Uncertainty in denoising of MRSI using low-rank methods
<p><strong>Purpose:</strong> Low-rank denoising of MRSI data results in an apparent increase in spectral SNR. However, it is not clear if this translates to a lower uncertainty in metabolite concentrations after spectroscopic fitting. Estimation of the true uncertainty after denois...
| Main Authors: | , |
|---|---|
| Format: | Journal article |
| Language: | English |
| Published: |
Wiley
2021
|
| _version_ | 1826289499456405504 |
|---|---|
| author | Clarke, WT Chiew, M |
| author_facet | Clarke, WT Chiew, M |
| author_sort | Clarke, WT |
| collection | OXFORD |
| description | <p><strong>Purpose:</strong> Low-rank denoising of MRSI data results in an apparent increase in spectral SNR. However, it is not clear if this translates to a lower uncertainty in metabolite concentrations after spectroscopic fitting. Estimation of the true uncertainty after denoising is desirable for downstream analysis in spectroscopy.
In this work the uncertainty reduction from low-rank denoising methods based on spatio-temporal separability and linear predictability in MRSI are assessed. A new method for estimating metabolite concentration uncertainty after denoising is proposed. Automatic rank threshold selection methods are also assessed in simulated low SNR regimes.</p>
<p><strong>Methods:</strong> Assessment of denoising methods is conducted using Monte Carlo simulation of proton MRSI data, and by reproducibility of repeated in vivo acquisitions in five subjects.</p>
<p><strong>Results:</strong> In simulated and in vivo data, spatio-temporal based denoising is shown to reduce the concentration uncertainty, but linear prediction denoising increases uncertainty. Uncertainty estimates provided by fitting algorithms after denoising consistently under-estimate actual metabolite uncertainty. However, the proposed uncertainty estimation, based on an analytical expression for entry-wise variance after denoising, is more accurate. It is also shown automated rank threshold selection using Marchenko-Pastur distribution can bias the data in low SNR conditions. An alternative soft-thresholding function is proposed.</p>
<p><strong>Conclusion:</strong> Low-rank denoising methods based on spatio-temporal separability do reduce uncertainty in MRS(I) data. However, thorough assessment is needed as assessment by SNR measured from residual baseline noise is insufficient given the presence of non-uniform variance. It is also important to select the right rank thresholding method in low SNR cases.</p> |
| first_indexed | 2024-03-07T02:29:48Z |
| format | Journal article |
| id | oxford-uuid:a6d9f770-cbbd-4b93-bf7a-005478287d0c |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-07T02:29:48Z |
| publishDate | 2021 |
| publisher | Wiley |
| record_format | dspace |
| spelling | oxford-uuid:a6d9f770-cbbd-4b93-bf7a-005478287d0c2022-03-27T02:50:29ZUncertainty in denoising of MRSI using low-rank methodsJournal articlehttp://purl.org/coar/resource_type/c_dcae04bcuuid:a6d9f770-cbbd-4b93-bf7a-005478287d0cEnglishSymplectic ElementsWiley2021Clarke, WTChiew, M<p><strong>Purpose:</strong> Low-rank denoising of MRSI data results in an apparent increase in spectral SNR. However, it is not clear if this translates to a lower uncertainty in metabolite concentrations after spectroscopic fitting. Estimation of the true uncertainty after denoising is desirable for downstream analysis in spectroscopy. In this work the uncertainty reduction from low-rank denoising methods based on spatio-temporal separability and linear predictability in MRSI are assessed. A new method for estimating metabolite concentration uncertainty after denoising is proposed. Automatic rank threshold selection methods are also assessed in simulated low SNR regimes.</p> <p><strong>Methods:</strong> Assessment of denoising methods is conducted using Monte Carlo simulation of proton MRSI data, and by reproducibility of repeated in vivo acquisitions in five subjects.</p> <p><strong>Results:</strong> In simulated and in vivo data, spatio-temporal based denoising is shown to reduce the concentration uncertainty, but linear prediction denoising increases uncertainty. Uncertainty estimates provided by fitting algorithms after denoising consistently under-estimate actual metabolite uncertainty. However, the proposed uncertainty estimation, based on an analytical expression for entry-wise variance after denoising, is more accurate. It is also shown automated rank threshold selection using Marchenko-Pastur distribution can bias the data in low SNR conditions. An alternative soft-thresholding function is proposed.</p> <p><strong>Conclusion:</strong> Low-rank denoising methods based on spatio-temporal separability do reduce uncertainty in MRS(I) data. However, thorough assessment is needed as assessment by SNR measured from residual baseline noise is insufficient given the presence of non-uniform variance. It is also important to select the right rank thresholding method in low SNR cases.</p> |
| spellingShingle | Clarke, WT Chiew, M Uncertainty in denoising of MRSI using low-rank methods |
| title | Uncertainty in denoising of MRSI using low-rank methods |
| title_full | Uncertainty in denoising of MRSI using low-rank methods |
| title_fullStr | Uncertainty in denoising of MRSI using low-rank methods |
| title_full_unstemmed | Uncertainty in denoising of MRSI using low-rank methods |
| title_short | Uncertainty in denoising of MRSI using low-rank methods |
| title_sort | uncertainty in denoising of mrsi using low rank methods |
| work_keys_str_mv | AT clarkewt uncertaintyindenoisingofmrsiusinglowrankmethods AT chiewm uncertaintyindenoisingofmrsiusinglowrankmethods |