, and bone mineral using a phantom-independent and post-reconstruction DEQCT method is feasible. DEQCT-I has the advantage of not requiring prior information about the X-ray spectra or the detector sensitivity function, as is the case with spectral-based DEQCT methods. Instead, DEQCT-I, similar to other DEQCT methods depends on the chemical description of reference materials and a beam hardening correction function. DEQCT-I method provides an individualized and non-invasive procedure using a (SPECT/)CT system to apply S values based on the patient-specific volume fractions of yellow marrow, red marrow, and bone mineral in red marrow dosimetry.

What You See Is Not What You Get: On the Accuracy of Voxel-Based Dosimetry in Molecular Radiotherapy.

Improvements in quantitative SPECT/CT have aroused growing interest in voxel-based dosimetry for radionuclide therapies, because it promises visualization of absorbed doses at a voxel level. In this work, SPECT/CT-based voxel-level dosimetry of a 3-dimensional (3D) printed 2-compartment kidney phantom was performed, and the resulting absorbed dose distributions were examined. Additionally, the potential of the PETPVC partial-volume correction tool was investigated. Methods: Both kidney compartments (70% cortex, 30% medulla) were filled with different activity concentrations, and SPECT/CT imaging was performed. The images were reconstructed using varying settings (iterations, subsets, and postfiltering). On the basis of these activity concentration maps, absorbed dose distributions were calculated with precalculated 177Lu voxel S values and an empiric kidney half-life. An additional set of absorbed doses was calculated after applying PETPVC for partial-volume correction of the SPECT reconstructions. Results: SPECT/CT imaging blurs the 2 discrete suborgan absorbed dose values into a continuous distribution. Although this effect is slightly improved by applying more iterations, it is enhanced by additional postfiltering. By applying PETPVC, the absorbed dose values are separated into 2 peaks. Although this leads to a better agreement between SPECT/CT-based and nominal values, considerable discrepancies remain. In contrast to the calculated nominal absorbed doses of 7.8 and 1.6 Gy (in the cortex and medulla, respectively), SPECT/CT-based voxel-level dosimetry resulted in mean absorbed doses of 3.0-6.6 Gy (cortex) and 2.7-5.1 Gy (medulla). PETPVC led to improved ranges of 6.1-8.9 Gy (cortex) and 2.1-5.4 Gy (medulla). Conclusion: Our study showed that 177Lu quantitative SPECT/CT imaging leads to voxel-based dose distributions largely differing from the real organ distribution. SPECT/CT imaging and reconstruction deficiencies might directly translate into unrealistic absorbed dose distributions, thus questioning the reliability of SPECT-based voxel-level dosimetry. Therefore, SPECT/CT reconstructions should be adapted to ensure an accurate quantification of the underlying activity and, therefore, absorbed dose in a volume of interest of the expected object size (e.g., organs, organ substructures, lesions, or voxels). As an example, PETPVC largely improves the match between SPECT/CT-based and nominal dose distributions. In conclusion, the concept of voxel-based dosimetry should be treated with caution. Specifically, one should remember that the absorbed dose distribution is mainly a convolved version of the underlying SPECT reconstruction.

Quantification of the trabecular bone volume fraction for bone marrow dosimetry in molecular radiotherapy by using a dual-energy (SPECT/)CT.

A complete characterization of spongiosa (bone marrow plus trabecular bone) is required to calculate the absorbed dose to active bone marrow. Due to the complex microanatomy, it is necessary to apply non-conventional imaging methods in nuclear medicine. The aim of this study is validating a phantomless quantification method using dual-energy quantitative computed tomography (DEQCT) for the quantification of trabecular bone volume fraction for bone marrow dosimetry in molecular radiotherapy. First, a phantomless quantification method (mass fraction method) based on x-ray beam and detector sensitivity was validated in an integrated dual energy SPECT/CT and in a dual source computed tomography (DSCT) system for comparison. The validation was performed in a phantom consisting of different water, fat and hydroxyapatite compositions. Moreover, the European spine phantom (ESP) was used to simulate the spine geometry. Bone mineral content (BMC) of the whole vertebra and bone mineral density (BMD) in the spongiosa region of each phantom vertebra were measured using DEQCT and dual energy x-ray absorptiometry (DEXA). Lastly, BMC was measured in a patient using DEQCT and DEXA. Measured values of hydroxyapatite fraction and nominal values in the homemade phantom showed a good correlation. The relative error remained below 14.2%. Quantification of BMC (in whole vertebra) and BMD (in spongiosa) in the ESP showed a good agreement between measured values and nominal values. The relative error remained between 0.7% and 7.5% for BMCSPECT/CT, 1.1% and 7.7% for BMCDSCT, 5.4% and 32.0 for BMDSPECT/CT, and 59.4% and 10.0% for BMDDSCT. Quantification of BMC in lumbar vertebrae 1 and 2 of a patient showed relative errors of 7.6% and -8.4% between DEXA and DSCT. Our study shows that DEQCT using a mass fraction method (phantomless) enables quantification of hydroxyapatite in a clinical nuclear medicine setting. An overestimation of the hydroxyapatite volume fraction was observed in all quantified regions, in particular in the spongiosa region of ESP. This result might be related to insufficient information about the x-ray spectra and the detector sensitivity function.

Quantification of fat fraction in lumbar vertebrae: correlation with age and implications for bone marrow dosimetry in molecular radiotherapy.

Absorbed dose to active bone marrow is a predictor of hematological toxicity in molecular radiotherapy. Due to the complex composition of bone marrow tissue, the necessity to improve the personalized dosimetry has led to the application of non-conventional imaging methods in nuclear medicine. The aim of this study is to apply magnetic resonance imaging (MRI) for quantification of the fat fraction in lumbar vertebrae and to analyze its implications for bone marrow dosimetry. First, a highly accelerated two-point Dixon MRI sequence for fat-water separation was validated in a 3T system against the magnetic resonance spectroscopy (MRS) gold standard. The validation was performed in a fat-water phantom composed of 11 vials with different fat fractions between 0% and 100%, and subsequently repeated in the lumbar vertebrae of three healthy volunteers. Finally, a retrospective study was performed by analyzing the fat fraction in five lumbar vertebrae of 44 patients scanned with the two-point Dixon sequence. The two-point Dixon phantom acquisition showed a good agreement (maximum difference = 2.9%) between the nominal fat fraction and MRS. In the volunteers, a statistical analysis showed a non-significant difference (p = 0.19) between MRI and MRS. In the patients, gender-specific linear fits for female and male data indicated that the age-dependent marrow conversion (red → yellow marrow) is slower in males (0.3% per year) than in females (0.5% per year). Lastly, the fat fraction values showed a considerable variability in patients of similar ages and the same gender. Two-point Dixon MRI enables a non-invasive and spatially resolved quantification of the fat fraction in bone marrow. Our study provides important evidence on the differences in marrow conversion between females and males. In addition, differences were observed in the cellularity values of the International Commission on Radiological Protection (ICRP) reference man (0.7) and the median values obtained in our patient group. These observations lead to the conclusion that the fat fraction in bone marrow should be considered as a patient-specific variable in clinical dosimetry procedures.