BACKGROUND:Susceptibility-induced geometric distortions and non-physiologic hyperintensities confound acute ischemic stroke diagnosis on diffusion weighted imaging (DWI). Reverse phase-encoded imaging needed for artifact correction is rarely acquired acutely. Structural imaging-based correction is validated in healthy subjects, but its efficacy in acute stroke remains unclear. PURPOSE:To investigate whether structural imaging-based artifact correction reduces DWI distortions and non-physiologic hyperintensities in clinical acute stroke imaging. STUDY TYPE:Retrospective. POPULATION:214 patients with acute ischemic stroke and clinical MRI from two centers. FIELD STRENGTH/SEQUENCES:1.5 T/3 T; single-shot echo planar DWI; 3D MP-RAGE and 2D spin echo, gradient echo, turbo spin echo, and turbo inversion recovery T1-weighted sequences; 2D turbo spin echo and BLADE FLAIR sequences. ASSESSMENT:DWI distortions were corrected using the Synb0-DisCo and TOPUP algorithms with b0 and T1-weighted images. Mutual information evaluated geometric agreement against FLAIR and T1-weighted images before and after correction. The diagnostic value of hyperintense DWI lesions for apparent diffusion coefficient (ADC)-defined reference infarct was assessed. STATISTICAL TESTS:Wilcoxon sign rank or rank sum tests of mutual information, sensitivity, specificity, and positive and negative predictive value; Bland-Altman plots; and linear regression with coefficients of determination (R2) in linear and log space. Bonferroni-adjusted p < 0.05 was significant. RESULTS:After correction, DWI demonstrated significantly increased mutual information against FLAIR images [0.31 (IQR, 0.25-0.38) versus 0.30 (0.24-0.35)] and T1-weighted images [0.32 (IQR, 0.29-0.36) versus 0.29 (0.24-0.33)]. ADC-defined infarct volumes before and after correction correlated strongly (R2 0.998, log 0.541) with a negative bias of 0.51 (95% CI, 0.18-0.85) mL. DWI hyperintensities before and after correction correlated strongly (R2 0.996, log 0.360) but demonstrated a negative bias of 2.47 (95% CI, 1.95-3.00) mL. After correction, DWI hyperintensities demonstrated significantly improved specificity [0.998 (IQR, 0.994-1.000) versus 0.997 (0.992-0.999)] and positive predictive value [0.53 (IQR, 0.31-0.77) versus 0.39 (0.16-0.63)] for ADC-defined infarct. DATA CONCLUSION:Structural imaging-based distortion correction of routinely acquired clinical DWI in acute ischemic stroke improves anatomic geometry and reduces false positive hyperintensities. EVIDENCE LEVEL:3. STAGE OF TECHNICAL EFFICACY:3.
BACKGROUND:Right ventricular (RV) to left ventricular (LV) end-diastolic volume ratio (RV/LV) is a validated marker of RV dilation in pulmonary arterial hypertension (PAH), yet its impact on LV myocardial function and the underlying mediating pathway is incompletely characterized. PURPOSE:To investigate the association between RV/LV volume ratio and LV dysfunction in PAH patients, with a focus on regional strain patterns and mediating pathways. STUDY TYPE:Retrospective. POPULATION:188 PAH patients were divided into normal (ratio < 1.27, n = 74) and elevated (ratio ≥ 1.27, n = 114) RV/LV volume ratio groups, and 81 age- and sex-matched controls. FIELD STRENGTH/SEQUENCE:3.0-T/balanced steady-state free precession cine sequence. ASSESSMENT:LV global longitudinal, circumferential, and radial strains (LVGLS, LVGCS, and LVGRS, respectively) and regional strains were measured using feature tracking. Strain parameters were compared among controls and PAH subgroups. STATISTICAL TESTS:One-way ANOVA with Bonferroni-corrected post hoc tests, Kruskal-Wallis test, and Pearson or Spearman correlation. Multivariable linear regression was used to identify independent determinants and mediation analysis to examine the mediating pathways. A two-tailed p < 0.05 was deemed statistically significant. RESULTS:PAH patients with elevated volume ratio had significantly reduced LVGCS and LVGLS. Regional analysis showed that the most extensive strain impairment was at the mid-ventricular septum. The RV/LV volume ratio was independently associated with LVGCS (β = -0.194) and LVGLS (β = -0.307). Mediation analysis demonstrated that the mPAP-LVGLS relationship was statistically mediated by the RV/LV volume ratio, with a significant indirect effect (β = -0.192, 95% confidence interval: -0.302 to -0.096) and a non-significant (p = 0.544) direct effect. DATA CONCLUSION:Elevated RV/LV volume ratio in PAH is associated with spatially heterogeneous LV strain impairment, with the mid-septum being most vulnerable. The ratio mediates the adverse impact of pulmonary pressure on LV longitudinal strain and may serve as an imaging marker for assessing LV dysfunction. EVIDENCE LEVEL:3. TECHNICAL EFFICACY:Stage 3.
BACKGROUND:Pediatric posterior fossa tumors vary in malignancy, treatment, and prognosis across tumor types and molecular subtypes, yet noninvasive preoperative differentiation remains challenging. PURPOSE:To develop a deep learning (DL) pipeline using T2-weighted (T2w) MR images to automatically segment pediatric posterior fossa tumors, differentiate tumor types (medulloblastoma [MB], ependymoma [EP], pilocytic astrocytoma [PA]), and classify molecular subtypes of MB and EP. STUDY TYPE:Retrospective and prospective. POPULATION:1305 patients (M/F: 828/477; 490 MB, 327 EP, and 488 PA) from three centers. For tumor segmentation and classification, PF-nnU-Net was developed on the training set (n = 880) and validated on a validation set (n = 220), an internal prospective test set (n = 90), and two external independent test sets (n = 68, n = 47). MB-nnU-Net was trained on 338 patients and tested on 63 patients for MB subtyping; a prior developed EP-nnU-Net was tested on 38 patients for EP subtyping. FIELD STRENGTH/SEQUENCE:1.5 T or 3 T MRI, axial T2w images (turbo spin echo). ASSESSMENT:Three nnU-Net-based models: PF-nnU-Net and MB-nnU-Net for development, EP-nnU-Net for validation. Five-fold cross-validation was performed on training sets, followed by testing on independent test sets. STATISTICAL TESTS:Dice similarity coefficient for segmentation. Accuracy, sensitivity, specificity, the area under the receiver operating characteristic curve (AUC), and Cohen's kappa for classification. A two-sided p < 0.05 was considered significant. RESULTS:PF-nnU-Net achieved Dice scores of 0.94-0.96 and overall classification accuracy of 0.824-0.918 (multiclass Cohen's kappa: 0.722-0.873). MB-nnU-Net attained an overall accuracy of 0.794 (multiclass Cohen's kappa: 0.605), and EP-nnU-Net achieved an accuracy of 0.789 (Cohen's kappa: 0.538). DATA CONCLUSION:A fully automated DL pipeline was developed and validated to accurately segment pediatric posterior fossa tumors, differentiate tumor types (MB, EP, PA), and classify MB and EP molecular subtypes. EVIDENCE LEVEL:3. TECHNICAL EFFICACY:Stage 2.
BACKGROUND:Intracranial solitary fibrous tumors (ISFTs) are frequently misdiagnosed as meningiomas on MRI. Existing MRI signs have been evaluated individually, without a standardized and externally validated diagnostic framework. PURPOSE:To develop and validate an MRI-based diagnostic score for intracranial solitary fibrous tumors (ISFT-DS) for the preoperative identification of ISFTs. STUDY TYPE:Retrospective multicenter diagnostic accuracy study. POPULATION:Five hundred with ISFTs (mean age, 46.4 ± 11.7 years; 269 men), 250 with low-grade meningiomas (mean age, 57.0 ± 11.1 years; 51 men), and 250 with high-grade meningiomas (mean age, 49.6 ± 14.3 years; 108 men). The development and external validation cohorts included 840 (84%) and 160 (16%) patients, respectively. FIELDSTRENGTH/SEQUENCE:Spin-echo T1-weighted imaging (T1WI), fast or turbo spin-echo T2-weighted imaging, and contrast-enhanced spin-echo T1-weighted imaging. ASSESSMENT:Three potential MRI features-T1WI hyperintensity, honeycomb-like cystic change, and trans-compartmental growth pattern-were assessed by three neuroradiologists and combined into the 0-3-point ISFT-DS. Diagnostic performance and the effects of standardized training in six readers were evaluated. STATISTICAL TESTS:Sensitivity, specificity, accuracy, kappa statistics, receiver operating characteristic analysis, DeLong tests, χ2and Fisher exact tests, analysis of variance, McNemar test, Kruskal-Wallis tests. p < 0.05 indicated statistical significance. RESULTS:At an ISFT-DS cutoff of ≥ 1, sensitivity/specificity were 81.4%/83.3% in the development cohort and 76.3%/86.3% in the external validation cohort, respectively. Specificity reached 98.3%/98.8% at ≥ 2 and 100%/100% at 3. Interreader agreement was good to excellent for the three MRI features and the ISFT-DS (κ = 0.80-0.87). Standardized reader training on the ISFT-DS increased the overall AUC from 0.67 to 0.80 (ΔAUC = 0.13), with larger gains for Readers 1 and 2 (ΔAUC = 0.21 and 0.15, respectively). DATA CONCLUSION:The ISFT-DS provides a reproducible and teachable MRI-based scoring system for standardized preoperative identification of ISFTs. EVIDENCE LEVEL:3. TECHNICAL EFFICACY:Stage 2.
BACKGROUND:Noninvasive quantitative assessment of marrow recovery after curative therapies such as total marrow irradiation (TMI) and bone marrow transplantation (BMT) remain poorly defined, particularly in sickle cell disease (SCD). PURPOSE:To evaluate MRI-derived proton density fat fraction (PDFF) as a biomarker of marrow recovery following organ-sparing total marrow irradiation-based transplantation (OS-TMI/BMT) in an SCD mouse model and to assess initial clinical feasibility in patients. STUDY TYPE:Exploratory preclinical study with descriptive pilot feasibility analysis in humans. SUBJECTS:Townes SCD mice (n = 10 controls, n = 8 treated) and 2 adults (male, 24 ± 1 years old) with SCD. FIELD STRENGTH/SEQUENCE:Preclinical 7-T small-bore MRI used a spoiled gradient echo (GRE) multi-echo sequence for PDFF and R2* mapping, while clinical 3-T MRI employed a spoiled GRE-based sequence. ASSESSMENT:Marrow recovery in an animal study was assessed using MRI-derived PDFF with ROI analysis, complemented by micro-CT and histologic evaluation. Translational feasibility was assessed with longitudinal PDFF MRI in 2 adults with severe SCD undergoing marrow transplantation. STATISTICAL TESTS:Parametric data were analyzed using t-tests and Pearson correlation, nonparametric data using Mann-Whitney U and Spearman correlation, and two-way ANOVA assessed group, region, and interaction effects, with significance set at p < 0.05. RESULTS:Right femur MRI showed partial restoration of adipose-rich marrow in the OS-TMI/BMT group versus controls, higher PDFF in the proximal metaphysis (42.5 ± 2.7 vs. 34.8 ± 6.5), midshaft diaphysis (44.8 ± 3.0 vs. 26.9 ± 7.0), and distal metaphysis (45.1 ± 2.6 vs. 36.2 ± 4.9). OS-TMI/BMT also improved trabecular structure, with increased trabecular thickness, reduced bone surface-to-volume ratio, lower structure model index and anisotropy, and increased distal bone volume density. In 2 patients, PDFF increased post-transplant, and 1-year biopsy showed trilineage hematopoiesis without fibrosis. DATA CONCLUSION:Marrow-selective transplantation was associated with partial normalization of the marrow and skeletal microenvironment in SCD. PDFF may provide a practical noninvasive approach for monitoring post-transplant marrow recovery. LEVEL OF EVIDENCE: 4: TECHNICAL EFFICACY:Stage 2.
BACKGROUND:Respiratory-triggered T2-weighted imaging (T2WI) may use rectilinear sampling (respiratory-triggered fast recovery fast spin echo [RT-FRFSE]) or radial sampling (periodically rotated overlapping parallel lines with enhanced reconstruction [PROPELLER]). Their relative performance across breathing patterns is unclear. PURPOSE:To compare image quality of PROPELLER and RT-FRFSE according to pre-scan breathing pattern. STUDY TYPE:Retrospective. SUBJECTS:200 patients undergoing liver MRI (124 males; mean age, 58.1 ± 13.3 years). FIELD STRENGTH/SEQUENCE:1.5 T; fat-suppressed T2WI using PROPELLER and RT-FRFSE, both respiratory-triggered. ASSESSMENT:Patients were stratified by pre-scan respiratory rate and regularity: Group A (regular, ≤ 14 breaths/min, n = 49), B (15-17, n = 57), C (≥ 18, n = 49), D (irregular, n = 45). Three radiologists, blinded to respiratory group, graded ghost artifacts, sharpness, noise, and overall quality on five-point scales. Liver signal-to-noise (SNR) and liver-to-gallbladder contrast-to-noise ratios (CNR) were measured. STATISTICAL TESTS:Wilcoxon signed-rank test with Holm correction; Kruskal-Wallis analysis of within-patient difference scores and a linear mixed-effects model for SNR/CNR interactions; quadratic weighted kappa and intraclass correlation coefficient for interobserver agreement. p < 0.05 indicated significance. RESULTS:In Groups A and B, RT-FRFSE scored higher qualitatively, except ghost artifacts in Group A (p = 0.197). In Groups C and D, PROPELLER was superior (overall quality 3.96 vs. 3.41 and 3.93 vs. 3.01). Quantitatively (PROPELLER vs. RT-FRFSE), median SNR was 79.2 vs. 95.6 (A) and 70.0 vs. 58.6 (C), and median CNR 57.6 vs. 64.4 (A) and 55.5 vs. 42.7 (D); SNR did not differ in Groups B (p = 0.504) or D (p = 0.480), and CNR did not differ in Groups B or C (both p = 1.000). Interobserver agreement was substantial to almost perfect (kappa: 0.64-0.83; ICC: 0.89-0.93). DATA CONCLUSION:RT-FRFSE performed better with slower, regular breathing and PROPELLER with faster or irregular breathing. Pre-scan respiratory assessment may inform sequence selection. EVIDENCE LEVEL:3. TECHNICAL EFFICACY:Stage 1.
BACKGROUND:Quantitative susceptibility mapping (QSM) characterizes multiple sclerosis (MS) lesions through susceptibility changes related to iron and myelin. However, the prevalence, regional distribution, and relationship of QSM-defined lesion subtypes with choroid plexus alterations remain unclear. PURPOSE:To characterize MS lesion subtypes using QSM and evaluate their prevalence, regional distribution, and associations with choroid plexus volume changes. STUDY TYPE:Prospective longitudinal study. POPULATION:Forty-six patients with MS (44 relapsing-remitting, 2 primary progressive; mean age, 37 ± 8 years). QSM data were available for lesion classification in 32 participants. FIELD STRENGTH/SEQUENCE:1.5T MRI with 3D T2-FLAIR, 3D T1-MPRAGE, and multi-echo gradient-echo sequences for QSM. ASSESSMENT:Following whole-brain segmentation, white matter was parcellated into periventricular, juxtacortical, and deep regions. White matter lesions were segmented using the SAMSEG algorithm and classified on QSM as isointense, hypointense, hyperintense, or paramagnetic rim lesions. Lesion counts, lesion volumes, regional distribution, and associations with choroid plexus volume changes were evaluated. STATISTICAL TESTS:Normality was assessed using the Kolmogorov-Smirnov test. Spearman correlation and Mann-Whitney U tests were used. Bonferroni correction was applied for multiple comparisons. RESULTS:Eight hundred ninety lesions were identified: 512 isointense (57.1%), 65 hypointense (7.2%), 291 hyperintense (32.4%), and 29 paramagnetic rim lesions (3.2%). Isointense and hypointense lesions, potentially reflecting different stages of tissue repair, accounted for 64.3% of all lesions. Juxtacortical white matter was most frequently affected (43.4%), followed by subcortical (40.4%) and periventricular (16.2%) regions. Hyperintense lesion volume correlated significantly with baseline lesion volume (r = 0.71), follow-up lesion volume (r = 0.81), and total lesion volume (r = 0.87). Total lesion volume was significantly associated with bilateral choroid plexus volume changes (r = 0.515-0.595). DATA CONCLUSION:QSM has the potential to characterize MS lesion heterogeneity by identifying distinct lesion susceptibility patterns. Total lesion volume was associated with choroid plexus volume changes. EVIDENCE LEVEL:2. TECHNICAL EFFICACY:Stage 2.
BACKGROUND:Proton density fat fraction (PDFF) is typically measured using confounder-corrected, multi-echo, gradient-recalled-echo chemical-shift-encoded (CSE)-MRI. Ferumoxytol, an iron-based MRI contrast agent, increases liver and bone marrow (BM) R2*, a major confounder of PDFF. PURPOSE:To assess the impact of ferumoxytol-induced R2* increase on BM and liver PDFF. STUDY TYPE:Retrospective. POPULATION:Eleven healthy volunteers (6 female) imaged before and 1, 2, 4, 7, and 30 days after ferumoxytol; 25 patients (10 female) imaged before and 2-3 days after ferumoxytol. FIELD STRENGTH/SEQUENCE:1.5-T and 3.0-T; 3D GRE-CSE-MRI. ASSESSMENT:Regions of interest were placed on liver and spinal BM to measure PDFF and R2* before and after ferumoxytol. Dual-R2* simulations evaluated the effect of differential water and fat R2* relaxation on PDFF. STATISTICAL TESTS:Liver and BM PDFF and R2* were compared before and after ferumoxytol using regression analyses and F-tests. Changes from baseline were assessed with Wilcoxon signed-rank tests; p < 0.05 was considered significant. RESULTS:Among volunteers, BM R2* increased significantly after ferumoxytol, peaking at Day 1, 222 (84) s-1 at 1.5-T and 328 (106) s-1 at 3.0-T (baseline 72 [26] s-1 and 182 [64] s-1, respectively). BM PDFF increased significantly after ferumoxytol: 49.7 (11.3) % at 1.5-T and 47.5 (12.9) % at 3.0-T (baseline 41.1 [13.2] % and 38.4 [11.0] %, respectively). Post-ferumoxytol liver R2* also increased, yet liver PDFF did not significantly change (p range 0.579-0.983). Among patients, post-ferumoxytol BM PDFF demonstrated a systematic positive bias relative to baseline, yet this effect was not observed in liver. Simulations confirmed that a differential increase in water relative to fat R2* results in PDFF bias when a single-R2* model is fitted. DATA CONCLUSION:BM PDFF demonstrates a systematic bias in the presence of ferumoxytol-induced R2* increase and should be interpreted with caution. This bias was not observed in the liver. EVIDENCE LEVEL:3. TECHNICAL EFFICACY:Stage 1.
BACKGROUND:Cerebrovascular aging is characterized by progressive reductions in cerebral blood flow and vascular compliance. Although long-term endurance exercise has been associated with improved brain health, its relationship with cerebrovascular aging remains incompletely understood. PURPOSE:To assess associations between long-term endurance exercise and intracranial arterial structural and hemodynamic characteristics, and to determine whether these associations differ with age. STUDY TYPE:Prospective cross-sectional study. POPULATION:Fifty-one endurance runners (≥ 5 h/week endurance exercise for ≥ 3 years) and 51 matched sedentary controls. FIELD STRENGTH/SEQUENCE:Intracranial four-dimensional velocity-encoded gradient echo sequence (4D Flow MRI) at 3.0 T. ASSESSMENT:Luminal cross-sectional area, blood flow, pulsatility index (PI), and resistive index (RI) were quantified in the bilateral internal carotid arteries and basilar artery. Total cerebral blood flow (TCBF) was calculated from these vessels. STATISTICAL TESTS:Univariable analysis was used to compare groups. Multivariable linear regression and linear mixed-effects models were used to assess associations of age and long-term endurance exercise with cerebrovascular structural and hemodynamic parameters, with additional testing for age-by-exercise interactions. A p value < 0.05 was considered significant. RESULTS:Endurance runners had significantly higher TCBF than sedentary controls (586.2 ± 85.7 vs. 540.7 ± 85.6 mL/min). Older age was independently associated with lower cerebral blood flow (β = -2.636 mL/min per year) and higher PI (β = 0.003 per year), whereas long-term endurance exercise was independently associated with larger luminal cross-sectional area (β = 1.119 mm2) and higher TCBF (β = 52.929 mL/min) and higher vessel-level blood flow (β = 19.313 mL/min). In addition, endurance exercise was associated with higher RI (β = 0.054) but not PI (β = 0.026). No significant age-by-exercise interactions were observed. DATA CONCLUSION:Long-term endurance exercise was associated with larger intracranial arterial luminal dimensions and higher cerebral blood flow, without significant differences in age-related hemodynamic changes between groups. LEVEL OF EVIDENCE: 2: STAGE OF TECHNICAL EFFICACY:2.
BACKGROUND:Dual-type deep learning reconstruction (DLR) may improve the pancreatic MRI image quality; however, the effects of different DLR strengths at 1.5 T compared with the conventional 1.5 T and 3 T MRI without DLR remain unclear. PURPOSE:To evaluate the image quality of 1.5 T MRI reconstructed with different DLR strengths for pancreatic cystic lesions, compared with conventional 1.5 T and 3 T MRI without DLR. STUDY TYPE:Retrospective, single-center study. POPULATION:Ninety-nine patients with pancreatic cystic lesion (53 men, 46 women; age range, 29.2-93.8 years) underwent 1.5 T MRI; 54 had undergone 3 T MRI within the preceding 2 years. FIELD STRENGTH/SEQUENCE:1.5 T and 3 T MRI; axial single-shot turbo spin-echo T2-weighted imaging and heavily T2-weighted single-shot turbo spin-echo two-dimensional MRCP. The 1.5 T images were reconstructed using dual-type DLR with 80% and 50% denoising levels and compressed sensing (CS) without DLR. The 3 T images used CS without DLR. ASSESSMENT:Three blinded radiologists independently assessed eight image-quality items using a 5-point Likert scale. Signal-to-noise ratio, contrast-to-noise ratio, and edge rise distance were measured. STATISTICAL TESTS:Friedman tests followed by Wilcoxon signed-rank tests with Holm correction were used. RESULTS:At 1.5 T, 80% DLR yielded significantly higher scores than CS for all qualitative items except for texture naturalness with higher SNR (40.20 vs. 23.23) and CNR (24.67 vs. 13.57) and lower ERD (1.59 vs. 1.98 mm). ERD did not differ between 80% and 50% DLR (p = 0.825). Compared with 3 T CS, 1.5 T 80% DLR showed significantly higher scores for seven of eight items, significantly higher SNR and CNR, and lower ERD (39.69 vs. 30.62; CNR, 24.33 vs. 16.17; and 1.49 vs. 2.16 mm, respectively). Texture naturalness was significantly lower with 80% DLR, indicating a trade-off between image enhancement and naturalness. DATA CONCLUSION:In this exploratory, non-simultaneous comparison, 1.5 T DLR showed favorable image-quality metrics relative to 3 T CS without DLR. EVIDENCE LEVEL:3. TECHNICAL EFFICACY:Stage 1.
BACKGROUND:Deep learning reconstruction can shorten breath-hold MRI for liver proton density fat fraction (DL-PDFF), but agreement with conventional PDFF (Conv-PDFF) and the impact of measurement approach (regional vs. whole-liver segmentation) remain unclear. PURPOSE:To evaluate linearity and agreement of DL-PDFF with Conv-PDFF and MR spectroscopy PDFF (MRS-PDFF), and reconstruction-dependent whole-liver distribution metrics. STUDY TYPE:Retrospective. POPULATION:Fat phantom three vials per nominal fat mass fraction (27.86%, 18.32%, and 9.12%); 96 adults (53 males; median age, 64). FIELD STRENGTH/SEQUENCE:3 T; Chemical shift-encoded multi-echo gradient-echo; acquisition time, 10 s (DL-PDFF; acceleration factor [AF], 6) and 14 s (Conv-PDFF; AF, 4). ASSESSMENT:Central region of interest (ROI) for phantom PDFF; In vivo, single ROI by radiology technologists, multiple ROIs by a resident and a medical student, and automated whole-liver segmentation, yielding voxel-based histograms and residual fitting error; inter-observer agreement. STATISTICAL TESTS:Wilcoxon signed-rank tests, linear regression (slope/intercept/R2), Bland-Altman bias and 95% limits of agreement (LoA) with proportional-bias slope (β), Lin's concordance correlation coefficient (CCC), Hodges-Lehmann method, and Spearman (ρ). Two-sided p < 0.05. RESULTS:DL-PDFF versus Conv-PDFF showed excellent phantom linearity (R2 = 1.00). In patients, within the same approach: single ROI (bias, -0.24%; widest LoA, -2.76 to 2.28; CCC, 0.94); multiple ROIs (bias, -0.002; LoA, -1.04 to 1.04; CCC, 0.99); and whole-liver segmentation (bias, 0.12; LoA, -0.54 to 0.78; CCC, 0.99). Cross-method comparison (DL-PDFF segmentation vs. Conv-PDFF multiple ROIs) showed larger bias (0.57; LoA, -0.77 to 1.91; CCC, 0.97). Against MRS-PDFF, bias was similar (-0.05 to -0.02; LoA, -3.00 to 2.96). For DL-PDFF, whole-liver SD (Hodges-Lehmann difference, 0.55%; ρ = 0.95) and residual fitting error significantly higher (median, 2.80 vs. 1.92). Inter-observer agreement was high (CCC, 0.99). DATA CONCLUSION:DL-PDFF demonstrated high agreement with Conv-PDFF while reducing acquisition time. However, measurement approach (segmentation vs. ROI) contributes larger systematic differences relevant to longitudinal and cross-study comparisons. EVIDENCE LEVEL:3. STAGE OF TECHNICAL EFFICACY:2.
