{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,10]],"date-time":"2026-01-10T09:20:02Z","timestamp":1768036802388,"version":"3.49.0"},"reference-count":24,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2024,6,12]],"date-time":"2024-06-12T00:00:00Z","timestamp":1718150400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Neuroinform."],"abstract":"<jats:sec><jats:title>Background<\/jats:title><jats:p>Quantitative maps obtained with diffusion weighted (DW) imaging, such as fractional anisotropy (FA) \u2013calculated by fitting the diffusion tensor (DT) model to the data,\u2014are very useful to study neurological diseases. To fit this map accurately, acquisition times of the order of several minutes are needed because many noncollinear DW volumes must be acquired to reduce directional biases. Deep learning (DL) can be used to reduce acquisition times by reducing the number of DW volumes. We already developed a DL network named \u201cone-minute FA,\u201d which uses 10 DW volumes to obtain FA maps, maintaining the same characteristics and clinical sensitivity of the FA maps calculated with the standard method using more volumes. Recent publications have indicated that it is possible to train DL networks and obtain FA maps even with 4 DW input volumes, far less than the minimum number of directions for the mathematical estimation of the DT.<\/jats:p><\/jats:sec><jats:sec><jats:title>Methods<\/jats:title><jats:p>Here we investigated the impact of reducing the number of DW input volumes to 4 or 7, and evaluated the performance and clinical sensitivity of the corresponding DL networks trained to calculate FA, while comparing results also with those using our one-minute FA. Each network training was performed on the human connectome project open-access dataset that has a high resolution and many DW volumes, used to fit a ground truth FA. To evaluate the generalizability of each network, they were tested on two external clinical datasets, not seen during training, and acquired on different scanners with different protocols, as previously done.<\/jats:p><\/jats:sec><jats:sec><jats:title>Results<\/jats:title><jats:p>Using 4 or 7 DW volumes, it was possible to train DL networks to obtain FA maps with the same range of values as ground truth - map, only when using HCP test data; pathological sensitivity was lost when tested using the external clinical datasets: indeed in both cases, no consistent differences were found between patient groups. On the contrary, our \u201cone-minute FA\u201d did not suffer from the same problem.<\/jats:p><\/jats:sec><jats:sec><jats:title>Conclusion<\/jats:title><jats:p>When developing DL networks for reduced acquisition times, the ability to generalize and to generate quantitative biomarkers that provide clinical sensitivity must be addressed.<\/jats:p><\/jats:sec>","DOI":"10.3389\/fninf.2024.1415085","type":"journal-article","created":{"date-parts":[[2024,6,12]],"date-time":"2024-06-12T04:42:20Z","timestamp":1718167340000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Finding the limits of deep learning clinical sensitivity with fractional anisotropy (FA) microstructure maps"],"prefix":"10.3389","volume":"18","author":[{"given":"Marta","family":"Gaviraghi","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Antonio","family":"Ricciardi","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Fulvia","family":"Palesi","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wallace","family":"Brownlee","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Paolo","family":"Vitali","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Ferran","family":"Prados","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Baris","family":"Kanber","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Claudia A. M.","family":"Gandini Wheeler-Kingshott","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1965","published-online":{"date-parts":[[2024,6,12]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"316","DOI":"10.1021\/jf505777p","article-title":"Diffusion tensor imaging of the brain.","volume":"4","author":"Alexander","year":"2007","journal-title":"Neurotherapeutics"},{"key":"B2","doi-asserted-by":"publisher","first-page":"845","DOI":"10.1002\/mrm.28470","article-title":"Extracting diffusion tensor fractional anisotropy and mean diffusivity from 3-direction DWI scans using deep learning.","volume":"85","author":"Aliotta","year":"2021","journal-title":"Magn. Reson. Med."},{"key":"B3","doi-asserted-by":"publisher","first-page":"247","DOI":"10.1006\/jmrb.1994.1037","article-title":"Estimation of the effective self-diffusion tensor from the NMR spin echo.","volume":"103","author":"Basser","year":"1994","journal-title":"J. Magn. Reson. Ser. B"},{"key":"B4","doi-asserted-by":"publisher","first-page":"755","DOI":"10.1136\/jnnp-2018-319634","article-title":"Cortical grey matter sodium accumulation is associated with disability and secondary progressive disease course in relapse-onset multiple sclerosis\u2019.","volume":"90","author":"Brownlee","year":"2019","journal-title":"J. Neurol. Neurosurg. Psychiatry"},{"key":"B5","first-page":"164","article-title":"Camino: Diffusion MRI reconstruction and processing.","volume":"2","author":"Cook","year":"2005","journal-title":"Statistics"},{"key":"B6","doi-asserted-by":"publisher","first-page":"131","DOI":"10.1007\/s00234-007-0317-y","article-title":"Feasibility of 1.6-mm isotropic voxel diffusion tensor tractography in depicting limbic fibers.","volume":"50","author":"Fujiwara","year":"2008","journal-title":"Neuroradiology"},{"key":"B7","doi-asserted-by":"publisher","DOI":"10.3390\/diagnostics11122181","article-title":"\u2018Deep learning applications in magnetic resonance imaging: Has the future become present?\u2019.","volume":"11","author":"Gassenmaier","year":"2021","journal-title":"Diagnostics"},{"key":"B8","doi-asserted-by":"publisher","DOI":"10.3389\/fninf.2022.891234","article-title":"A generalized deep learning network for fractional anisotropy reconstruction: Application to epilepsy and multiple sclerosis.","volume":"16","author":"Gaviraghi","year":"2022","journal-title":"Front. Neuroinform."},{"key":"B9","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-73018-5_21","article-title":"Automatic segmentation of dentate nuclei for microstructure assessment: Example of application to temporal lobe epilepsy patients","author":"Gaviraghi","year":"2021","journal-title":"Computational diffusion MRI. Mathematics and visualization"},{"key":"B10","first-page":"176","article-title":"Dependence of brain DTI maps of fractional anisotropy and mean diffusivity on the number of diffusion weighting directions.","author":"Giannelli","year":"2010"},{"key":"B11","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9781107415324.004","author":"Jones","year":"2011","journal-title":"Diffusion MRI: Theory, methods, and applications"},{"key":"B12","doi-asserted-by":"publisher","first-page":"7","DOI":"10.1002\/jmri.1880010103","article-title":"Imaging of diffusion and microcirculation with gradient sensitization: Design, strategy, and significance.","volume":"1","author":"Le Bihan","year":"1991","journal-title":"J. Magn. Reson. Imaging"},{"key":"B13","doi-asserted-by":"publisher","first-page":"474","DOI":"10.1002\/mrm.23254","article-title":"Six is enough? Comparison of diffusion parameters measured using six or more diffusion-encoding gradient directions with deterministic tractography.","volume":"68","author":"Lebel","year":"2012","journal-title":"Magn. Reson. Med."},{"key":"B14","doi-asserted-by":"publisher","first-page":"71398","DOI":"10.1109\/ACCESS.2019.2919241","article-title":"Fast and robust diffusion kurtosis parametric mapping using a three-dimensional convolutional neural network.","volume":"7","author":"Li","year":"2019","journal-title":"IEEE Access"},{"key":"B15","doi-asserted-by":"publisher","first-page":"637","DOI":"10.1148\/radiology.201.3.8939209","article-title":"Diffusion tensor MR imaging of the human brain.","volume":"201","author":"Pierpaoli","year":"1996","journal-title":"Radiology"},{"key":"B16","first-page":"234","article-title":"U-Net: Convolutional networks for biomedical image segmentation","author":"Ronneberger","year":"2015","journal-title":"Medical image computing and computer-assisted intervention \u2013 MICCAI 2015"},{"key":"B17","doi-asserted-by":"publisher","first-page":"1532","DOI":"10.1002\/mrm.22924.Diffusion","article-title":"Diffusion tensor imaging and beyond.","volume":"65","author":"Tournier","year":"2011","journal-title":"Magn. Reson. Med."},{"key":"B18","doi-asserted-by":"publisher","first-page":"62","DOI":"10.1016\/j.neuroimage.2013.05.041","article-title":"The WU-Minn human connectome project: An overview.","volume":"80","author":"Van Essen","year":"2013","journal-title":"Neuroimage"},{"key":"B19","doi-asserted-by":"publisher","first-page":"2222","DOI":"10.1115\/JRC2014-3865","article-title":"The human connectome project: A data acquisition perspective.","volume":"62","author":"Van Essen","year":"2012","journal-title":"Neuroimage"},{"key":"B20","doi-asserted-by":"publisher","first-page":"138","DOI":"10.1002\/mrm.22603","article-title":"More accurate estimation of diffusion tensor parameters using diffusion Kurtosis imaging.","volume":"65","author":"Veraart","year":"2011","journal-title":"Magn. Reson. Med."},{"key":"B21","doi-asserted-by":"publisher","first-page":"600","DOI":"10.1109\/TIP.2003.819861","article-title":"\u2018Image quality assessment: From error visibility to structural similarity\u2019.","volume":"13","author":"Wang","year":"2004","journal-title":"IEEE Trans. Image Process."},{"key":"B22","first-page":"1","year":"2017","journal-title":"WU-Minn HCP 1200 subjects data release: Reference manual"},{"key":"B23","doi-asserted-by":"publisher","first-page":"1357","DOI":"10.1016\/j.neuroimage.2009.09.057.HOW","article-title":"How does angular resolution affect diffusion imaging measures?","volume":"49","author":"Zhan","year":"2011","journal-title":"Neuroimage"},{"key":"B24","doi-asserted-by":"publisher","first-page":"2242","DOI":"10.1109\/ICCV.2017.244","article-title":"Unpaired image-to-image translation using cycle-consistent adversarial networks","author":"Zhu","year":"2017","journal-title":"Proceedings of the IEEE international conference on computer vision, 2017-October"}],"container-title":["Frontiers in Neuroinformatics"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fninf.2024.1415085\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,6,12]],"date-time":"2024-06-12T04:42:24Z","timestamp":1718167344000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fninf.2024.1415085\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,6,12]]},"references-count":24,"alternative-id":["10.3389\/fninf.2024.1415085"],"URL":"https:\/\/doi.org\/10.3389\/fninf.2024.1415085","relation":{},"ISSN":["1662-5196"],"issn-type":[{"value":"1662-5196","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,6,12]]},"article-number":"1415085"}}