{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,23]],"date-time":"2026-09-23T18:10:27Z","timestamp":1790187027434,"version":"4.1.0"},"reference-count":27,"publisher":"MDPI AG","issue":"6","license":[{"start":{"date-parts":[[2026,3,16]],"date-time":"2026-03-16T00:00:00Z","timestamp":1773619200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Deanship of Scientific Research, Islamic University of Madinah, Saudi Arabia"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Diagnostics"],"abstract":"<jats:p>Background: The generation of synthetic CT images from MRI scans represents a crucial step toward enabling MRI-only clinical workflows and supporting multi-modal integration in medical imaging, particularly in radiotherapy planning. Despite significant advancements in deep learning models, many current methods still struggle to reconstruct high-density structures, especially bone, and exhibit limited accuracy in density values. This shortcoming is largely attributed to the passage of excessive or noisy features through skip connections in the traditional U-Net architecture, which degrade the quality of information transmitted to the decoder, negatively impacting the clarity of anatomical boundaries and the pixel-wise accuracy of the resulting synthetic image. Methods: In this work, we propose an enhanced 3D U-Net architecture in which the Convolutional Block Attention Module (CBAM) is systematically integrated within each skip connection. The CBAM sequentially applies channel and spatial attention to adaptively reweight encoder feature maps before fusion with the decoder, thereby emphasizing anatomically relevant structures while suppressing irrelevant feature propagation. The model was trained and evaluated on the SynthRAD2023 (Task 1\u2014Brain) MRI\u2013CT dataset. To rigorously assess the contribution of the attention mechanism, a dedicated ablation study was conducted comparing three variants: 3D U-Net with Squeeze-and-Excitation (SE), Coordinate Attention (CA), and the proposed CBAM module. Performance was evaluated using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), and Normalized Cross-Correlation (NCC). Results: The ablation study demonstrated that the CBAM-enhanced model consistently outperformed both SE- and CA-based variants across all quantitative metrics. Specifically, the proposed method achieved an MAE of 38.2\u00b15.4 HU and an RMSE of 51.0\u00b112.0 HU, representing the lowest reconstruction errors among the evaluated models. In addition, it obtained a PSNR of 29.45\u00b12.10 dB, SSIM of 0.940\u00b10.031, and NCC of 0.967\u00b10.015, indicating superior structural preservation and strong voxel-wise correspondence between synthesized and reference CT volumes. These results confirm that the sequential integration of channel and spatial attention provides a statistically and practically meaningful improvement for high-fidelity MRI-to-CT synthesis. Conclusions: Generating high-resolution brain CT images from brain MRI scans using a 3D U-Net network enhanced with a CBAM module can contribute to supporting the clinical workflow by providing additional diagnostic data without the need for extra radiological examinations, thereby enhancing diagnostic efficiency and reducing radiation exposure. This technique helps reduce patient exposure to radiation and improves accessibility in resource-limited settings. Furthermore, this method is valuable for retrospective studies, surgical planning, and image-guided therapy, where complete multi-modal data may not always be available.<\/jats:p>","DOI":"10.3390\/diagnostics16060875","type":"journal-article","created":{"date-parts":[[2026,3,16]],"date-time":"2026-03-16T13:32:07Z","timestamp":1773667927000},"page":"875","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":1,"title":["An Adaptive Attention 3D U-Net for High-Fidelity MRI-to-CT Synthesis: Bridging the Anatomical Gap with CBAM"],"prefix":"10.3390","volume":"16","author":[{"ORCID":"https:\/\/orcid.org\/0009-0009-4318-0471","authenticated-orcid":false,"given":"Chaima","family":"Bensebihi","sequence":"first","affiliation":[{"name":"LiM Laboratory, Department of Computer Science, Faculty of Science and Technology, University of Souk Ahras, Souk Ahras 41000, Algeria"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7148-3212","authenticated-orcid":false,"given":"Nacer Eddine","family":"Benzebouchi","sequence":"additional","affiliation":[{"name":"LabGED Laboratory, Badji Mokhtar\u2014Annaba University, Annaba 23000, Algeria"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2827-4946","authenticated-orcid":false,"given":"Nawel","family":"Zemmal","sequence":"additional","affiliation":[{"name":"LabGED Laboratory, Badji Mokhtar\u2014Annaba University, Annaba 23000, Algeria"},{"name":"Department of Computer Science, Faculty of Science and Technology, University of Souk Ahras, Souk Ahras 41000, Algeria"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7050-0532","authenticated-orcid":false,"given":"Abdallah","family":"Namoun","sequence":"additional","affiliation":[{"name":"AI Center, Islamic University of Madinah, Madinah 42351, Saudi Arabia"},{"name":"Faculty of Computer and Information Systems, Islamic University of Madinah, Madinah 42351, Saudi Arabia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2667-3196","authenticated-orcid":false,"given":"Aida","family":"Chefrour","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Faculty of Science and Technology, University of Souk Ahras, Souk Ahras 41000, Algeria"},{"name":"LISCO Laboratory, Badji Mokhtar\u2014Annaba University, Annaba 23000, Algeria"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0008-6112-0141","authenticated-orcid":false,"given":"Siham","family":"Amrouch","sequence":"additional","affiliation":[{"name":"LiM Laboratory, Department of Computer Science, Faculty of Science and Technology, University of Souk Ahras, Souk Ahras 41000, Algeria"},{"name":"Department of Computer Science, Faculty of Science and Technology, University of Souk Ahras, Souk Ahras 41000, Algeria"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,3,16]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"1","DOI":"10.4018\/IJIIT.309583","article-title":"Unbalanced Learning for Diabetes Diagnosis Based on Enhanced Resampling and Stacking Classifier","volume":"18","author":"Zemmal","year":"2022","journal-title":"Int. J. Intell. Inf. Technol. (IJIIT)"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"92","DOI":"10.1186\/s13014-019-1308-y","article-title":"MR-guidance in clinical reality: Current treatment challenges and future perspectives","volume":"14","author":"Corradini","year":"2019","journal-title":"Radiat. Oncol."},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"McRobbie, D.W., Moore, E.A., Graves, M.J., and Prince, M.R. (2017). MRI from Picture to Proton, Cambridge University Press.","DOI":"10.1017\/9781107706958"},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"93","DOI":"10.1186\/s13014-020-01524-4","article-title":"Medical physics challenges in clinical MR-guided radiotherapy","volume":"15","author":"Kurz","year":"2020","journal-title":"Radiat. Oncol."},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"e14499","DOI":"10.1002\/acm2.14499","article-title":"Advancements in synthetic CT generation from MRI: A review of techniques, and trends in radiation therapy planning","volume":"25","author":"Bahloul","year":"2024","journal-title":"J. Appl. Clin. Med. Phys."},{"key":"ref_6","doi-asserted-by":"crossref","unstructured":"Nie, D., Cao, X., Gao, Y., Wang, L., and Shen, D. (2016). Estimating CT image from MRI data using 3D fully convolutional networks. Deep Learning in Medical Image Analysis, Springer.","DOI":"10.1007\/978-3-319-46976-8_18"},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"030005","DOI":"10.1063\/5.0183090","article-title":"CT synthesis from MRI using GAN architecture","volume":"2938","author":"Navneet","year":"2023","journal-title":"AIP Conf. Proc."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"101953","DOI":"10.1016\/j.compmedimag.2021.101953","article-title":"CT synthesis from MRI using multi-cycle GAN for head-and-neck radiation therapy","volume":"91","author":"Liu","year":"2021","journal-title":"Comput. Med. Imaging Graph."},{"key":"ref_9","first-page":"636","article-title":"Synthetic CT generation from MRI using 3D diffusion model","volume":"Volume 12926","author":"Pan","year":"2024","journal-title":"Medical Imaging 2024: Image Processing"},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Mahalle, P.N., Lohani, M.C., Hemelatha, S., Sharma, D., and Jayakumar, S.S. (2024). Converter Networks for Multi-Scale Tokens-Aware Multi-Region and Multi-Sequence MR-to-CT Synthesis in a Single Model. 2024 Global Conference on Communications and Information Technologies (GCCIT), IEEE.","DOI":"10.1109\/GCCIT63234.2024.10862644"},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"1429","DOI":"10.1002\/mrm.28008","article-title":"Deep learning\u2013based MR-to-CT synthesis: The influence of varying gradient echo\u2013based MR images as input channels","volume":"83","author":"Florkow","year":"2020","journal-title":"Magn. Reson. Med."},{"key":"ref_12","unstructured":"Oktay, O., Schlemper, J., Folgoc, L.L., Lee, M., Heinrich, M., Misawa, K., Mori, K., McDonagh, S., Hammerla, N.Y., and Kainz, B. (2018). Attention U-Net: Learning Where to Look for the Pancreas. arXiv."},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"120578","DOI":"10.1016\/j.ins.2024.120578","article-title":"CAT-Unet: An enhanced U-Net architecture with coordinate attention and skip-neighborhood attention transformer for medical image segmentation","volume":"670","author":"Ding","year":"2024","journal-title":"Inf. Sci."},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"10076","DOI":"10.1109\/TPAMI.2024.3435571","article-title":"Medical Image Segmentation Review: The Success of U-Net","volume":"46","author":"Azad","year":"2024","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_15","doi-asserted-by":"crossref","unstructured":"Woo, S., Park, J., Lee, J.-Y., and Kweon, I.S. (2018, January 8\u201314). CBAM: Convolutional Block Attention Module. Proceedings of the European Conference on Computer Vision (ECCV), Munich, Germany.","DOI":"10.1007\/978-3-030-01234-2_1"},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"5427","DOI":"10.1007\/s40747-024-01445-9","article-title":"An attention mechanism module with spatial perception and channel information interaction","volume":"10","author":"Wang","year":"2024","journal-title":"Complex Intell. Syst."},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"691","DOI":"10.1109\/TRPMS.2025.3533749","article-title":"Synthetic CT image generation from CBCT: A Systematic Review","volume":"9","author":"Altalib","year":"2025","journal-title":"IEEE Trans. Radiat. Plasma Med. Sci."},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Wang, J., Yu, Z., Luan, Z., Ren, J., Zhao, Y., and Yu, G. (2022). RDAU-Net: Based on a Residual Convolutional Neural Network with DFP and CBAM for Brain Tumor Segmentation. Front. Oncol., 12.","DOI":"10.3389\/fonc.2022.805263"},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Zhang, B., Qiu, S., and Liang, T. (2024). Dual attention-based 3D U-Net liver segmentation algorithm on CT images. Bioengineering, 11.","DOI":"10.3390\/bioengineering11070737"},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"3671","DOI":"10.1007\/s13042-023-01857-y","article-title":"ICUnet++: An Inception-CBAM network based on U-Net++ for MR spine image segmentation","volume":"14","author":"Li","year":"2023","journal-title":"Int. J. Mach. Learn. Cybern."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Balaji, P., Alsid, L.E.G., Mishra, S., Obaid, A.J., and Alkhafaji, M.A. (2024). Brain Tumor MRI Segmentation Using Deep Instance Segmentation with Bioinspired Optimization Algorithm. Doctoral Symposium on Computational Intelligence, Springer.","DOI":"10.1007\/978-981-97-6726-7_4"},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"832","DOI":"10.1007\/s11227-025-07339-y","article-title":"DPF-Unet: A CNN-swin transformer fusion network for 3D brain tumor segmentation in MRI images","volume":"81","author":"Li","year":"2025","journal-title":"J. Supercomput."},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"4664","DOI":"10.1002\/mp.16529","article-title":"SynthRAD2023 Grand Challenge dataset: Generating synthetic CT for radiotherapy","volume":"50","author":"Thummerer","year":"2023","journal-title":"Med. Phys."},{"key":"ref_24","doi-asserted-by":"crossref","unstructured":"Gul, M.S.K., Mukati, M.U., B\u00e4tz, M., Forchhammer, S., and Keinert, J. (2021, January 19\u201322). Light-field view synthesis using a convolutional block attention module. Proceedings of the 2021 IEEE International Conference on Image Processing (ICIP), Anchorage, AK, USA.","DOI":"10.1109\/ICIP42928.2021.9506586"},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"103276","DOI":"10.1016\/j.media.2024.103276","article-title":"Generating synthetic computed tomography for radiotherapy: SynthRAD2023 challenge report","volume":"97","author":"Huijben","year":"2024","journal-title":"Med. Image Anal."},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"2538","DOI":"10.1002\/mp.16847","article-title":"Synthetic CT generation from MRI using 3D transformer-based denoising diffusion model","volume":"51","author":"Pan","year":"2023","journal-title":"Med. Phys."},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"3772","DOI":"10.1002\/mp.17867","article-title":"Generation of synthetic CT from MRI for MRI-based attenuation correction of brain PET images using radiomics and machine learning","volume":"52","author":"Hoseinipourasl","year":"2025","journal-title":"Med. Phys."}],"container-title":["Diagnostics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2075-4418\/16\/6\/875\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,20]],"date-time":"2026-03-20T10:59:20Z","timestamp":1774004360000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2075-4418\/16\/6\/875"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,16]]},"references-count":27,"journal-issue":{"issue":"6","published-online":{"date-parts":[[2026,3]]}},"alternative-id":["diagnostics16060875"],"URL":"https:\/\/doi.org\/10.3390\/diagnostics16060875","relation":{},"ISSN":["2075-4418"],"issn-type":[{"value":"2075-4418","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,3,16]]}}}