Browsing by Author "Smith, Taylor"

Now showing 1 - 4 of 4
  • Thumbnail Image
    ItemOpen 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, Ehsan
    Image 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.
  • Thumbnail Image
    ItemOpen 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, Ehsan
    PURPOSE 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.
  • Thumbnail Image
    ItemOpen Access
    Scientific Abstracts and Sessions
    (Medical Physics, 2020-06) Ria, Francesco; Smith, Taylor; Abadi, Ehsan; Solomon, Justin; Samei, ehsan
    Purpose 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.
  • Thumbnail Image
    ItemOpen Access
    Total risk index: a mathematical model for decision making based on clinical and radiation risk assessment in CT
    (2019-12-04) Ria, Francesco; Smith, Taylor; Hoye, Jocelyn; Marin, Daniele; Samei, Ehsan
    Purpose. Radiological risk is a combination of radiation and clinical risk (likelihood of not delivering a proper diagnosis), which together may be characterized as a total risk index (TRI). While many strategies have been developed to ascertain radiation risk, there has been a paucity of studies assessing the clinical risk. This knowledge gap makes impossible to determine the total radiological procedure risk and, thus, to perform a comprehensive optimization. The purpose of this study was to develop a mathematical model to ascertain TRI and to identify the minimum TRI (mTRI) in a clinical CT population. Materials and Methods. This IRB approved study included 21 adults abdomen exams performed on a dual-source single energy CT at two different dose levels (84 CT series). Virtual liver lesions were inserted into projection data to simulate localized stage liver cancer (LSLC). The detectability index (d') was calculated in each series and converted to percentage of correct observer answers (AUC) in a two-alternative forced-choice model. The AUC was converted into the loss of 5-year relative survival rate (SEER, NCI), considering an upper bound on patient's risk for a misdiagnosis of LSLC (false positive + false negative). Concerning radiation risk, organ doses were estimated using a Monte Carlo method and the Risk Index was calculated and converted in 5-year relative survival rate for cancer. Finally, the two risks were weighted equally into a combined TRI curve per each patient as a function of CTDIvol. The analytical minimum of each TRI curve provided the patient mTRI. Results. The mTRI for LSLC patients that underwent an abdominal CT exhibited a rapid rise at low radiation dose due to enhanced clinical risk of under-dosed examinations. Increasing dose offered less risk with mortality per 100 patients between 2.1 and 6.5 (mean 4.5) at CTDIvol=5mGy, between 1.1 and 5.9 (mean 3.5) at CTDIvol=10mGy and between 0.5 and 5.4 (mean 3.0) at CTDIvol=20 mGy. Conclusion. The clinical risk seems to play a more dominant factor in designing optimum CT protocols. The TRI may provide an objective and quantifiable metric of the interplay of radiation and clinical risks during the optimization of the CT technique for individual patients. Clinical Relevance statement. CT risk-based optimization can be made possible by first quantifying both radiation and clinical risk using comparable units, then calculating an overall risk, and finally minimizing the total risk.