Monalisa: An Open-Source, User-Friendly, Well-Documented MATLAB Library for MRI Reconstruction
Poster No:
1872
Submission Type:
Abstract Submission
Authors:
Mauro Leidi1,2, Bastien Milani1,2, Yiwei Jia1,2, Dominik Helbing1, Jaime Barranco1,2,3, Benedetta Franceschiello1,2,4
Institutions:
1Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland, Sion, Switzerland, 2The Sense Innovation and Research Center, Lausanne and Sion, Switzerland, 3CIBM Center for Biomedical Imaging, Lausanne, Switzerland, 4Department of Radiology, Lausanne University Hospital (CHUV) and University of Lausanne (UNIL), Lausanne, Switzerland
First Author:
Mauro Leidi
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland|The Sense Innovation and Research Center
Sion, Switzerland|Lausanne and Sion, Switzerland
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland|The Sense Innovation and Research Center
Sion, Switzerland|Lausanne and Sion, Switzerland
Co-Author(s):
Bastien Milani
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland|The Sense Innovation and Research Center
Sion, Switzerland|Lausanne and Sion, Switzerland
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland|The Sense Innovation and Research Center
Sion, Switzerland|Lausanne and Sion, Switzerland
Yiwei Jia
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland|The Sense Innovation and Research Center
Sion, Switzerland|Lausanne and Sion, Switzerland
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland|The Sense Innovation and Research Center
Sion, Switzerland|Lausanne and Sion, Switzerland
Dominik Helbing
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland
Sion, Switzerland
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland
Sion, Switzerland
Jaime Barranco
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland|The Sense Innovation and Research Center|CIBM Center for Biomedical Imaging
Sion, Switzerland|Lausanne and Sion, Switzerland|Lausanne, Switzerland
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland|The Sense Innovation and Research Center|CIBM Center for Biomedical Imaging
Sion, Switzerland|Lausanne and Sion, Switzerland|Lausanne, Switzerland
Benedetta Franceschiello
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland|The Sense Innovation and Research Center|Department of Radiology, Lausanne University Hospital (CHUV) and University of Lausanne (UNIL)
Sion, Switzerland|Lausanne and Sion, Switzerland|Lausanne, Switzerland
Institute of Systems Engineering, School of Engineering, HES-SO Valais-Wallis, Sion, Switzerland|The Sense Innovation and Research Center|Department of Radiology, Lausanne University Hospital (CHUV) and University of Lausanne (UNIL)
Sion, Switzerland|Lausanne and Sion, Switzerland|Lausanne, Switzerland
Introduction:
MRI provides insights into tissue properties, pathologies, and neural activity, with data captured in k-space and reconstructed into diagnostic images (Ogawa et al. 1990, Ogawa et al., 1992, Plewes et al., 2012). Techniques like parallel imaging, compressed sensing, and deep learning have addressed challenges like long acquisition times and motion artifacts, enabling faster and higher-quality reconstructions (Montalt-Tordera et al., 2021, Yang et al., 2016, Zaitsev et al., 2015). Despite these advancements, MRI reconstruction is often treated as a "black box", with many researchers relying on opaque, proprietary methods implemented by MRI manufacturers. To address this, several open-source initiatives have emerged, aiming to foster transparent innovation and democratize access to MRI reconstruction by providing tools and frameworks that make these processes more accessible and comprehensible. Among these, BART (Uecker et al., 2015) is the reference toolkit for parallel imaging, non-Cartesian reconstruction, and machine learning integrations, widely adopted in the research community.
Toolboxes like BART still face limitations in accessibility. Their reliance on low-level programming languages and modular frameworks adds complexity, making the code hard to digest. Additionally their limited documentation, poses barriers for new developers seeking to learn or contribute.
In this article, we introduce Monalisa, a comprehensive, open-source, and user-friendly MATLAB library for MRI reconstruction, supporting Cartesian and non-Cartesian sampling, compressed sensing, and iterative methods.
Toolboxes like BART still face limitations in accessibility. Their reliance on low-level programming languages and modular frameworks adds complexity, making the code hard to digest. Additionally their limited documentation, poses barriers for new developers seeking to learn or contribute.
In this article, we introduce Monalisa, a comprehensive, open-source, and user-friendly MATLAB library for MRI reconstruction, supporting Cartesian and non-Cartesian sampling, compressed sensing, and iterative methods.
Methods:
The Monalisa reconstruction pipeline, illustrated in Figure 1, consists of four key steps:
1. Coil Sensitivity Estimation (C) determines the spatial sensitivity profiles of imaging coils, essential for combining coil signals. Sensitivity profiles are derived either from pre-scan images or k-space data and are represented as a complex matrix of size [ImageSize,nCh], where nCh is the number of coils.
2. Binning (B) partitions k-space readouts into groups using a logical mask of acquisition lines, defined as a matrix of size [nBins,nLines].
3. Mitosius organizes raw data, k-space trajectories, and computed volume elements into a standardized structure ready for reconstruction. Outputs are arranged in a cell array of size [nBins].
4. Image Reconstruction transforms frequency-domain data into spatial-domain images, supporting iterative techniques like iterative-SENSE and compressed sensing methods. Regularization options include spatial and temporal constraints, solved using conjugate gradient descent and ADMM. Outputs are nBins matrices, each of size [ImageSize].
We compared Monalisa with BART as in (Knopp et al., 2021), performing L2-regularized reconstruction using conjugate gradient descent. Simulated undersampled k-space data (30 radial lines) from N=4 different images were used. Regularization parameters were optimized via a linear search based on SSIM. Reconstructed images were rescaled to match the ground truth's mean within the ROI, enabling robust visual and quantitative comparisons.
1. Coil Sensitivity Estimation (C) determines the spatial sensitivity profiles of imaging coils, essential for combining coil signals. Sensitivity profiles are derived either from pre-scan images or k-space data and are represented as a complex matrix of size [ImageSize,nCh], where nCh is the number of coils.
2. Binning (B) partitions k-space readouts into groups using a logical mask of acquisition lines, defined as a matrix of size [nBins,nLines].
3. Mitosius organizes raw data, k-space trajectories, and computed volume elements into a standardized structure ready for reconstruction. Outputs are arranged in a cell array of size [nBins].
4. Image Reconstruction transforms frequency-domain data into spatial-domain images, supporting iterative techniques like iterative-SENSE and compressed sensing methods. Regularization options include spatial and temporal constraints, solved using conjugate gradient descent and ADMM. Outputs are nBins matrices, each of size [ImageSize].
We compared Monalisa with BART as in (Knopp et al., 2021), performing L2-regularized reconstruction using conjugate gradient descent. Simulated undersampled k-space data (30 radial lines) from N=4 different images were used. Regularization parameters were optimized via a linear search based on SSIM. Reconstructed images were rescaled to match the ground truth's mean within the ROI, enabling robust visual and quantitative comparisons.
Results:
As presented in Figure 2, Monalisa outperforms Bart on the 4 test examples, achieving higher SSIM in 75% of the cases and lower L2 distance in 100% of the cases. Toghether with the best regularisation images, a gridded reconstruction is presented for qualitative comparison.
Conclusions:
Monalisa contributes to MRI reconstruction with its accessible, transparent, and user-friendly design. Its modular pipeline, supporting advanced techniques like compressed sensing for non-Cartesian data, offers high-quality imaging comparable to BART. This open-source milestone promotes community engagement, advancing both research and education in MRI.
Documentation and Tutorials: https://mattechlab.github.io/monalisa/ (Sphinx (Brandl, 2021))
Source code: https://github.com/MattechLab/monalisa
Documentation and Tutorials: https://mattechlab.github.io/monalisa/ (Sphinx (Brandl, 2021))
Source code: https://github.com/MattechLab/monalisa
Neuroinformatics and Data Sharing:
Workflows
Novel Imaging Acquisition Methods:
Anatomical MRI 1
BOLD fMRI 2
Keywords:
Data analysis
Development
Informatics
Learning
MRI
MRI PHYSICS
Open-Source Code
Open-Source Software
Other - Image Reconstruction
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
Which processing packages did you use for your study?
Provide references using APA citation style.
2. Knopp, T., & Grosser, M. (2021). MRIReco.jl: An MRI reconstruction framework written in Julia. Magnetic resonance in medicine, 86(3), 1633-1646.
3. Lauterbur, P. C. (1973). Image formation by induced local interactions: examples employing nuclear magnetic resonance. Nature, 242(5394), 190-191.
4. Montalt-Tordera, J., Muthurangu, V., Hauptmann, A., & Steeden, J. A. (2021). Machine learning in magnetic resonance imaging: image reconstruction. Physica Medica, 83, 79-87.
5. Ogawa, S., Lee, T. M., Nayak, A. S., & Glynn, P. (1990). Oxygenation‐sensitive contrast in magnetic resonance image of rodent brain at high magnetic fields. Magnetic resonance in medicine, 14(1), 68-78.
6. Ogawa, S., Tank, D. W., Menon, R., Ellermann, J. M., Kim, S. G., Merkle, H., & Ugurbil, K. (1992). Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proceedings of the National Academy of Sciences, 89(13), 5951-5955.
7. Plewes, D. B., & Kucharczyk, W. (2012). Physics of MRI: a primer. Journal of magnetic resonance imaging, 35(5), 1038-1054.
8.Uecker, M., Ong, F., Tamir, J. I., Bahri, D., Virtue, P., Cheng, J. Y., ... & Lustig, M. (2015, May). Berkeley advanced reconstruction toolbox. In Proc. Intl. Soc. Mag. Reson. Med (Vol. 23, No. 2486, p. 9).
9. Yang, A. C., Kretzler, M., Sudarski, S., Gulani, V., & Seiberlich, N. (2016). Sparse reconstruction techniques in magnetic resonance imaging: methods, applications, and challenges to clinical adoption. Investigative radiology, 51(6), 349-364.
10. Zaitsev, M., Maclaren, J., & Herbst, M. (2015). Motion artifacts in MRI: A complex problem with many partial solutions. Journal of Magnetic Resonance Imaging, 42(4), 887-901.