Daniele Marin

Overview:

Liver Imaging
Dual Energy CT
CT Protocol Optimization
Dose Reduction Strategies for Abdominal CT Applications

Positions:

Associate Professor of Radiology

Radiology, Abdominal Imaging
School of Medicine

Member of the Duke Cancer Institute

Duke Cancer Institute
School of Medicine

Education:

M.D. 2003

Sapienza University of Rome (Italy)

Grants:

LowEr Administered Dose with highEr Relaxivity: Gadovist vs Dotarem (LEADER 75)

Administered By
Radiology, Abdominal Imaging
Awarded By
Bayer Healthcare Pharmaceuticals Inc
Role
Principal Investigator
Start Date
End Date

Dual-Shot NCOM Power Contract Injector Study

Administered By
Radiology, Abdominal Imaging
Awarded By
Nemoto Kyorindo Co., Ltd.
Role
Principal Investigator
Start Date
End Date

Optimization of a Frequency-Based Fusion Technique for Improving the Image Quality on Low Energy Virtual Monochromatic Images from Dual Energy CT

Administered By
Radiology, Abdominal Imaging
Awarded By
Radiological Society of North America
Role
Principal Investigator
Start Date
End Date

Toward Precise and Accurate Assessment of Dose Reduction Using Iterative Reconstruction Methods for Abdominal Imaging Applications

Administered By
Radiology, Abdominal Imaging
Awarded By
Society of Abdominal Radiology
Role
Principal Investigator
Start Date
End Date

CT Research Fellowship in Dual Energy and Deep Learning Image Reconstruction

Administered By
Radiology, Abdominal Imaging
Awarded By
GE Healthcare
Role
Principal Investigator
Start Date
End Date

Publications:

Deep learning based spectral extrapolation for dual-source, dual-energy x-ray computed tomography.

PURPOSE: Data completion is commonly employed in dual-source, dual-energy computed tomography (CT) when physical or hardware constraints limit the field of view (FoV) covered by one of two imaging chains. Practically, dual-energy data completion is accomplished by estimating missing projection data based on the imaging chain with the full FoV and then by appropriately truncating the analytical reconstruction of the data with the smaller FoV. While this approach works well in many clinical applications, there are applications which would benefit from spectral contrast estimates over the larger FoV (spectral extrapolation)-e.g. model-based iterative reconstruction, contrast-enhanced abdominal imaging of large patients, interior tomography, and combined temporal and spectral imaging. METHODS: To document the fidelity of spectral extrapolation and to prototype a deep learning algorithm to perform it, we assembled a data set of 50 dual-source, dual-energy abdominal x-ray CT scans (acquired at Duke University Medical Center with 5 Siemens Flash scanners; chain A: 50 cm FoV, 100 kV; chain B: 33 cm FoV, 140 kV + Sn; helical pitch: 0.8). Data sets were reconstructed using ReconCT (v14.1, Siemens Healthineers): 768 × 768 pixels per slice, 50 cm FoV, 0.75 mm slice thickness, "Dual-Energy - WFBP" reconstruction mode with dual-source data completion. A hybrid architecture consisting of a learned piecewise linear transfer function (PLTF) and a convolutional neural network (CNN) was trained using 40 scans (five scans reserved for validation, five for testing). The PLTF learned to map chain A spectral contrast to chain B spectral contrast voxel-wise, performing an image domain analog of dual-source data completion with approximate spectral reweighting. The CNN with its U-net structure then learned to improve the accuracy of chain B contrast estimates by copying chain A structural information, by encoding prior chain A, chain B contrast relationships, and by generalizing feature-contrast associations. Training was supervised, using data from within the 33-cm chain B FoV to optimize and assess network performance. RESULTS: Extrapolation performance on the testing data confirmed our network's robustness and ability to generalize to unseen data from different patients, yielding maximum extrapolation errors of 26 HU following the PLTF and 7.5 HU following the CNN (averaged per target organ). Degradation of network performance when applied to a geometrically simple phantom confirmed our method's reliance on feature-contrast relationships in correctly inferring spectral contrast. Integrating our image domain spectral extrapolation network into a standard dual-source, dual-energy processing pipeline for Siemens Flash scanner data yielded spectral CT data with adequate fidelity for the generation of both 50 keV monochromatic images and material decomposition images over a 30-cm FoV for chain B when only 20 cm of chain B data were available for spectral extrapolation. CONCLUSIONS: Even with a moderate amount of training data, deep learning methods are capable of robustly inferring spectral contrast from feature-contrast relationships in spectral CT data, leading to spectral extrapolation performance well beyond what may be expected at face value. Future work reconciling spectral extrapolation results with original projection data is expected to further improve results in outlying and pathological cases.
Authors
Clark, DP; Schwartz, FR; Marin, D; Ramirez-Giraldo, JC; Badea, CT
MLA Citation
Clark, Darin P., et al. “Deep learning based spectral extrapolation for dual-source, dual-energy x-ray computed tomography.Med Phys, June 2020. Pubmed, doi:10.1002/mp.14324.
URI
https://scholars.duke.edu/individual/pub1448109
PMID
32531114
Source
pubmed
Published In
Med Phys
Published Date
DOI
10.1002/mp.14324

Variability of quantitative measurements of metastatic liver lesions: a multi-radiation-dose-level and multi-reader comparison.

PURPOSE: To evaluate the variability of quantitative measurements of metastatic liver lesions by using a multi-radiation-dose-level and multi-reader comparison. METHODS: Twenty-three study subjects (mean age, 60 years) with 39 liver lesions who underwent a single-energy dual-source contrast-enhanced staging CT between June 2015 and December 2015 were included. CT data were reconstructed with seven different radiation dose levels (ranging from 25 to 100%) on the basis of a single CT acquisition. Four radiologists independently performed manual tumor measurements and two radiologists performed semi-automated tumor measurements. Interobserver, intraobserver, and interdose sources of variability for longest diameter and volumetric measurements were estimated and compared using Wilcoxon rank-sum tests and intraclass correlation coefficients. RESULTS: Inter- and intraobserver variabilities for manual measurements of the longest diameter were higher compared to semi-automated measurements (p < 0.001 for overall). Inter- and intraobserver variabilities of volume measurements were higher compared to the longest diameter measurement (p < 0.001 for overall). Quantitative measurements were statistically different at < 50% radiation dose levels for semi-automated measurements of the longest diameter, and at 25% radiation dose level for volumetric measurements. The variability related to radiation dose was not significantly different from the inter- and intraobserver variability for the measurements of the longest diameter. CONCLUSION: The variability related to radiation dose is comparable to the inter- and intraobserver variability for measurements of the longest diameter. Caution should be warranted in reducing radiation dose level below 50% of a conventional CT protocol due to the potentially detrimental impact on the assessment of lesion response in the liver.
Authors
Ding, Y; Marin, D; Vernuccio, F; Gonzalez, F; Williamson, HV; Becker, H-C; Patel, BN; Solomon, J; Ramirez-Giraldo, JC; Samei, E; Nelson, RC; Meyer, M
MLA Citation
Ding, Yuqin, et al. “Variability of quantitative measurements of metastatic liver lesions: a multi-radiation-dose-level and multi-reader comparison.Abdom Radiol (Ny), June 2020. Pubmed, doi:10.1007/s00261-020-02601-8.
URI
https://scholars.duke.edu/individual/pub1447335
PMID
32524151
Source
pubmed
Published In
Abdom Radiol (Ny)
Published Date
DOI
10.1007/s00261-020-02601-8

CT material identification

© Springer Nature Switzerland AG 2020. Technical advances in CT imaging have improved ability to discriminate different materials, going beyond the attenuation imaging provided by most current systems. Nowadays, dual-energy CT systems allow for material identification and quantification providing qualitative and quantitative information about tissue composition and contrast agent distribution. This chapter explores the general principles of material decomposition analysis, as well as the technical implementations of single- and dual-source CT systems and the clinical advantages and prospects.
Authors
Vernuccio, F; Marin, D
MLA Citation
Vernuccio, F., and D. Marin. “CT material identification.” Computed Tomography: Approaches, Applications, and Operations, 2019, pp. 305–18. Scopus, doi:10.1007/978-3-030-26957-9_16.
URI
https://scholars.duke.edu/individual/pub1448504
Source
scopus
Published Date
Start Page
305
End Page
318
DOI
10.1007/978-3-030-26957-9_16

Noise and spatial resolution properties of a commercially available deep learning-based CT reconstruction algorithm.

PURPOSE: To characterize the noise and spatial resolution properties of a commercially available deep learning-based computed tomography (CT) reconstruction algorithm. METHODS: Two phantom experiments were performed. The first used a multisized image quality phantom (Mercury v3.0, Duke University) imaged at five radiation dose levels (CTDIvol : 0.9, 1.2, 3.6, 7.0, and 22.3 mGy) with a fixed tube current technique on a commercial CT scanner (GE Revolution CT). Images were reconstructed with conventional (FBP), iterative (GE ASiR-V), and deep learning-based (GE True Fidelity) reconstruction algorithms. Noise power spectrum (NPS), high-contrast (air-polyethylene interface), and intermediate-contrast (water-polyethylene interface) task transfer functions (TTF) were measured for each dose level and phantom size and summarized in terms of average noise frequency (fav ) and frequency at which the TTF was reduced to 50% (f50% ), respectively. The second experiment used a custom phantom with low-contrast rods and lung texture sections for the assessment of low-contrast TTF and noise spatial distribution. The phantom was imaged at five dose levels (CTDIvol : 1.0, 2.1, 3.0, 6.0, and 10.0 mGy) with 20 repeated scans at each dose, and images reconstructed with the same reconstruction algorithms. The local noise stationarity was assessed by generating spatial noise maps from the ensemble of repeated images and computing a noise inhomogeneity index, η , following AAPM TG233 methods. All measurements were compared among the algorithms. RESULTS: Compared to FBP, noise magnitude was reduced on average (± one standard deviation) by 74 ± 6% and 68 ± 4% for ASiR-V (at "100%" setting) and True Fidelity (at "High" setting), respectively. The noise texture from ASiR-V had substantially lower noise frequency content with 55 ± 4% lower NPS fav compared to FBP while True Fidelity had only marginally different noise frequency content with 9 ± 5% lower NPS fav compared to FBP. Both ASiR-V and True Fidelity demonstrated locally nonstationary noise in a lung texture background at all radiation dose levels, with higher noise near high-contrast edges of vessels and lower noise in uniform regions. At the 1.0 mGy dose level η values were 314% and 271% higher in ASiR-V and True Fidelity compared to FBP, respectively. High-contrast spatial resolution was similar between all algorithms for all dose levels and phantom sizes (<3% difference in TTF f50% ). Compared to FBP, low-contrast spatial resolution was lower for ASiR-V and True Fidelity with a reduction of TTF f50% of up to 42% and 36%, respectively. CONCLUSIONS: The deep learning-based CT reconstruction demonstrated a strong noise magnitude reduction compared to FBP while maintaining similar noise texture and high-contrast spatial resolution. However, the algorithm resulted in images with a locally nonstationary noise in lung textured backgrounds and had somewhat degraded low-contrast spatial resolution similar to what has been observed in currently available iterative reconstruction techniques.
Authors
Solomon, J; Lyu, P; Marin, D; Samei, E
MLA Citation
Solomon, Justin, et al. “Noise and spatial resolution properties of a commercially available deep learning-based CT reconstruction algorithm.Med Phys, June 2020. Pubmed, doi:10.1002/mp.14319.
URI
https://scholars.duke.edu/individual/pub1447264
PMID
32506661
Source
pubmed
Published In
Med Phys
Published Date
DOI
10.1002/mp.14319

Impact of dual energy cardiac CT for metal artefact reduction post aortic valve replacement.

PURPOSE: Assess image quality of dual-energy (DE) and single-energy (SE) cardiac multi-detector computed tomographic (MDCT) post aortic valve replacement (AVR) on a dual source MDCT scanner. METHODS: Eighty patients with cardiac MDCT acquisitions (ECG gated, dual-source) post-surgical and transcatheter AVR were retrospectively identified. Forty DE (cohort 1) and 40 SE acquisitions (cohort 2; 100 or 120 kVp) were reviewed. Metal artefact at valve coaptation (VC) and valve insertion site (VIS), and contrast enhancement were assessed. Valve leaflet edge definition was graded on a 4-point scale by three radiologists. RESULTS: The mean percentage valve area obscured by metal artifact differed between the cohorts; cohort 1 DE blended, high keV and low keV: 14.8 %, 11.1 % and 17.8 % at VC and 16.4 %, 13 %, 20.4 % at VIS respectively. Cohort 2: 25.8 % and 33.6 % (VC and VIS); each DE reconstruction vs SE: P < 0.0001. Average contrast opacification and coefficient of variance for cohort 1: 562.9 ± 144.7, 281.1 ± 60.3 and 1132.7 ± 300.8 Hounsfield Units (HU) and 9.6 %, 10 % and 8.9 %. For cohort 2: 437.2 ± 119.2 HU and 10.8 % (P < 0.01). Average leaflet edge definition cohort 1: 2.3 ± 0.4, 2.7 ± 0.2 and 2.3 ± 0.2, and cohort 2: 2.9 ± 0.2. CONCLUSION: DE high keV renderings can result in up to 17.2 % less metal artefact compared to standard SE acquisition for cardiac CT. Contrast opacification and homogeneity is higher for DE blended and low keV renderings compared to SE acquisition with leaflet visibility preferred for low keV and blended DE renderings.
Authors
Schwartz, FR; Tailor, T; Gaca, JG; Kiefer, T; Harrison, K; Hughes, GC; Ramirez-Giraldo, J-C; Marin, D; Hurwitz, LM
MLA Citation
Schwartz, Fides Regina, et al. “Impact of dual energy cardiac CT for metal artefact reduction post aortic valve replacement.Eur J Radiol, vol. 129, Aug. 2020, p. 109135. Pubmed, doi:10.1016/j.ejrad.2020.109135.
URI
https://scholars.duke.edu/individual/pub1448832
PMID
32590257
Source
pubmed
Published In
Eur J Radiol
Volume
129
Published Date
Start Page
109135
DOI
10.1016/j.ejrad.2020.109135