EUVIMEDAtrasti EUVIMED
Open access · CC BY-NC 4.0

Raw data consistent deep learning‐based field of view extension for dual‐source dual‐energy CT

Joscha Maier, Julien Erath, Stefan Sawall, Eric Fournié, Karl Stierstorfer, Marc Kachelrieß

Medical Physics · 2024

DOI: 10.1002/mp.16684 ↗

At a glance

Journal article in Medical Physics, 2024. Written by Joscha Maier and 5 co-authors. Institution: German Cancer Research Center (DKFZ), Heidelberg, Germany. Major MEDLINE topics: Deep Learning. Cited 4 times so far (Crossref); 28 references. The full text is freely available (open access, CC BY-NC 4.0). Funded by: Siemens Healthineers.

SantraukaSubjectsBibliografiniai duomenysCiteRelated studies

Abstract (original language)

Background

Due to technical constraints, dual‐source dual‐energy CT scans may lack spectral information in the periphery of the patient.

Purpose

Here, we propose a deep learning‐based iterative reconstruction to recover the missing spectral information outside the field of measurement (FOM) of the second source‐detector pair.

Methods

In today's Siemens dual‐source CT systems, one source‐detector pair (referred to as A) typically has a FOM of about 50 cm, while the FOM of the other pair (referred to as B) is limited by technical constraints to a diameter of about 35 cm. As a result, dual‐energy applications are currently only available within the small FOM, limiting their use for larger patients. To derive a reconstruction at B's energy for the entire patient cross‐section, we propose a deep learning‐based iterative reconstruction. Starting with A's reconstruction as initial estimate, it employs a neural network in each iteration to refine the current estimate according to a raw data fidelity measure. Here, the corresponding mapping is trained using simulated chest, abdomen, and pelvis scans based on a data set containing 70 full body CT scans. Finally, the proposed approach is tested on simulated and measured dual‐source dual‐energy scans and compared against existing reference approaches.

Results

For all test cases, the proposed approach was able to provide artifact‐free CT reconstructions of B for the entire patient cross‐section. Considering simulated data, the remaining error of the reconstructions is between 10 and 17 HU on average, which is about half as low as the reference approaches. A similar performance with an average error of 8 HU could be achieved for real phantom measurements.

Conclusions

The proposed approach is able to recover missing dual‐energy information for patients exceeding the small 35 cm FOM of dual‐source CT systems. Therefore, it potentially allows to extend dual‐energy applications to the entire‐patient cross section.

Subjects

MeSH terms (MEDLINE)

Deep LearningHumansTomography, X-Ray ComputedThoraxPhantoms, ImagingAlgorithmsImage Processing, Computer-Assisted

Author keywords

deep learningdual-source dual-energy CTfield of view extension

Authors and institutions

  • German Cancer Research Center (DKFZ), Heidelberg, Germany.
  • Heidelberg University, Heidelberg, Germany.
  • Siemens Healthineers, Forchheim, Germany.

Funding

  • Siemens Healthineers

Cite

APA

Maier, J., Erath, J., Sawall, S., Fournié, E., Stierstorfer, K., & Kachelrieß, M. (2024). Raw data consistent deep learning‐based field of view extension for dual‐source dual‐energy CT. Medical Physics. https://doi.org/10.1002/mp.16684

Vancouver

Maier J, Erath J, Sawall S, Fournié E, Stierstorfer K, Kachelrieß M. Raw data consistent deep learning‐based field of view extension for dual‐source dual‐energy CT. Medical Physics. 2024. doi:10.1002/mp.16684

BibTeX
@article{Maier2024,
  author = {Maier, Joscha and Erath, Julien and Sawall, Stefan and Fournié, Eric and Stierstorfer, Karl and Kachelrieß, Marc},
  title = {Raw data consistent deep learning‐based field of view extension for dual‐source dual‐energy CT},
  journal = {Medical Physics},
  year = {2024},
  doi = {10.1002/mp.16684}
}
Start for free →