Partial-volume correction in dynamic PET-CT: effect on tumor kinetic parameter estimation and validation of simplified metrics

Deep learning-based partial volume correction in standard and low-dose positron emission tomography-computed tomography imaging

BACKGROUND

Positron emission tomography (PET) imaging encounters the obstacle of partial volume effects, arising from its limited intrinsic resolution, giving rise to (I) considerable bias, particularly for structures comparable in size to the point spread function (PSF) of the system; and (II) blurred image edges and blending of textures along the borders. We set out to build a deep learning-based framework for predicting partial volume corrected full-dose (FD + PVC) images from either standard or low-dose (LD) PET images without requiring any anatomical data in order to provide a joint solution for partial volume correction and de-noise LD PET images.

METHODS

We trained a modified encoder-decoder U-Net network with standard of care or LD PET images as the input and FD + PVC images by six different PVC methods as the target. These six PVC approaches include geometric transfer matrix (GTM), multi-target correction (MTC), region-based voxel-wise correction (RBV), iterative Yang (IY), reblurred Van-Cittert (RVC), and Richardson-Lucy (RL). The proposed models were evaluated using standard criteria, such as peak signal-to-noise ratio (PSNR), root mean squared error (RMSE), structural similarity index (SSIM), relative bias, and absolute relative bias.

RESULTS

Different levels of error were observed for these partial volume correction methods, which were relatively smaller for GTM with a SSIM of 0.63 for LD and 0.29 for FD, IY with an SSIM of 0.63 for LD and 0.67 for FD, RBV with an SSIM of 0.57 for LD and 0.65 for FD, and RVC with an SSIM of 0.89 for LD and 0.94 for FD PVC approaches. However, large quantitative errors were observed for multi-target MTC with an RMSE of 2.71 for LD and 2.45 for FD and RL with an RMSE of 5 for LD and 3.27 for FD PVC approaches.

CONCLUSIONS

We found that the proposed framework could effectively perform joint de-noising and partial volume correction for PET images with LD and FD input PET data (LD vs. FD). When no magnetic resonance imaging (MRI) images are available, the developed deep learning models could be used for partial volume correction on LD or standard PET-computed tomography (PET-CT) scans as an image quality enhancement technique.

Validation of image-derived input function using a long axial field of view PET/CT scanner for two different tracers

BACKGROUND

Accurate image-derived input function (IDIF) from highly sensitive large axial field of view (LAFOV) PET/CT scanners could avoid the need of invasive blood sampling for kinetic modelling. The aim is to validate the use of IDIF for two kinds of tracers, 3 different IDIF locations and 9 different reconstruction settings.

METHODS

Eight [18F]FDG and 10 [18F]DPA-714 scans were acquired respectively during 70 and 60 min on the Vision Quadra PET/CT system. PET images were reconstructed using various reconstruction settings. IDIFs were taken from ascending aorta (AA), descending aorta (DA), and left ventricular cavity (LV). The calibration factor (CF) extracted from the comparison between the IDIFs and the manual blood samples as reference was used for IDIFs accuracy and precision assessment. To illustrate the effect of various calibrated-IDIFs on Patlak linearization for [18F]FDG and Logan linearization for [18F]DPA-714, the same target time-activity curves were applied for each calibrated-IDIF.

RESULTS

For [18F]FDG, the accuracy and precision of the IDIFs were high (mean CF ≥ 0.82, SD ≤ 0.06). Compared to the striatum influx (Ki) extracted using calibrated AA IDIF with the updated European Association of Nuclear Medicine Research Ltd. standard reconstruction (EARL2), Ki mean differences were < 2% using the other calibrated IDIFs. For [18F]DPA714, high accuracy of the IDIFs was observed (mean CF ≥ 0.86) except using absolute scatter correction, DA and LV (respectively mean CF = 0.68, 0.47 and 0.44). However, the precision of the AA IDIFs was low (SD ≥ 0.10). Compared to the distribution volume (VT) in a frontal region obtained using calibrated continuous arterial sampler input function as reference, VT mean differences were small using calibrated AA IDIFs (for example VT mean difference = -5.3% using EARL2), but higher using calibrated DA and LV IDIFs (respectively + 12.5% and + 19.1%).

CONCLUSIONS

For [18F]FDG, IDIF do not need calibration against manual blood samples. For [18F]DPA-714, AA IDIF can replace continuous arterial sampling for simplified kinetic quantification but only with calibration against arterial blood samples. The accuracy and precision of IDIF from LAFOV PET/CT system depend on tracer, reconstruction settings and IDIF VOI locations, warranting careful optimization.

Clinical feasibility study of early 30-minute dynamic FDG-PET scanning protocol for patients with lung lesions

PURPOSE

This study aimed to evaluate the clinical feasibility of early 30-minute dynamic 2-deoxy-2-[18F]fluoro-D-glucose (18F-FDG) positron emission tomography (PET) scanning protocol for patients with lung lesions in comparison to the standard 65-minute dynamic FDG-PET scanning as a reference.

METHODS

Dynamic 18F-FDG PET images of 146 patients with 181 lung lesions (including 146 lesions confirmed by histology) were analyzed in this prospective study. Dynamic images were reconstructed into 28 frames with a specific temporal division protocol for the scan data acquired 65 min post-injection. Ki images and quantitative parameters Ki based on two different acquisition durations [the first 30 min (Ki-30 min) and 65 min (Ki-65 min)] were obtained by applying the irreversible two-tissue compartment model using in-house Matlab software. The two acquisition durations were compared for Ki image quality (including visual score analysis and number of lesions detected) and Ki value (including accuracy of Ki, the value of differential diagnosis of lung lesions and prediction of PD-L1 status) by Wilcoxon's rank sum test, Spearman's rank correlation analysis, receiver operating characteristic (ROC) curve, and the DeLong test. The significant testing level (alpha) was set to 0.05.

RESULTS

The quality of the Ki-30 min images was not significantly different from the Ki-65 min images based on visual score analysis (P > 0.05). In terms of Ki value, among 181 lesions, Ki-65 min was statistically higher than Ki-30 min (0.027 ± 0.017 ml/g/min vs. 0.026 ± 0.018 ml/g/min, P < 0.05), while a very high correlation was obtained between Ki-65 min and Ki-30 min (r = 0.977, P < 0.05). In the differential diagnosis of lung lesions, ROC analysis was performed on 146 histologically confirmed lesions, the area under the curve (AUC) of Ki-65 min, Ki-30 min, and SUVmax was 0.816, 0.816, and 0.709, respectively. According to the Delong test, no significant differences in the diagnostic accuracies were found between Ki-65 min and Ki-30 min (P > 0.05), while the diagnostic accuracies of Ki-65 min and Ki-30 min were both significantly higher than that of SUVmax (P < 0.05). In 73 (NSCLC) lesions with definite PD-L1 expression results, the Ki-65 min, Ki-30 min, and SUVmax in PD-L1 positivity were significantly higher than that in PD-L1 negativity (P < 0.05). And no significant differences in predicting PD-L1 positivity were found among Ki-65 min, Ki-30 min, and SUVmax (AUC = 0.704, 0.695, and 0.737, respectively, P > 0.05), according to the results of ROC analysis and Delong test.

CONCLUSIONS

This study indicates that an early 30-minute dynamic FDG-PET acquisition appears to be sufficient to provide quantitative images with good-quality and accurate Ki values for the assessment of lung lesions and prediction of PD-L1 expression. Protocols with a shortened early 30-minute acquisition time may be considered for patients who have difficulty with prolonged acquisitions to improve the efficiency of clinical acquisitions.

A cycle-consistent adversarial network for brain PET partial volume correction without prior anatomical information

PURPOSE

Partial volume effect (PVE) is a consequence of the limited spatial resolution of PET scanners. PVE can cause the intensity values of a particular voxel to be underestimated or overestimated due to the effect of surrounding tracer uptake. We propose a novel partial volume correction (PVC) technique to overcome the adverse effects of PVE on PET images.

METHODS

Two hundred and twelve clinical brain PET scans, including 50 18F-Fluorodeoxyglucose (18F-FDG), 50 18F-Flortaucipir, 36 18F-Flutemetamol, and 76 18F-FluoroDOPA, and their corresponding T1-weighted MR images were enrolled in this study. The Iterative Yang technique was used for PVC as a reference or surrogate of the ground truth for evaluation. A cycle-consistent adversarial network (CycleGAN) was trained to directly map non-PVC PET images to PVC PET images. Quantitative analysis using various metrics, including structural similarity index (SSIM), root mean squared error (RMSE), and peak signal-to-noise ratio (PSNR), was performed. Furthermore, voxel-wise and region-wise-based correlations of activity concentration between the predicted and reference images were evaluated through joint histogram and Bland and Altman analysis. In addition, radiomic analysis was performed by calculating 20 radiomic features within 83 brain regions. Finally, a voxel-wise two-sample t-test was used to compare the predicted PVC PET images with reference PVC images for each radiotracer.

RESULTS

The Bland and Altman analysis showed the largest and smallest variance for 18F-FDG (95% CI: - 0.29, + 0.33 SUV, mean = 0.02 SUV) and 18F-Flutemetamol (95% CI: - 0.26, + 0.24 SUV, mean =  - 0.01 SUV), respectively. The PSNR was lowest (29.64 ± 1.13 dB) for 18F-FDG and highest (36.01 ± 3.26 dB) for 18F-Flutemetamol. The smallest and largest SSIM were achieved for 18F-FDG (0.93 ± 0.01) and 18F-Flutemetamol (0.97 ± 0.01), respectively. The average relative error for the kurtosis radiomic feature was 3.32%, 9.39%, 4.17%, and 4.55%, while it was 4.74%, 8.80%, 7.27%, and 6.81% for NGLDM_contrast feature for 18F-Flutemetamol, 18F-FluoroDOPA, 18F-FDG, and 18F-Flortaucipir, respectively.

CONCLUSION

An end-to-end CycleGAN PVC method was developed and evaluated. Our model generates PVC images from the original non-PVC PET images without requiring additional anatomical information, such as MRI or CT. Our model eliminates the need for accurate registration or segmentation or PET scanner system response characterization. In addition, no assumptions regarding anatomical structure size, homogeneity, boundary, or background level are required.

Immunocyte count combined with CT features for distinguishing pulmonary tuberculoma from malignancy among non-calcified solitary pulmonary solid nodules

Background

Tuberculoma is the most common type of surgically removed benign solid solitary pulmonary nodule (SPN) and can lead to a high risk of misdiagnoses for clinicians. This study aimed to discuss the value of the immunocyte count combined with computed tomography (CT) features in distinguishing pulmonary tuberculoma from malignancy among non-calcified solid SPNs.

Methods

Forty-eight patients with pulmonary tuberculoma and 52 patients with lung cancer were retrospectively included in our study. Univariate and multivariate analyses were conducted to screen the independent predictors. Receiver operating characteristic (ROC) analysis was performed to investigate the validity of the predictive model.

Results

The univariate and multivariate analyses revealed that a coarse margin, vacuole, lobulation, pleural indentation, cluster of differentiation (CD)3⁺ T-lymphocyte count, and CD4⁺ T-lymphocyte count were independent predictors for distinguishing pulmonary tuberculoma from malignancy. The sensitivity, specificity, accuracy, and the area under the ROC curve of the model comprising the CD3⁺ T-lymphocyte count were 79.2%, 75%, 74.5%, and 0.845 [95% confidence interval (CI), 0.759-0.910], respectively, and those of the model involving the CD4⁺ T-lymphocyte count were 77.1%, 78.8%, 77.1%, and 0.857 (95% CI, 0.773-0.919), respectively.

Conclusions

Immunocyte count combined with CT features is efficient in distinguishing pulmonary tuberculoma from malignancy among non-calcified solid SPNs and has applicable clinical value.

Perspectives of brain imaging with PET systems

In this partial review and partial attempt at vision of what may be the future of dedicated brain PET scanners, the key implementations of the PET technique, we postulate that we are still on a development path and there is still a lot to be done in order to develop optimal brain imagers. Optimized for particular imaging tasks and protocols, and also mobile, that can be used outside the PET center, in addition to the expected improvements in sensitivity and resolution. For this multi-application concept to be more practical, flexible, adaptable designs are preferred. This task is greatly facilitated by the improved TOF performance that allows for more open, adjustable, limited angular coverage geometries without creating image artifacts. As achieving uniform very high resolution in the whole body is not practical due to technological limits and high costs, hybrid systems using a moderate-resolution total body scanner (such as J-PET) combined with a very high performing brain imager could be a very attractive approach. As well, as using magnification inserts in the total body or long-axial length imagers to visualize selected targets with higher resolution. In addition, multigamma imagers combining PET with Compton imaging should be developed to enable multitracer imaging.

Proof of concept of a multimodal intravital molecular imaging system for tumour transpathology investigation

Background

Transpathology highlights the interpretation of the underlying physiology behind molecular imaging. However, it remains challenging due to the discrepancies between in vivo and in vitro measurements and difficulties of precise co-registration between trans-scaled images. This study aims to develop a multimodal intravital molecular imaging (MIMI) system as a tool for in vivo tumour transpathology investigation.

Methods

The proposed MIMI system integrates high-resolution positron imaging, magnetic resonance imaging (MRI) and microscopic imaging on a dorsal skin window chamber on an athymic nude rat. The window chamber frame was designed to be compatible with multimodal imaging and its fiducial markers were customized for precise physical alignment among modalities. The co-registration accuracy was evaluated based on phantoms with thin catheters. For proof of concept, tumour models of the human colorectal adenocarcinoma cell line HT-29 were imaged. The tissue within the window chamber was sectioned, fixed and haematoxylin–eosin (HE) stained for comparison with multimodal in vivo imaging.

Results

The final MIMI system had a maximum field of view (FOV) of 18 mm × 18 mm. Using the fiducial markers and the tubing phantom, the co-registration errors are 0.18 ± 0.27 mm between MRI and positron imaging, 0.19 ± 0.22 mm between positron imaging and microscopic imaging and 0.15 ± 0.27 mm between MRI and microscopic imaging. A pilot test demonstrated that the MIMI system provides an integrative visualization of the tumour anatomy, vasculatures and metabolism of the in vivo tumour microenvironment, which was consistent with ex vivo pathology.

Conclusions

The established multimodal intravital imaging system provided a co-registered in vivo platform for trans-scale and transparent investigation of the underlying pathology behind imaging, which has the potential to enhance the translation of molecular imaging.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00259-021-05574-y.

PET/CT-Based Response Evaluation in Cancer-a Systematic Review of Design Issues

Positron emission tomography/x-ray computed tomography (PET/CT) has long been discussed as a promising modality for response evaluation in cancer. When designing respective clinical trials, several design issues have to be addressed, especially the number/timing of PET/CT scans, the approach for quantifying metabolic activity, and the final translation of measurements into a rule. It is unclear how well these issues have been tackled in quest of an optimised use of PET/CT in response evaluation. Medline via Ovid and Science Citation Index via Web of Science were systematically searched for articles from 2015 on cancer patients scanned with PET/CT before and during/after treatment. Reports were categorised as being either developmental or evaluative, i.e. focusing on either the establishment or the evaluation of a rule discriminating responders from non-responders. Of 124 included papers, 112 (90 %) were accuracy and/or prognostic studies; the remainder were response-curve studies. No randomised controlled trials were found. Most studies were prospective (62 %) and from single centres (85 %); median number of patients was 38.5 (range 5-354). Most (69 %) of the studies employed only one post-baseline scan. Quantification was mainly based on SUVmax (91 %), while change over time was most frequently used to combine measurements into a rule (79 %). Half of the reports were categorised as developmental, the other half evaluative. Most development studies assessed only one element (35/62, 56 %), most frequently the choice of cut-off points (25/62, 40 %). In summary, the majority of studies did not address the essential open issues in establishing PET/CT for response evaluation. Reasonably sized multicentre studies are needed to systematically compare the many different options when using PET/CT for response evaluation.

PET Imaging of Tumor Perfusion: A Potential Cancer Biomarker?

Perfusion, as measured by imaging, is considered a standard of care biomarker for the evaluation of many tumors. Measurements of tumor perfusion may be used in a number of ways, including improving the visual detection of lesions, differentiating malignant from benign findings, assessing aggressiveness of tumors, identifying ischemia and by extension hypoxia within tumors, and assessing treatment response. While most clinical perfusion imaging is currently performed with CT or MR, a number of methods for PET imaging of tumor perfusion have been described. The inert PET radiotracer 15O-water PET represents the recognized gold standard for absolute quantification of tissue perfusion in both normal tissue and a variety of pathological conditions including cancer. Other cancer PET perfusion imaging strategies include the use of radiotracers with high first-pass uptake, analogous to those used in cardiac perfusion PET. This strategy produces more visually pleasing high-contrast images that provide relative rather than absolute perfusion quantification. Lastly, multiple timepoint imaging of PET tracers such as 18F-FDG, are not specifically optimized for perfusion, but have advantages related to availability, convenience, and reimbursement. Multiple obstacles have thus far blocked the routine use of PET imaging for tumor perfusion, including tracer production and distribution, image processing, patient body coverage, clinical validation, regulatory approval and reimbursement, and finally feasible clinical workflows. Fortunately, these obstacles are being overcome, especially within larger imaging centers, opening the door for PET imaging of tumor perfusion to become standard clinical practice. In the foreseeable future, it is possible that whole-body PET perfusion imaging with 15O-water will be able to be performed in a single imaging session concurrent with standard PET imaging techniques such as 18F-FDG-PET. This approach could establish an efficient clinical workflow. The resultant ability to measure absolute tumor blood flow in combination with glycolysis will provide important complementary information to inform prognosis and clinical decisions.

Multiple metabolic parameters and visual assessment of 18F-FDG uptake heterogeneity of PET/CT in advanced gastric cancer and primary gastric lymphoma

PURPOSE

Advanced gastric cancer (AGC) and primary gastric lymphoma (PGL) are the two most common malignant tumors of the stomach. Conventional imaging examinations have difficulty distinguishing the two. This study explored the values of multiple parameters and visual assessment of ¹⁸F-fluorodeoxyglucose(¹⁸F-FDG) uptake heterogeneity of positron emission tomography/computed tomography(PET/CT) for differentiating between AGC and PGL.

METHODS

This retrospective study included 70 AGC and 26 PGL patients, all of whom had undergone ¹⁸F-FDG PET/CT before treatment. We analyzed the differences between AGC and PGL in the distribution of metastatic lesions and multiple metabolic parameters, including the maximum standardized uptake value (SUVmax), SUVmax/maximal thickness(THKmax), metabolic tumor volume and total lesion glycolysis (TLG). In addition, ¹⁸F-FDG uptake heterogeneity was visually assessed using a visual scoring method and a method of measuring SUVmax differences (SUVmax-d).

RESULTS

The most common metastasis of AGC patients were liver, bone, peritoneal and proximal lymph nodes; PGL patients had fewer peritoneal metastases and lymph node metastasis could appeared to be "skip metastasis." The metabolic parameters-SUVmax, SUVmax/THKmax and TLG-were higher in patients who had PGL, especially in diffuse large B-cell lymphoma (DLBCL). In the visual assessment of ¹⁸F-FDG uptake heterogeneity, the measurements of SUVmax-d in PGL were significantly higher than in AGC. Receiver operating characteristics curve analysis suggested that SUVmax has the highest comprehensive diagnostic efficiency due to having the highest value of area under the curve and the highest accuracy (77.2%).

CONCLUSION

¹⁸F-FDG PET/CT had a high diagnostic efficiency for discrimination of AGC and PGL, especially between DLBCL and other pathological subtypes. Visual assessment used to evaluate ¹⁸F-FDG uptake heterogeneity could help to distinguish the two types of tumors. In addition, our innovative method of measuring the heterogeneity of ¹⁸F-FDG uptake-namely, SUVmax-d-could contribute to identification of the two tumor types and should have its significance clarified by future studies.

Tau pathology and relative cerebral blood flow are independently associated with cognition in Alzheimer's disease

Purpose

We aimed to investigate associations between tau pathology and relative cerebral blood flow (rCBF), and their relationship with cognition in Alzheimer’s disease (AD), by using a single dynamic [¹⁸F]flortaucipir positron emission tomography (PET) scan.

Methods

Seventy-one subjects with AD (66 ± 8 years, mini-mental state examination (MMSE) 23 ± 4) underwent a dynamic 130-min [¹⁸F]flortaucipir PET scan. Cognitive assessment consisted of composite scores of four cognitive domains. For tau pathology and rCBF, receptor parametric mapping (cerebellar gray matter reference region) was used to create uncorrected and partial volume-corrected parametric images of non-displaceable binding potential (BP_ND) and R₁, respectively. (Voxel-wise) linear regressions were used to investigate associations between BP_ND and/or R₁ and cognition_.

Results

Higher [¹⁸F]flortaucipir BP_ND was associated with lower R₁ in the lateral temporal, parietal and occipital regions. Higher medial temporal BP_ND was associated with worse memory, and higher lateral temporal BP_ND with worse executive functioning and language. Higher parietal BP_ND was associated with worse executive functioning, language and attention, and higher occipital BP_ND with lower cognitive scores across all domains. Higher frontal BP_ND was associated with worse executive function and attention. For [¹⁸F]flortaucipir R₁, lower values in the lateral temporal and parietal ROIs were associated with worse executive functioning, language and attention, and lower occipital R₁ with lower language and attention scores. When [¹⁸F]flortaucipir BP_ND and R₁ were modelled simultaneously, associations between lower R₁ in the lateral temporal ROI  and worse attention remained, as well as for lower parietal R₁ and worse executive functioning and attention.

Conclusion

Tau pathology was associated with locally reduced rCBF. Tau pathology and low rCBF were both independently associated with worse cognitive performance. For tau pathology, these associations spanned widespread neocortex, while for rCBF, independent associations were restricted to lateral temporal and parietal regions and the executive functioning and attention domains. These findings indicate that each biomarker may independently contribute to cognitive impairment in AD.

Electronic supplementary material

The online version of this article (10.1007/s00259-020-04831-w) contains supplementary material, which is available to authorized users.

Spatially-variant image-based modeling of PSF deformations with application to a limited angle geometry from a dual-panel breast-PET imager

Dual-panel PET system configuration can lead to spatially variable point-spread functions (PSF) of considerable deformations due to depth-of-interaction effects and limited angular coverage. If not modelled properly, these effects result in decreased and inconsistent recovery of lesion activity across the field-of-view (FOV), as well as mispositioning of lesions in the reconstructed image caused by strong PSF asymmetries. We implemented and evaluated models of such PSF deformations with spatially-variant image-based resolution modeling (IRM) within reconstruction (varRM) using the Direct Image REConstruction for Time-of-flight (DIRECT) method and within post-reconstruction deconvolution methods. In addition, DIRECT reconstruction was performed with a spatially-invariant IRM (invRM) and without resolution modeling (noRM) for comparison. The methods were evaluated using simulated data for a realistic breast model with a set of 5 mm lesions located throughout the FOV of a dual-panel Breast-PET scanner. We simulated high-count data to focus on the ability of each method to correctly recover the PSF deformations, and a clinically realistic count level to assess the impact of low count data on the quantitative performance of the evaluated techniques. Performance of the methods evaluated herein was assessed by comparing lesion activity recovery (%BIAS), consistency (%SD) across the FOV, overall error (%RMSE), and recovery of each lesion location. As expected, all techniques using IRM provide considerable improvement over the noRM reconstruction. For the high-count cases, the overall quantitative performance of all IRM techniques, whether within reconstruction or within post-reconstruction, is similar if the lesion location misplacements are ignored. However, invRM provides less consistent performance on activity across lesions and is not able to recover accurate lesion locations. For a clinically realistic count level, varRM reconstruction consistently outperforms all compared approaches, while the post-reconstruction IRM approaches exhibit higher %SD and %RMSE values due to being more affected by the data noise than the within-reconstruction IRM approaches.