Browsing by Author "Abadi, Ehsan"
Results Per Page
Sort Options
Item Open Access A comparison of COVID-19 and imaging radiation risk in clinical patient populations.(J Radiol Prot, 2020-10-07) Ria, Francesco; Fu, Wanyi; Chalian, Hamid; Abadi, Ehsan; Segars, Paul W; Fricks, Rafael; Khoshpouri, Pegah; Samei, EhsanOBJECTIVE: The outbreak of coronavirus SARS-COV2 affected more than 180 countries necessitating fast and accurate diagnostic tools. Reverse transcriptase polymerase chain reaction (RT-PCR) has been identified as a gold standard test with Chest CT and Chest Radiography showing promising results as well. However, radiological solutions have not been used extensively for the diagnosis of COVID-19 disease, partly due to radiation risk. This study aimed to provide quantitative comparison of imaging radiation risk versus COVID risk. METHODS: The analysis was performed in terms of mortality rate per age group. COVID-19 mortality was extracted from epidemiological data across 299,004 patients published by ISS-Integrated surveillance of COVID-19 in Italy. For radiological risk, the study considered 659 Chest CT performed in adult patients. Organ doses were estimated using a Monte Carlo method and then used to calculate Risk Index that was converted into an upper bound for related mortality rate following NCI-SEER data. RESULTS: COVID-19 mortality showed a rapid rise for ages >30 years old (min:0.30%; max:30.20%), whereas only 1 death was reported in the analyzed patient cohort for ages <20 years old. The rates decreased for radiation risk across age groups. The median mortality rate across all ages for Chest-CT and Chest-Radiography were 0.007% (min:0.005%; max:0.011%) and 0.0003% (min:0.0002%; max:0.0004%), respectively. CONCLUSIONS: COVID-19, Chest Radiography, and Chest CT mortality rates showed different magnitudes and trends across age groups. In higher ages, the risk of COVID-19 far outweighs that of radiological exams. Based on risk comparison alone, Chest Radiography and CT for COVID-19 care is justified for patients older than 20 and 30 years old, respectively. Notwithstanding other aspects of diagnosis, the present results capture a component of risk consideration associated with the use of imaging for COVID. Once integrated with other diagnostic factors, they may help inform better management of the pandemic.Item Open Access A patient-informed approach to predict iodinated-contrast media enhancement in the liver(European Journal of Radiology, 2022-10) Setiawan, Hananiel; Chen, Chaofan; Abadi, Ehsan; Fu, Wanyi; Marin, Daniele; Ria, Francesco; Samei, EhsanItem Open Access A systematic assessment and optimization of photon-counting CT for lung density quantifications.(Medical physics, 2024-02) Sotoudeh-Paima, Saman; Segars, W Paul; Ghosh, Dhrubajyoti; Luo, Sheng; Samei, Ehsan; Abadi, EhsanBackground
Photon-counting computed tomography (PCCT) has recently emerged into clinical use; however, its optimum imaging protocols and added benefits remains unknown in terms of providing more accurate lung density quantification compared to energy-integrating computed tomography (EICT) scanners.Purpose
To systematically assess the performance of a clinical PCCT scanner for lung density quantifications and compare it against EICT.Methods
This cross-sectional study involved a retrospective analysis of subjects scanned (August-December 2021) using a clinical PCCT system. The influence of altering reconstruction parameters was studied (reconstruction kernel, pixel size, slice thickness). A virtual CT dataset of anthropomorphic virtual subjects was acquired to demonstrate the correspondence of findings to clinical dataset, and to perform systematic imaging experiments, not possible using human subjects. The virtual subjects were imaged using a validated, scanner-specific CT simulator of a PCCT and two EICT (defined as EICT A and B) scanners. The images were evaluated using mean absolute error (MAE) of lung and emphysema density against their corresponding ground truth.Results
Clinical and virtual PCCT datasets showed similar trends, with sharper kernels and smaller voxel sizes increasing percentage of low-attenuation areas below -950 HU (LAA-950) by up to 15.7 ± 6.9% and 11.8 ± 5.5%, respectively. Under the conditions studied, higher doses, thinner slices, smaller pixel sizes, iterative reconstructions, and quantitative kernels with medium sharpness resulted in lower lung MAE values. While using these settings for PCCT, changes in the dose level (13 to 1.3 mGy), slice thickness (0.4 to 1.5 mm), pixel size (0.49 to 0.98 mm), reconstruction technique (70 keV-VMI to wFBP), and kernel (Qr48 to Qr60) increased lung MAE by 15.3 ± 2.0, 1.4 ± 0.6, 2.2 ± 0.3, 4.2 ± 0.8, and 9.1 ± 1.6 HU, respectively. At the optimum settings identified per scanner, PCCT images exhibited lower lung and emphysema MAE than those of EICT scanners (by 2.6 ± 1.0 and 9.6 ± 3.4 HU, compared to EICT A, and by 4.8 ± 0.8 and 7.4 ± 2.3 HU, compared to EICT B). The accuracy of lung density measurements was correlated with subjects' mean lung density (p < 0.05), measured by PCCT at optimum setting under the conditions studied.Conclusion
Photon-counting CT demonstrated superior performance in density quantifications, with its influences of imaging parameters in line with energy-integrating CT scanners. The technology offers improvement in lung quantifications, thus demonstrating potential toward more objective assessment of respiratory conditions.Item Open Access Accuracy of Noise Magnitude Measurements from Patient CT Images(https://aapm.onlinelibrary.wiley.com/doi/epdf/10.1002/mp.16525, 2023-07-23) Ria, Francesco; Setiawan, Hananiel; Abadi, Ehsan; Samei, EhsanPurpose Noise magnitude is a main CT image quality indicator. In vivo measurements emerged as a patient-specific methodology to assess and qualify CT noise, yet methods to do so vary. Current noise measurement methods in soft tissues and air surrounding the patient use distinct image segmentations, HU thresholds, and region-of-interests, resulting in noise estimation variations. In this study, we compared two noise magnitude calculation methods against the gold standard ensemble noise in two cohorts of virtually-generated patient images across 36 imaging conditions. Methods 1800 image datasets were generated using a virtual trial platform based on anthropomorphic phantoms (XCAT) and a validated, scanner-specific CT simulator (DukeSim). XCAT phantoms were repeatedly imaged 50 times using Chest and Abdominopelvic protocols, three dose levels, and three reconstruction kernels, using both FBP and IR algorithms. Noise magnitudes were calculated in the air surrounding the patient (An) and soft tissues (GNI) by applying HU<-900 and -300Item Open Access Classification of COVID-19 in chest radiographs: assessing the impact of imaging parameters using clinical and simulated images(Medical Imaging 2021: Computer-Aided Diagnosis, 2021-02-15) Fricks, Rafael; Abadi, Ehsan; Ria, Francesco; Samei, EhsanAs computer-aided diagnostics develop to address new challenges in medical imaging, including emerging diseases such as COVID-19, the initial development is hampered by availability of imaging data. Deep learning algorithms are particularly notorious for performance that tends to improve proportionally to the amount of available data. Simulated images, as available through advanced virtual trials, may present an alternative in data-constrained applications. We begin with our previously trained COVID-19 x-ray classification model (denoted as CVX) that leveraged additional training with existing pre-pandemic chest radiographs to improve classification performance in a set of COVID-19 chest radiographs. The CVX model achieves demonstrably better performance on clinical images compared to an equivalent model that applies standard transfer learning from ImageNet weights. The higher performing CVX model is then shown to generalize effectively to a set of simulated COVID-19 images, both quantitative comparisons of AUCs from clinical to simulated image sets, but also in a qualitative sense where saliency map patterns are consistent when compared between sets. We then stratify the classification results in simulated images to examine dependencies in imaging parameters when patient features are constant. Simulated images show promise in optimizing imaging parameters for accurate classification in data-constrained applications.Item Open Access Clinical and radiation risk across one million patients in Computed Tomography: influence of age, size, and race(2023-11-26) Ria, Francesco; Lerebours, Reginald; Zhang, Anru; Erkanli, Alaattin; Abadi, Ehsan; SOLOMON, justin; Marin, Daniele; Samei, EhsanPurpose. We recently developed a mathematical model to balance radiation risk and clinical risk, namely the risk of misdiagnosis due to insufficient image quality. In this work, we applied this model to a population of one million CT imaging cases to evaluate the risk stratification with different ages, sexes, and races. Materials and Methods. The demographics were informed by literature and census information simulating a clinical liver cancer population. The Total Risk (TR) was calculated as the linear combination of radiation risk and clinical risk. The model included factors for the radiation burden for different age and sex; the prevalence of the disease; the false positive rate; the expected life-expectancy loss for an incorrect diagnosis for different ages, sex, and race; and a typical false positive rate of 5%. It was assumed that each case received an average radiologist interpretative performance of 0.75 AUC for a hypothetical lesion without any changes in radiation dose beyond routine practice. We further, for each patient, simulated 2,000 imaging conditions with CTDIvol varying from 0.1 and 200 mGy with 0.1 mGy increments. Per each CTDIvol value, the anticipated AUC was calculated by applying the established asymptotic relationships between CTDIvol and image quality. The AUC distribution was then used to calculate the theoretical minimum total risk (TRmin) per each patient. Results. For the routine practice, the median theoretical total risk was estimated to be 0.058 deaths per 100 patients (range: 0.002 – 0.154) comprising of the median radiation risk of 0.009 (range: 0.001 – 0.069), and of the median clinical risk of 0.049 (range: 7.0x10-5 – 0.094). Considering the varying scanner output conditions, the median TRmin was 0.054 deaths per 100 patients for White male patients, 0.054 for Blacks, 0.057 for Hispanics, and 0.065 for Asians. For female patients, the median TRmin values were 0.049, 0.056, 0.054, and 0.061 deaths per 100 patients, respectively. Conclusion. For each demography condition, the clinical risk was found to largely outweigh the radiation risk by at least 500%. Total risk showed different stratifications with patient age and race. Clinical Relevance Statement. To optimize CT conditions for specific patients and/or population, both radiation risk and clinical risks should be all accounted for together with demographic information. We demonstrated a methodology that allows a complete depiction of total risk in CT, considering radiation and clinical risks at comparable units, and patient demographic.Item Open Access Comparative Risk Assessment of Clinical and Radiation Risk across a Cohort of Patient and Individualized Risk Optimization(https://aapm.onlinelibrary.wiley.com/doi/epdf/10.1002/mp.16525, 2023-07-23) Ria, Francesco; Lerebours, Reginald; Zhang, Anru; Erkanli, Alaattin; Abadi, Ehsan; SOLOMON, justin; Marin, daniele; Samei, EhsanPurpose Informed by a recent mathematical framework, we formulated an imaging strategy to balance interpretative performance-based clinical risk (i.e., false positive and false negative rates) and radiation risk as a risk-versus-risk assessment. The model was applied to a population of one million cases simulating a clinical liver cancer scenario. Moreover, a model was developed to predict individualized risk-versus-risk optimization. Methods The proposed model defined a Total Risk (TR) as the linear combination of radiation risk and clinical risk defined as functions of the radiation burden, the disease prevalence disease, the false positive rate, the expected life-expectancy loss for an incorrect diagnosis, and the radiologist interpretative performance (i.e., AUC). The mathematical framework was applied to a simulated dataset of 1,000,000 CT studies investigating localized stage liver cancer assuming a typical false positive rate of 5% and optimal imaging conditions (AUC=0.75). Demographic information was simulated according with literature and census data including male and female for different patient races (white, black, Asian, and Hispanic). Following BEIR-VII report, organ-specific radiation doses were used to calculate the radiation Risk Index per each patient. The model was then extended to predict the optimal scanner output associated with the TR for specific patients. Results Across all races and sexes, median radiation risk ranged between 0.008 and 0.012 number of deaths per 100 patients; median clinical risk ranged between 0.042 and 0.076; and medial total risk ranged between 0.010 and 0.088 deaths per 100 patients. The mathematical model was then generalized to estimate individualized optimal imaging condition minimizing TR. Conclusion A mathematical framework to describe total risk in CT was robustly tested in a simulated dataset of 1,000,000 CT studies. The results highlighted the dominance of clinical risk at typical CT examination dose levels. The generalization of the mathematical model allowed the prediction of individualized risk optimization.Item Open Access Deep learning classification of COVID-19 in chest radiographs: performance and influence of supplemental training(Journal of Medical Imaging, 2021-12-01) Fricks, Rafael B; Ria, Francesco; Chalian, Hamid; Khoshpouri, Pegah; Abadi, Ehsan; Bianchi, Lorenzo; Segars, William P; Samei, EhsanItem Open Access Development and Application of Realistic Anatomical and Imaging Models for Virtual Clinical Trials in Computed Tomography(2018) Abadi, EhsanThe purpose of this dissertation was to develop comprehensive toolsets for performing quality based virtual clinical trials in computed tomography. The developed toolsets in this dissertation enable rigorous quantification and evaluation of computed tomography scanners which not possible using ground-truth limited clinical trials or simplistic physical phantoms. This projection was outlined in three sections: 1) modeling high-resolution human models with intra-organ heterogeneities, 2) modeling a scanner-specific and realistic computed tomography simulator, and 3) performing a virtual clinical trial to evaluate and optimize geometrical imaging parameters in computed tomography.
In chapter 2, an anatomically-informed mathematical model was developed to extend the non-parenchyma structures, including airways, arteries, and veins to the level of terminal branches while avoiding intersections. A geometrical validation was done to ensure that the generated lung models have anatomical attributes close to morphometry studies. Additionally, a texture synthesis algorithm, informed by a high-resolution lung specimen, was used to develop a heterogenous parenchyma background within the lung regions.
In chapter 3, an algorithm, informed by a high-resolution bone dataset, was developed to model the trabecular and cortical bone within the human models. In chapter 4, a realistic and scanner-specific energy-integrating computed tomography simulator, DukeSim, was developed to synthesize projection images of the high-resolution human models developed in chapters 2 and 3. DukeSim calculates projection images using a combination of ray-tracing and Monte Carlo techniques. It accounts for the geometry and physics of a specific scanner. To validate DukeSim, clinical and simulated computed tomography scans of a phantom was imaged and quantitatively compared against each other. The results showed that DukeSim is capable of simulating computed tomography images with image quality metrics close to clinical images.
In chapter 5, DukeSim was extended to synthesize photon-counting projection images. Similar to chapter 4, the photon-counting feature of the DukeSim was validated by comparing the quality of the real and simulated images. The results showed that DukeSim is capable of simulating photon-counting computed tomography images with image quality metrics close to real images.
In chapter 6, the developed human and imaging models were integrated to perform a virtual clinical trial. The purpose of this chapter was to investigate the effects of beam collimation and pitch on image quality in computed tomography under different respiratory and cardiac motion levels. A realistic human model with added lung lesions was used to cardiac and respiratory motions. Each case was imaged using DukeSim at multiple pitches and beam collimations. The images were compared against the known ground truth using task-generic and task-specific metrics. All task-generic metrics degraded by increasing pitch. When imaged with motion, increasing pitch reduced motion artifacts. The image quality metrics remained largely unchanged with changes in beam collimations studied. Patient motion exhibited negative effects on the image quality metrics. The study concluded that while desirable for fast imaging, high pitch and large beam collimations can negatively affect image quality of CT images.
In conclusion, this dissertation provides a set of realistic toolsets that can be used to study, investigate, and optimize computed tomography technology and protocols in with a known-ground-truth, in a cost-effective manner, and without any radiation safety concerns.
Item Open Access Development and Testing of a Clinical Tool to Predict and Optimize Liver Contrast-Enhanced CT Imaging(https://aapm.onlinelibrary.wiley.com/doi/epdf/10.1002/mp.16525, 2023-07-23) Setiawan, Hananiel; Abadi, Ehsan; Marin, Daniele; Molvin, Lior; Ria, Francesco; Samei, EhsanAchieving consistent and sufficient hepatic parenchyma contrast enhancement (HPCE) level can improve diagnostic performance and reduce enhancement variability; this raises the baseline image quality and optimize injection practices, both carries economic and safety implications. Patient factors, Iodine injection and scanning parameters (e.g. tube potential, scanning delay) affect HPCE in CT imaging. In this study, we developed and prospectively tested a clinical graphical user interface (GUI) tool which predicts enhancement level and suggests alternative injection/scanning parameters based on patient attributes (height, weight, sex, age). Methods: This work was based on our retrospectively-validated neural-network prediction model. We built a GUI to combine our model with an optimization algorithm, which suggests alternative injection/scanning parameters for patients with predicted-insufficient enhancement. The system was clinically-deployed and prospectively-tested in 24 patients considering a 110HU+/-10HU target portal-venous HPCE. For each patient, HPCE was calculated as the average HU-value of three ROIs and compared against the target value. Additionally, we compared the outcome with the patient’s previous similarly-protocoled scan to assess improvement and consistency. Results: The system suggested adjustment for 15 patients with median 8.8% and 9.1% reductions to volume and injection rate, respectively. All scan delays were reduced by an average of 42.6%. Comparison with previous scans shows increased consistency (CV=0.21 v. 0.11,p=0.012) while median enhancement remained relatively unchanged (111.3HU v. 108.7HU). The number of under-enhanced patients was halved, and all previously over-enhanced patients received enhancement reductions. Conclusion: We developed and tested a patient-informed clinical framework which predicts optimal patient’s HPCE; and suggests empiric injection/scanning parameters when predicted enhancement is deemed insufficient. The system improved HPCE consistency and decreased the number of under-enhanced patients as compared to their previous scans. This study demonstrated that the patient-informed clinical framework can predict an optimal patient's HPCE and suggest empiric injection/scanning parameters to achieve consistent and sufficient HPCE levels.Item Open Access Estimation of in vivo noise in clinical CT images: comparison and validation of three different methods against ensemble noise gold-standard(Proc. SPIE 11595, Medical Imaging 2021: Physics of Medical Imaging, 115952P, 2021-02-15) Ria, Francesco; Smith, Taylor; Abadi, Ehsan; Solomon, Justin; Samei, EhsanImage quality estimation is crucial in modern CT with noise magnitude playing a key role. Several methods have been proposed to estimate noise surrogates in vivo. This study aimed to ascertain the accuracy of three different noise-magnitude estimation methods. We used ensemble noise as the ground truth. The most accurate approach to assess ensemble noise is to scan a patient repeatedly and assess the noise for each pixel across the ensemble of images. This process is ethically undoable on actual patients. In this study, we surmounted this impasse using Virtual Imaging Trials (VITs) that simulate clinical scenarios using computer-based simulations. XCAT phantoms were imaged 47 times using a scanner-specific simulator (DukeSim) and reconstructed with filtered back projection (FBP) and iterative (IR) algorithms. Noise magnitudes were calculated in lung (ROIn), soft tissues (GNI), and air surrounding the patient (AIRn), applying different HU thresholds and techniques. The results were compared with the ensemble noise magnitudes within soft tissue (En). For the FBP-reconstructed images, median En was 30.6 HU; median ROIn was 46.6 HU (+52%), median GNI was 40.1 HU (+31%), and median AIRn 25.1 HU (-18%). For the IR images, median En was 19.5 HU; median ROIn was 31.2 HU (+60%), median GNI was 25.1 HU (+29%), and median AIRn 18.8 HU (-4%). Compared to ensemble noise, GNI and ROIn overestimate the tissue noise, while AIRn underestimates it. Air noise was least representative of variations in tissue noise due to imaging condition. These differences may be applied as adjustment or calibration factors to better represent clinical results.Item Open Access Impact of CT Simulation Parameters on the Realism of Virtual Imaging Trials(2023) Montero, Isabel SeraphinaVirtual imaging trials (VITs) provide the opportunity to conduct medical imaging experiments otherwise not feasible through patient images. The reliability of these virtual trials is directly dependent upon their ability to replicate clinical imaging experiments. The combined effect of various key simulation parameters on the closeness of virtual images to experimental images has not yet been explicitly quantified, which this sensitivity study aimed to address. To do so, a physical phantom, Mercury 3.0 (Sun Nuclear), was scanned using a clinical scanner (Siemens Force). Meanwhile, utilizing a validated CT simulator (DukeSim), a computational version of the Mercury 3.0 phantom was virtually imaged, emulating the same scanner model and imaging acquisition settings. The simulations were performed with varied parameters for the x-ray source, phantom model, and detector characteristics, evaluating their impact on the realism of the final reconstructed virtual images. Simulations were explicitly conducted and evaluated various source and detector subsampling (1 – 5 per side), phantom voxel resolution (0.1 mm – 0.5mm), anode heel severity (0% - 40% over anode-cathode axis), aluminum filtration (0.9cm - 1.1cm), and pixel-to-pixel detector crosstalk (0 – 10.5%, 0 – 15% per dimension). The real and simulated projections were then reconstructed, employing a vendor-specific reconstruction software (Siemens ReconCT), with identical reconstruction settings. The real and simulated images were then compared in terms of modulation transfer function (MTF), noise magnitude, noise power spectrum (NPS), and CT number accuracy. When the optimal simulation parameters were selected, the simulated images closely replicated real images (0.80% relative error in f50air metric). The error in the f50 measurements were highly sensitive to the variation of source and detector subsampling and phantom voxel size. The relative error in the noise magnitude measurements were not highly sensitive to the variation of source and detector subsampling or phantom voxel size but were sensitive to the modeling of the anode heel effect severity. The error in the nNPS measurements were not highly sensitive to the variation of source and detector subsampling, phantom voxel size, degree of anode heel severity, aluminum filtration, or detector cross talk. Finally, the error in the CT number accuracy measurements were not highly sensitive to the variation of source and detector subsampling, phantom voxel size, aluminum filtration, or degree of detector cross talk, but were sensitive to the modeling of anode heel severity. Through this study, the effects of various key simulation parameters on the realism of scanner-specific simulations were assessed. Certain simulation parameters, such as source and detector subsampling, and degree of anode heel severity, exert greater influence on simulation realism than others, thus they should be prioritized when exploring novel modeling avenues.
Item Open Access Implementation and Applications of Virtual Imaging Trials in Computed Tomography(2023) Felice, NicholasVirtual imaging is a powerful tool that offers an alternative to clinical trials by simulating medical image acquisitions. Virtual trials allow for the generation of large data sets without the need for human subjects, and generated images can be compared to ground truth which is not known in clinical trials. The goal of this thesis is to build upon prior work in developments of virtual trials and utilize them for cardiac and abdominal CT imaging applications. In particular, integration tools were developed to enable automatic CT simulation for large scale trials. With these tools, a pilot study was conducted to investigate the effects of image acquisition parameters on cardiac CT quantifications, laying the groundwork for a comprehensive evaluation. Further, a virtual trial was performed to objectively evaluate photon-counting CT in terms of the image quality improvements and liver lesion detectability. This study provided valuable information about how photon-counting CT can be effectively implemented in the clinic. These studies demonstrated the viability of virtual trials for medical imaging research.
Item Open Access Modeling Patient-Informed Liver Contrast Perfusion in Contrast-enhanced Computed Tomography.(J Comput Assist Tomogr, 2020-11) Setiawan, Hananiel; Ria, Francesco; Abadi, Ehsan; Fu, Wanyi; Smith, Taylor B; Samei, EhsanOBJECTIVE: To determine the correlation between patient attributes and contrast enhancement in liver parenchyma and demonstrate the potential for patient-informed prediction and optimization of contrast enhancement in liver imaging. METHODS: The study included 418 chest/abdomen/pelvis computed tomography scans, with 75% to 25% training-testing split. Two regression models were built to predict liver parenchyma contrast enhancement over time: first model (model A) utilized patient attributes (height, weight, sex, age, bolus volume, injection rate, scan times, body mass index, lean body mass) and bolus-tracking data. A second model (model B) only used the patient attributes. Pearson coefficient was used to assess predictive accuracy. RESULTS: Weight- and height-related features were found to be statistically significant predictors (P < 0.05), weight being the strongest. Of the 2 models, model A (r = 0.75) showed greater accuracy than model B (r = 0.42). CONCLUSIONS: Patient attributes can be used to build prediction model for liver parenchyma contrast enhancement. The model can have utility in optimization and improved consistency in contrast-enhanced liver imaging.Item Open Access Patient-informed modelling of hepatic contrast dynamics in contrast-enhanced CT imaging(Medical Imaging 2020: Physics of Medical Imaging, 2020-03-16) Setiawan, Hananiel; Ria, Francesco; Abadi, Ehsan; Fu, Wanyi; Smith, Taylor; Samei, EhsanPURPOSE Iodinated contrast agents are commonly used in CT imaging to enhance tissue contrast. Consistency in contrast enhancement (CE) is critical in radiological diagnosis. Contrast material circulation in individual patients is affected by factors such as patient body habitus and anatomy leading to significant variability in organ contrast enhancement, image quality, and dose. Toward the goal of improving CE consistency in clinical populations, in this work we developed a contrast dynamics model to predict CT HU enhancement of liver parenchyma in abdominopelvic CE CT scans. METHOD AND MATERIALS This study included 700 adult abdominopelvic contrast CT exams performed in 2014-2018 using two scanner models from two vendors. Each CT image was segmented using a deep learning-based segmentation algorithm and the hepatic parenchyma HU values were acquired from the segmentations. A two-layer neural network-based algorithm was used to identify the relationship between patient attributes (height, weight, BMI, age, sex), scan parameters (slice thickness, scanner model), contrast injection protocols (bolus volume, injection-to-scan wait time), and the liver HU CE. We randomly selected 60% studies for training, 10% validation, and 30% for testing the accuracy. The training output was the extracted HU values. The goodness-of-fit of the model was evaluated in terms of R^2, Adjusted R^2, Mean Absolute Error (MAE), and Mean Squared Error (MSE) between the model prediction and ground truth. In addition, the generalizability of the model was evaluated by comparing the R^2 in the training data (leave-one-out validation) and the testing data. RESULTS This preliminary model has an 0.51 R^2, 0.40 adjusted R^2, 10.0 HU MAE, 159.1 HU MSE, 0.6±12.8 HU Mean Error, and 2.5 HU Median Error on test data. For training data, the model has 0.59 R^2, 0.56 Adjusted R^2, and 0.5 predicted R^2. The close R^2 between testing and training data results indicate a reasonable generalizability. CONCLUSION Results showed considerable predictability of liver CE from patient attributes, scanning parameters, and contrast administration protocol. We envision to expand the model to include other major organs toward a comprehensive predictive model. CLINICAL RELEVANCE/APPLICATION A contrast dynamics model can be an essential tool to personalize contrast-enhanced CT protocol and to improve the consistency of contrast enhancement across different patients in diagnostics imaging.Item Open Access Proceedings Virtual Imaging Trials in Medicine 2024.(ArXiv, 2024-05-08) Abadi, Ehsan; Badano, Aldo; Bakic, Predrag; Bliznakova, Kristina; Bosmans, Hilde; Carton, Ann-Katherine; Frangi, Alejandro; Glick, Stephen; Kinahan, Paul; Lo, Joseph; Maidment, Andrew; Ria, Francesco; Samei, Ehsan; Sechopoulos, Ioannis; Segars, Paul; Tanaka, Rie; Vancoillie, LiesbethThis submission comprises the proceedings of the 1st Virtual Imaging Trials in Medicine conference, organized by Duke University on April 22-24, 2024. The listed authors serve as the program directors for this conference. The VITM conference is a pioneering summit uniting experts from academia, industry and government in the fields of medical imaging and therapy to explore the transformative potential of in silico virtual trials and digital twins in revolutionizing healthcare. The proceedings are categorized by the respective days of the conference: Monday presentations, Tuesday presentations, Wednesday presentations, followed by the abstracts for the posters presented on Monday and Tuesday.Item Open Access Scientific Abstracts and Sessions(Medical Physics, 2020-06) Ria, Francesco; Smith, Taylor; Abadi, Ehsan; Solomon, Justin; Samei, ehsanPurpose Image quality estimation in CT is crucial for technology assessment, procedure optimization, and overall radiological benefit evaluation, with noise magnitude playing a key role. Over the years, several methods have been proposed to estimate noise surrogates in vivo. The most accurate approach is to assess ensemble noise by scanning a patient multiple times and sampling each pixel noise within the ensemble of images, an ethically undoable repeated imaging process. Such impasse can be surmounted using Virtual Imaging Trials (VITs) that use computer-based simulations to simulate clinically realistic scenarios. The purpose of this study was to compare two different noise magnitude estimation methods with the ensemble noise measured in a VIT population. Methods This study included a set of 47 XCAT-phantom repeated chest exams acquired virtually using a scanner-specific simulator (DukeSim) modeling a commercial scanner geometry, reconstructed with FBP and IR algorithms. Noise magnitudes were calculated in soft tissues (GNI) and air surrounding the patient (AIRn), applying [-300,100]HU and HU<-900 thresholds, respectively. Furthermore, for each pixel in GNI threshold, the ensemble noise magnitudes in soft tissues (En) were calculated across images. Noise magnitude from different methods were compared in terms of percentage difference with correspondent En median values. Results For FBP reconstructed images, median En was 30.6 HU; median GNI was 40.1 HU (+31%) and median AIRn was 25.1 HU (-18%). For IR images, median En was 19.5 HU; median GNI was 25.1 HU (+29%) and median AIRn was 18.8 HU (-4%). Conclusion Compared to ensemble noise, GNI overestimates the tissue noise by about 30%, while AIRn underestimates it by 4 to 18%, depending on the reconstruction used. These differences may be applied as adjustment or calibration factors to the related noise estimation methods to most closely represent clinical results. However, air noise cannot be assumed to represent tissue noise.Item Open Access Synthesis of 3D Realistic High-resolution Lung Background Textures Using a Conditional Generative Adversarial Network (CGAN)(2022) Wang, YuhaoObjectives: We develop machine-learning based methods to synthesize lung textures within computational phantoms for improved realism in simulating high-resolution patient CT imaging data to evaluate and improve imaging devices and techniques.Methods: We first optimized a previously developed technique designed using a Conditional Generative Adversarial Network (CGAN), Project 1. The optimized model was trained and validated using clinical CT data. Generated texture images were evaluated qualitatively and quantitatively comparing them to the original CT data as well as to results from the previous work. Using what we learned from Project 1, in Project 2, we trained and validated a new generator using high-resolution micro-CT data of the lungs. The new generator was evaluated in a similar fashion. Results: For Project 1, the model was unable to produce results better than the previous work; lung textures were found to be blurry and lacked detail. For Project 2, the trained generator was found capable of simulating variable 3D lung background textures similar to the micro-CT both qualitatively and quantitatively. Conclusion: The CGAN method developed in this work, based on micro-CT data, can greatly improve the realism of computational phantoms by adding high-resolution background textures to the lungs. Such anatomical detail is necessary to evaluate higher-resolution CT imaging methods such as photon-counting CT.
Item Open Access Technology Characterization Through Diverse Evaluation Methodologies: Application to Thoracic Imaging in Photon-Counting Computed Tomography.(J Comput Assist Tomogr, 2024-04-15) Rajagopal, Jayasai R; Schwartz, Fides R; McCabe, Cindy; Farhadi, Faraz; Zarei, Mojtaba; Ria, Francesco; Abadi, Ehsan; Segars, Paul; Ramirez-Giraldo, Juan Carlos; Jones, Elizabeth C; Henry, Travis; Marin, Daniele; Samei, EhsanOBJECTIVE: Different methods can be used to condition imaging systems for clinical use. The purpose of this study was to assess how these methods complement one another in evaluating a system for clinical integration of an emerging technology, photon-counting computed tomography (PCCT), for thoracic imaging. METHODS: Four methods were used to assess a clinical PCCT system (NAEOTOM Alpha; Siemens Healthineers, Forchheim, Germany) across 3 reconstruction kernels (Br40f, Br48f, and Br56f). First, a phantom evaluation was performed using a computed tomography quality control phantom to characterize noise magnitude, spatial resolution, and detectability. Second, clinical images acquired using conventional and PCCT systems were used for a multi-institutional reader study where readers from 2 institutions were asked to rank their preference of images. Third, the clinical images were assessed in terms of in vivo image quality characterization of global noise index and detectability. Fourth, a virtual imaging trial was conducted using a validated simulation platform (DukeSim) that models PCCT and a virtual patient model (XCAT) with embedded lung lesions imaged under differing conditions of respiratory phase and positional displacement. Using known ground truth of the patient model, images were evaluated for quantitative biomarkers of lung intensity histograms and lesion morphology metrics. RESULTS: For the physical phantom study, the Br56f kernel was shown to have the highest resolution despite having the highest noise and lowest detectability. Readers across both institutions preferred the Br56f kernel (71% first rank) with a high interclass correlation (0.990). In vivo assessments found superior detectability for PCCT compared with conventional computed tomography but higher noise and reduced detectability with increased kernel sharpness. For the virtual imaging trial, Br40f was shown to have the best performance for histogram measures, whereas Br56f was shown to have the most precise and accurate morphology metrics. CONCLUSION: The 4 evaluation methods each have their strengths and limitations and bring complementary insight to the evaluation of PCCT. Although no method offers a complete answer, concordant findings between methods offer affirmatory confidence in a decision, whereas discordant ones offer insight for added perspective. Aggregating our findings, we concluded the Br56f kernel best for high-resolution tasks and Br40f for contrast-dependent tasks.Item Open Access Validation and Application of a Virtual Imaging Trial Platform for Accurate and Precise CT Quantifications in Lung Imaging(2021) Shankar, Sachin SureshComputed Tomography (CT) is a prevalent imaging technique in modern medicine that provides physicians a non-invasive method to evaluate and diagnose various clinical conditions. To aid in diagnosis, it is important to have a high accuracy and reliability in these images. In the first phase of this study, the variability of clinically-relevant imaging biomarkers was analyzed across different scanners and imaging parameters through usage of a customized anthropomorphic chest phantom with several experimental sample inserts. This phantom was scanned across 10 different scanners. Imaging biomarkers were computed for each scan. Intra and inter-scan variability was assessed by computing coefficients of variation and standard deviations of the measurements. It was found that LAA -950 and LAA -856 were the biomarkers with the highest levels of variability, while the majority of other biomarkers had variability less than 10 HU or 10% CV in both inter and intra-scan measurements. No clear trend was found between the variability of the biomarkers and radiation dose (i.e., CTDI).
Traditional assessments of CT technologies are limited in the sense that they work with real patient data and are not efficient. Alternatively, Virtual Imaging Trials (VITs), which use virtual scanners and patients, are more efficient and avoid unnecessary radiation exposure. DukeSim is a CT simulator that has been validated with simple cylindrical phantoms in the past, but not with more clinically-relevant phantoms and conditions. Biomarkers computed from real CT image data were compared to those from simulated CT scans of a computational version of an anthropomorphic chest phantom. Overall, relative percent errors ranged from 0.187% to 18.269%.
Having validated DukeSim in a clinically relevant context, the utility of DukeSim as a VIT tool was shown by investigating the effects of imaging and reconstruction parameters on the clinically relevant biomarkers. It was found that sharper reconstruction kernels and lower tube currents tended to reduce the accuracy of measured biomarkers. These findings will help to spark further studies in virtual imaging, which can help to yield further clinical insights to improve patient health outcomes.