Material decomposition from photon-counting CT using a convolutional neural network and energy-integrating CT training labels.

Nadkarni, Rohan; Allphin, Alex; Clark, Darin P; Badea, Cristian T

Material decomposition from photon-counting CT using a convolutional neural network and energy-integrating CT training labels.

dc.contributor.author	Nadkarni, Rohan
dc.contributor.author	Allphin, Alex
dc.contributor.author	Clark, Darin P
dc.contributor.author	Badea, Cristian T
dc.date.accessioned	2022-07-19T20:07:40Z
dc.date.available	2022-07-19T20:07:40Z
dc.date.issued	2022-06-29
dc.date.updated	2022-07-19T20:07:39Z
dc.description.abstract	Objective Photon-counting CT (PCCT) has better dose efficiency and spectral resolution than energy-integrating CT, which is advantageous for material decomposition. Unfortunately, the accuracy of PCCT-based material decomposition is limited due to spectral distortions in the photon-counting detector (PCD). Approach In this work, we demonstrate a deep learning (DL) approach that compensates for spectral distortions in the PCD and improves accuracy in material decomposition by using decomposition maps provided by high-dose multi-energy-integrating detector (EID) data as training labels. We use a 3D U-net architecture and compare networks with PCD filtered backprojection (FBP) reconstruction (FBP2Decomp), PCD iterative reconstruction (Iter2Decomp), and PCD decomposition (Decomp2Decomp) as the input. Main results We found that our Iter2Decomp approach performs best, but DL outperforms matrix inversion decomposition regardless of the input. Compared to PCD matrix inversion decomposition, Iter2Decomp gives 27.50% lower root mean squared error (RMSE) in the iodine (I) map and 59.87% lower RMSE in the photoelectric effect (PE) map. In addition, it increases the structural similarity (SSIM) by 1.92%, 6.05%, and 9.33% in the I, Compton scattering (CS), and PE maps, respectively. When taking measurements from iodine and calcium vials, Iter2Decomp provides excellent agreement with multi-EID decomposition. One limitation is some blurring caused by our DL approach, with a decrease from 1.98 line pairs/mm at 50% modulation transfer function (MTF) with PCD matrix inversion decomposition to 1.75 line pairs/mm at 50% MTF when using Iter2Decomp. Significance Overall, this work demonstrates that our DL approach with high-dose multi-EID derived decomposition labels is effective at generating more accurate material maps from PCD data. More accurate preclinical spectral PCCT imaging such as this could serve for developing nanoparticles that show promise in the field of theranostics (therapy and diagnostics).
dc.identifier.issn	0031-9155
dc.identifier.issn	1361-6560
dc.identifier.uri	https://hdl.handle.net/10161/25502
dc.language	eng
dc.publisher	IOP Publishing
dc.relation.ispartof	Physics in medicine and biology
dc.relation.isversionof	10.1088/1361-6560/ac7d34
dc.subject	CNN
dc.subject	deep learning
dc.subject	material decomposition
dc.subject	photon-counting CT
dc.subject	theranostics
dc.title	Material decomposition from photon-counting CT using a convolutional neural network and energy-integrating CT training labels.
dc.type	Journal article
duke.contributor.orcid	Badea, Cristian T\|0000-0002-1850-2522
pubs.organisational-group	Duke
pubs.organisational-group	Pratt School of Engineering
pubs.organisational-group	School of Medicine
pubs.organisational-group	Clinical Science Departments
pubs.organisational-group	Institutes and Centers
pubs.organisational-group	Biomedical Engineering
pubs.organisational-group	Radiology
pubs.organisational-group	Duke Cancer Institute
pubs.publication-status	Published