The Lancet medical imaging & theranostics最新文献

{"title":"Predicting brain tumour enhancement from non-contrast MRI with artificial intelligence: a multicohort, retrospective, diagnostic accuracy study.","authors":"James K Ruffle, Samia Mohinta, Guilherme Pombo, Asthik Biswas, Alan Campbell, Indran Davagnanam, David Doig, Ahmed Hammam, Harpreet Hyare, Farrah Jabeen, Emma Lim, Dermot Mallon, Stephanie Owen, Sophie Wilkinson, Sebastian Brandner, Parashkev Nachev","doi":"10.1016/j.lanmit.2026.100012","DOIUrl":"10.1016/j.lanmit.2026.100012","url":null,"abstract":"<p><strong>Background: </strong>Brain tumour imaging assessment typically requires both pre-contrast and post-contrast MRI, but gadolinium administration is not always desirable-for example, in patients requiring frequent follow-up, or those with renal impairment or allergy, or those who are of pediatric age [0-18 years]. We aimed to develop and validate a deep learning model to predict brain tumour contrast enhancement from non-contrast MRI sequences alone.</p><p><strong>Methods: </strong>In this multicohort, retrospective study, we compiled 11 089 brain MRI studies (acquired in 2006-24) from ten datasets across four countries (the UK, the USA, the Netherlands, and Nigeria) and three continents (North America, Europe, and Africa), spanning adult and paediatric populations with no age restriction, and including glioma, meningioma, metastases, and pre-treatment and post-treatment appearances. Three deep learning architectures (nnU-Net, SegResNet, and SwinUNETR) were trained to detect and segment enhancing tumour tissue from T1-weighted (T1w), T2-weighted (T2w), and T2/fluid-attenuated inversion recovery (FLAIR) sequences alone. Performance was assessed in a held-out test set of 1109 studies, with patient-level enhancement detection as the primary endpoint and voxel-level Dice score as the secondary endpoint. Eleven expert radiologists performed the same novel task on a 564-case subset, each reviewing 100 cases while being blinded to clinical history, previous imaging, and referral indication.</p><p><strong>Findings: </strong>The best performing model, nnU-Net, reached a balanced accuracy of 83·0% (95% CI 79·1-87·2; sensitivity 91·5% [89·9-93·1], specificity 74·4% [66·7-82·5]) for enhancement detection, with R<sup>2</sup>=0·859 for enhancement-volume prediction. Among the enhancing cases, 762 (77%) of 992 reached a Dice score of greater than or equal to 0·3, 670 (68%) of 992 had a Dice score of greater than or equal to 0·5, and 498 (50%) of 992 showed a Dice score of greater than or equal to 0·7. In the 564 matched cases, the radiologists' majority-vote performance was lower, with a balanced accuracy of 71·7% (sensitivity 77·6%, specificity 65·8%) under the same blinded conditions. Patient-level enhancement detection was generally consistent across pathologies (proportion of patients reaching voxel-level Dice ≥0·3: meningioma, 93 [93%] of 100; presurgical glioma, 526 [76%] of 691; metastases, 52 [74%] of 70; post-treatment glioma, 82 [74%] of 111), but was lower for paediatric cases (nine [45%] of 20).</p><p><strong>Interpretation: </strong>Deep learning can identify contrast-enhancing brain tumours from non-contrast MRI. These models show promise as a triage or decision-support aid-for example, flagging studies in which post-contrast enhancement is likely so that contrast can be added to a non-contrast-only protocol-and could reduce gadolinium dependence in neuro-oncology imaging. Future studies should optimise these models through close collabora","PeriodicalId":524237,"journal":{"name":"The Lancet medical imaging & theranostics","volume":"1 2","pages":"None"},"PeriodicalIF":0.0,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13637711/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149123913","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Low-dose CT for MRI-free quantification of tau PET in Alzheimer's disease: a multicentre, cross-sectional, comparison study.","authors":"Guilherme Povala, Livia Amaral, Bruna Bellaver, Firoza Z Lussier, Pamela C L Ferreira, Douglas T Leffa, Guilherme Bauer-Negrini, Raphael de Luca E Tuma, Alvaro de Oliveira Franco, Cécile Tissot, Quentin Finn, João Pedro Ferrari-Souza, Eduardo R Zimmer, Dana L Tudorascu, Joseph C Masdeu, Juan Fortea, David N Soleimani-Meigooni, Val Lowe, Hwamee Oh, Brian A Gordon, Belen Pascual, Pedro Rosa-Neto, Suzanne Baker, Tharick A Pascoal","doi":"10.1016/j.lanmit.2026.100007","DOIUrl":"10.1016/j.lanmit.2026.100007","url":null,"abstract":"<p><strong>Background: </strong>Tau PET quantification typically requires structural MRI for reference-region definition, normalisation, and segmentation. We evaluated whether low-dose CT can serve as a reliable MRI alternative for tau PET quantification when MRI is unavailable.</p><p><strong>Methods: </strong>In this multicentre cross-sectional methods comparison study, we developed a PET-CT pipeline using low-dose CT as the sole anatomical reference for tau PET quantification. Participants were recruited from six sites in the multicentre Head-to-head Harmonisation of tau PET tracers (HEAD) study. The cohort comprised cognitively unimpaired (CU) young adults (<28 years), CU middle-aged adults (50-65 years), CU elderly (>65 years), and participants with mild cognitive impairment (MCI) or dementia. Participants underwent [<sup>18</sup>F]MK6240 and [<sup>18</sup>F]Flortaucipir PET-CT, T1-weighted MRI, and amyloid PET. SUVRs were calculated for three AD-relevant composite regions and 34 individual cortical regions using a template-based cerebellar reference and compared with PET-MR pipelines using template-based or FreeSurfer subject-space segmentation. Agreement was assessed using linear regression, intraclass correlation coefficients (ICC), Bland-Altman analysis, equivalence testing (two one-sided tests, TOST), and Cohen's kappa (κ) for binary positivity concordance.</p><p><strong>Findings: </strong>A total of 387 participants were included, with recruitment spanning from November 11, 2022 and July 7, 2025 (205 [53%] female, 182 [47%] male; 6 [2%] aged 19-25 years, 130 [34%] aged 50-65 years and 251 [65%] aged 66-89 years). Against the template-based PET-MR pipeline, CT-derived SUVRs showed near-perfect linear correspondence with the template-based PET-MR pipeline for MK6240 (R<sup>2</sup> 0·959-0·975) and Flortaucipir (R<sup>2</sup> 0·925-0·964), with high absolute agreement (ICC 0·976-0·985 and 0·955-0·978, respectively). Against the FreeSurfer subject-space pipeline, correspondence remained strong for MK6240 (R<sup>2</sup> 0·952-0·971) and Flortaucipir (R<sup>2</sup> 0·909-0·949). Mean R<sup>2</sup> across 34 cortical regions was 0·917 (SD 0·049) for MK6240 and 0·874 (SD 0·057) for Flortaucipir. Tau-positivity concordance was high (92·0-96·6%; κ 0·802-0·895; all p<0·0001), with discordant cases near the classification threshold. Associations with amyloid burden and memory were preserved across pipelines (Steiger's test, all p>0·05). Equivalence was formally confirmed by TOST (all p<0·0001).</p><p><strong>Interpretation: </strong>CT-only processing enables robust tau PET quantification comparable to standard MRI-based pipelines, including computationally intensive FreeSurfer-based approaches. This method was validated across different radiotracers and represents a reliable alternative when MRI is unavailable or contraindicated. Because low-dose CT is routinely acquired during PET-CT sessions, this approach enables quantification of legacy datase","PeriodicalId":524237,"journal":{"name":"The Lancet medical imaging & theranostics","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13662961/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149198849","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}