Diagnostic performance of a novel deep learning attenuation correction software for MPI using a cardio dedicated CZT camera. Experience in the clinical practice.

PURPOSE: To evaluate the diagnostic performance of a novel deep learning attenuation correction software (DLACS) for myocardial perfusion imaging (MPI) using a cadmium-zinc-telluride (CZT) cardio dedicated camera with invasive coronary angiography (ICA) correlation for the diagnosis of coronary artery disease (CAD) in a high-risk population. METHODS: Retrospective study of 300 patients (196 males [65%], mean age 68 years) from September 2014 to October 2019 undergoing MPI, followed by ICA and evaluated by means of quantitative angiography software, within six months after the MPI. The mean pre-test probability score for coronary disease according to the European Society of Cardiology criteria was 37% for the whole cohort. The MPI was performed in a dedicated CZT cardio camera (D-SPECT Spectrum Dynamics) with a two-day protocol, according to the European Association of Nuclear Medicine guidelines. MPI was retrospectively evaluated with and without the DLACS. RESULTS: The overall diagnostic accuracy of MPI without DLACS to identify patients with any obstructive CAD at ICA was 87%, sensitivity 94%, specificity 57%, Positive Predictive Value 91% and Negative Predictive Value 64%. Using DLACS the overall diagnostic accuracy was 90%, sensitivity 91%, specificity 86%, Positive Predictive Value 97% and Negative Predictive Value 66%. CONCLUSION: Use of the novel DLACS enhances performance of the MPI using the CZT D-SPECT camera and achieves improved results, especially avoiding artefacts and reducing the number of false positive results.

Adopting transfer learning for neuroimaging: a comparative analysis with a custom 3D convolution neural network model.

BACKGROUND: In recent years, neuroimaging with deep learning (DL) algorithms have made remarkable advances in the diagnosis of neurodegenerative disorders. However, applying DL in different medical domains is usually challenged by lack of labeled data. To address this challenge, transfer learning (TL) has been applied to use state-of-the-art convolution neural networks pre-trained on natural images. Yet, there are differences in characteristics between medical and natural images, also image classification and targeted medical diagnosis tasks. The purpose of this study is to investigate the performance of specialized and TL in the classification of neurodegenerative disorders using 3D volumes of 18F-FDG-PET brain scans. RESULTS: Results show that TL models are suboptimal for classification of neurodegenerative disorders, especially when the objective is to separate more than two disorders. Additionally, specialized CNN model provides better interpretations of predicted diagnosis. CONCLUSIONS: TL can indeed lead to superior performance on binary classification in timely and data efficient manner, yet for detecting more than a single disorder, TL models do not perform well. Additionally, custom 3D model performs comparably to TL models for binary classification, and interestingly perform better for diagnosis of multiple disorders. The results confirm the superiority of the custom 3D-CNN in providing better explainable model compared to TL adopted ones.

A 3D deep learning model to predict the diagnosis of dementia with Lewy bodies, Alzheimer's disease, and mild cognitive impairment using brain 18F-FDG PET.

PURPOSE: The purpose of this study is to develop and validate a 3D deep learning model that predicts the final clinical diagnosis of Alzheimer's disease (AD), dementia with Lewy bodies (DLB), mild cognitive impairment due to Alzheimer's disease (MCI-AD), and cognitively normal (CN) using fluorine 18 fluorodeoxyglucose PET (18F-FDG PET) and compare model's performance to that of multiple expert nuclear medicine physicians' readers. MATERIALS AND METHODS: Retrospective 18F-FDG PET scans for AD, MCI-AD, and CN were collected from Alzheimer's disease neuroimaging initiative (556 patients from 2005 to 2020), and CN and DLB cases were from European DLB Consortium (201 patients from 2005 to 2018). The introduced 3D convolutional neural network was trained using 90% of the data and externally tested using 10% as well as comparison to human readers on the same independent test set. The model's performance was analyzed with sensitivity, specificity, precision, F1 score, receiver operating characteristic (ROC). The regional metabolic changes driving classification were visualized using uniform manifold approximation and projection (UMAP) and network attention. RESULTS: The proposed model achieved area under the ROC curve of 96.2% (95% confidence interval: 90.6-100) on predicting the final diagnosis of DLB in the independent test set, 96.4% (92.7-100) in AD, 71.4% (51.6-91.2) in MCI-AD, and 94.7% (90-99.5) in CN, which in ROC space outperformed human readers performance. The network attention depicted the posterior cingulate cortex is important for each neurodegenerative disease, and the UMAP visualization of the extracted features by the proposed model demonstrates the reality of development of the given disorders. CONCLUSION: Using only 18F-FDG PET of the brain, a 3D deep learning model could predict the final diagnosis of the most common neurodegenerative disorders which achieved a competitive performance compared to the human readers as well as their consensus.

18Fluorodeoxyglucose uptake in relation to fat fraction and R2* in atherosclerotic plaques, using PET/MRI: a pilot study.

Inflammation inside Atherosclerotic plaques represents a major pathophysiological process driving plaques towards rupture. Pre-clinical studies suggest a relationship between lipid rich necrotic core, intraplaque hemorrhage and inflammation, not previously explored in patients. Therefore, we designed a pilot study to investigate the feasibility of assessing the relationship between these plaque features in a quantitative manner using PET/MRI. In 12 patients with high-grade carotid stenosis the extent of lipid rich necrotic core and intraplaque hemorrhage was quantified from fat and R2* maps acquired with a previously validated 4-point Dixon MRI sequence in a stand-alone MRI. PET/MRI was used to measure 18F-FDG uptake. T1-weighted images from both scanners were used for registration of the quantitative Dixon data with the PET images. The plaques were heterogenous with respect to their volumes and composition. The mean values for the group were as follows: fat fraction (FF) 0.17% (± 0.07), R2* 47.6 s-1 (± 10.9) and target-to-blood pool ratio (TBR) 1.49 (± 0.48). At group level the correlation between TBR and FFmean was - 0.406, p 0.19 and for TBR and R2*mean 0.259, p 0.42. The lack of correlation persisted when analysed on a patient-by-patient basis but the study was not powered to draw definitive conclusions. We show the feasibility of analysing the quantitative relationship between lipid rich necrotic cores, intraplaque haemorrhage and plaque inflammation. The 18F-FDG uptake for most patients was low. This may reflect the biological complexity of the plaques and technical aspects inherent to 18F-FDG measurements.Trial registration: ISRCTN, ISRCTN30673005. Registered 05 January 2021, retrospectively registered.

In vitro measurement of glucose uptake after radiation and cetuximab treatment in head and neck cancer cell lines using 18F-FDG, gamma spectrometry and PET/CT.

The standard treatment for head and neck squamous cell carcinoma (HNSCC) is radiotherapy, often in combination with chemotherapy or surgery. However, a novel monoclonal antibody, cetuximab (Erbitux®), has also been approved for patient therapy. The aim of present study was to develop an in vitro method for the measurement of 18F-fluoro-2deoxy-D-glucose (FDG) to determine if cellular 18F-FDG uptake is associated with response to radiotherapy or cetuximab treatment. In the current study, HNSCC cell lines were treated with radiation or with cetuximab. Next, the uptake of 18F-FDG was measured using a gamma spectrometer (GS). Thereafter, uptake after radiation was measured first with GS and then compared with positron emission tomography (PET)/computed tomography (CT) imaging. Furthermore, the mRNA expression of glucose transporter 1 (GLUT1) was measured following cetuximab treatment via reverse transcription-quantitative PCR. A study protocol was developed to measure the cellular uptake of 18F-FDG via gamma-ray spectrometry and comparable results were obtained with those of clinical PET/CT. The results revealed a decrease in 18F-FDG after radiation and cetuximab treatment. The uptake of 18F-FDG following cetuximab treatment was significantly lower in the cetuximab-sensitive cell line UT-SCC-14 compared with the cetuximab-resistant cell lines UT-SCC-2 and UT-SCC-45. Furthermore, after treatment with cetuximab for 24 and 48 h, a significant increase in GLUT1 expression was detected in the sensitive cell line compared with the two resistant cell lines. In conclusion, a novel yet reliable method for the measurement of intracellular 18F-FDG via GS has been developed, and our results indicate that 18F-FDG uptake is associated with radiation and cetuximab response in HNSCC.

Shining dead bone-cause for cautious interpretation of [18F]NaF PET scans.

Background and purpose - [18F]Fluoride ([18F]NaF) PET scan is frequently used for estimation of bone healing rate and extent in cases of bone allografting and fracture healing. Some authors claim that [18F]NaF uptake is a measure of osteoblastic activity, calcium metabolism, or bone turnover. Based on the known affinity of fluoride to hydroxyapatite, we challenged this view. Methods - 10 male rats received crushed, frozen allogeneic cortical bone fragments in a pouch in the abdominal wall on the right side, and hydroxyapatite granules on left side. [18F]NaF was injected intravenously after 7 days. 60 minutes later, the rats were killed and [18F]NaF uptake was visualized in a PET/CT scanner. Specimens were retrieved for micro CT and histology. Results - MicroCT and histology showed no signs of new bone at the implant sites. Still, the implants showed a very high [18F]NaF uptake, on a par with the most actively growing and remodeling sites around the knee joint. Interpretation - [18F]NaF binds with high affinity to dead bone and calcium phosphate materials. Hence, an [18F]NaF PET/CT scan does not allow for sound conclusions about new bone ingrowth into bone allograft, healing activity in long bone shaft fractures with necrotic fragments, or remodeling around calcium phosphate coated prostheses.

Contrast biases the autocorrelation phase shift estimation in Doppler tissue imaging.

Quantitative assessment of regional myocardial function at rest and during stress with Doppler tissue imaging (DTI) plays an important role in daily routine echocardiography. However, reliable visual analysis is largely dependent on image quality and adequate border delineation, which still remains a challenge in a significant number of patients. In this respect, an ultrasound contrast agent (UCA) is often used to improve visualization in patients with suboptimal image quality. The knowledge of how DTI measurements will be affected by UCA present in the tissue is therefore of significant importance for an accurate interpretation of local myocardial motion. The aim of this paper was to investigate how signal contribution from UCA and nonlinear wave propagation influence the performance of the autocorrelation phase shift estimator used for DTI applications. Our results are based on model experiments with a clinical 2-D grayscale scanner and computational simulations of the DTI velocity estimator for synthetically-derived pulses, simulated bubble echoes and experimentally-sampled RF data of transmitted pulses and backscattered contrast echoes. The results show that destruction of UCA present in the tissue will give rise to an apparent bidirectional velocity bias of individual velocity estimates, but that spatial averaging of individual velocity measurements within a region-of-interest will result in a negative bias (away from the transducer) of the estimated mean or mean peak velocity. The UCA destruction will also have a significant impact on the measured integrated mean velocity over time, i.e., displacement. To achieve improved visualization with UCA during DTI-examinations, we either recommend that it is performed at low acoustic powers, mechanical index

Effects of ultrasound contrast agents on Doppler tissue velocity estimation.

The combination of Doppler tissue imaging and myocardial contrast echocardiography has the potential to provide information about motion and perfusion of the myocardium in a single examination. The purpose of this study was to establish how the presence of ultrasound contrast agent (UCA) affects measurements of Doppler tissue velocities in vivo and in vitro. We performed echocardiography in 12 patients with ischemic heart disease before and immediately after a slow intravenous infusion of the UCA Optison, using color Doppler tissue imaging to examine the effect of contrast agents in vivo. The myocardial peak systolic velocities and their integrals were analyzed in digitally stored cineloops before and after contrast administration. To distinguish between methodologic and physiologic factors affecting the measurement of tissue velocity in vitro, experiments with a rotating disk and a flow cone phantom were also carried out for the 3 contrast agents: Optison, Sonovue, and Sonazoid. In vivo results show that the values for peak systolic velocity increased by about 10% during contrast infusion, from mean 5.2 +/- 1.8 to 5.7 +/- 2.3 cm/s (P = .02, 95% confidence interval 2%-16%). The increase in myocardial peak systolic velocities was verified in experimental models in which the UCA increased the estimated mean velocity in the order of 5% to 20% for the motion interval of 5 to 7 cm/s, corresponding to the myocardial velocities studied in vivo. The response was similar for all 3 contrast agents and was not affected by moderate variations in concentration of the agent. We have shown that the presence UCA will affect Doppler tissue measurements in vivo and in vitro. The observed bias is presumed to be an effect of harmonic signal contribution from rupturing contrast agent microbubbles and does not indicate biologic or physiologic effects.

Absolute quantification of human liver metabolite concentrations by localized in vivo 31P NMR spectroscopy in diffuse liver disease.

Phosphorus-31 NMR spectroscopy using slice selection (DRESS) was used to investigate the absolute concentrations of metabolites in the human liver. Absolute concentrations provide more specific biochemical information compared to spectrum integral ratios. Nine patients with histopathologically proven diffuse liver disease and 12 healthy individuals were examined in a 1.5-T MR scanner (GE Signa LX Echospeed plus). The metabolite concentration quantification procedures included: (1) determination of optimal depth for the in vivo measurements, (2) mapping the detection coil characteristics, (3) calculation of selected slice and liver volume ratios using simple segmentation procedures and (4) spectral analysis in the time domain. The patients had significantly lower concentrations of phosphodiesters (PDE), 6.3+/-3.9 mM, and ATP-beta, 3.6+/-1.1 mM, (P<0.05) compared with the control group (10.0+/-4.2 mM and 4.2+/-0.3 mM, respectively). The concentrations of phosphomonoesters (PME) were higher in the patient group, although this was not significant. Constructing an anabolic charge (AC) based on absolute concentrations, [PME]/([PME] + [PDE]), the patients had a significantly larger AC than the control subjects, 0.29 vs. 0.16 (P<0.005). Absolute concentration measurements of phosphorus metabolites in the liver are feasible using a slice selective sequence, and the technique demonstrates significant differences between patients and healthy subjects.

Modelling of nonlinear effects and the response of ultrasound contrast micro bubbles: simulation and experiment.

The propagation of diagnostic ultrasonic imaging pulses in tissue and their interaction with contrast micro bubbles is a very complex physical process, which we assumed to be separable into three stages: pulse propagation in tissue, the interaction of the pulse with the contrast bubble, and the propagation of the scattered echo. The model driven approach is used to gain better knowledge of the complex processes involved. A simplified way of field simulation is chosen due to the complexity of the task and the necessity to estimate comparative contributions of each component of the process. Simulations are targeted at myocardial perfusion estimation. A modified method for spatial superposition of attenuated waves enables simulations of low intensity pulse pressure fields from weakly focused transducers in a nonlinear, attenuating, and liquid-like biological medium. These assumptions enable the use of quasi-linear calculations of the acoustic field. The simulations of acoustic bubble response are carried out with the Rayleigh-Plesset equation with the addition of radiation damping. Theoretical simulations with synthesised and experimentally sampled pulses show that the interaction of the excitation pulses with the contrast bubbles is the main cause of nonlinear scattering, and a 2-3 dB increase of second harmonic amplitude depends on nonlinear distortions of the incident pulse.

Backscattered ultrasound from contrast microbubbles: effects of tissue and bubble interaction.

The propagation of diagnostic ultrasonic imaging pulses in tissue and their interaction with contrast microbubbles is a complex physical process. Our model driven approach is used to gain better knowledge of the different processes involved in the generation of the backscattered contrast echo. It can be divided into three separable stages: linear and nonlinear wave propagation in tissue, the resulting echo from the pulse interaction with the contrast microbubble, and the propagation of the scattered echo. A simplified approach of field simulation is chosen due to the complexity of the task and necessity to estimate comparative contributions of each component of the process. A modified method for spatial superposition of attenuated waves was further developed to enable simulations of low intensity pulse fields in nonlinear attenuating and liquid-like biological medium using weakly focused transducers. Simulations of the acoustic bubble response are carried out with Rayleigh-Plesset equation with the addition of the radiation damping. Theoretical simulations show that contrast bubbles interaction with excitation pulses is the main cause of nonlinear distortions, and a 2-3 dB increase of second harmonic amplitude depends on nonlinear distortions of incident pulse.