Yanniklas Kravutske
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yanniklas Kravutske has a total of 29 co-authors. Here are the top co-authors by number of shared publications:
- Bruno Passebon Soares
Department of Radiology, Stanford University School of Medicine, Stanford, United States
4 shared publications - Mateus Aragão Esmeraldo
Stanford University School of Medicine
4 shared publications - Stefanie Chambers
Department of Neurosurgery, Medical University of Vienna, Vienna, Austria
4 shared publications - Lukas Haider
Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna, Vienna, Austria
3 shared publications - Ana Filipa Geraldo
Department of Radiology, Pediatric Neuroradiology Unit, Lucile Packard Children’s Hospital, Stanford University School of Medicine, Palo Alto, USA
2 shared publications - Eduardo Pontes Reis
Department of Radiology, Stanford University School of Medicine, Stanford, CA, USA
2 shared publications
Latest Publications
- Open Access
- 14 October 2026
- Published by MDPI AG in Journal of Imaging
- Vol. 12 (10) , 510
Deep Learning-Based Multiple Sclerosis Lesion Segmentation at 7 Tesla: Comparison of Single- and Multi-Contrast MRI Inputs
Automated lesion segmentation can support rapid and reproducible quantitative assessment and long-term monitoring of multiple sclerosis. Ultra-high-field 7T MRI provides increased spatial resolution and lesion visibility but introduces challenges for automated analysis, including field inhomogeneity, susceptibility-related artifacts, and spatially varying image contrast. Although deep learning methods perform well at conventional field strengths, their application to 7T MRI remains underexplored. We evaluated nnUNet using FLAIR alone, MP2RAGE alone, and combined MP2RAGE+FLAIR. The dataset included 115 longitudinal 7T MRI scans from 57 patients with MS, with 20 scans from 11 independent patients reserved for testing. DeepMedic was included as a comparative architecture. On the test set, nnUNet achieved mean DSC values of 0.72, 0.71, and 0.66 for combined, MP2RAGE-only, and FLAIR-only inputs, respectively. No significant DSC differences were observed between nnUNet and DeepMedic. For both architectures, adding FLAIR to MP2RAGE did not show a statistically significant additional benefit in the present test cohort, whereas MP2RAGE-containing inputs significantly outperformed FLAIR-only input. The findings of the present study provide preliminary evidence suggesting that the observed voxel-level segmentation performance was primarily associated with the inclusion of MP2RAGE. However, lesion-level performance depended on lesion size, indicating further investigation of contrast-specific effects on small-lesion detection.
- 14 October 2026
- Published by MDPI AG in Journal of Imaging
- Vol. 12 (10) , 510
Open AccessDeep Learning-Based Multiple Sclerosis Lesion Segmentation at 7 Tesla: Comparison of Single- and Multi-Contrast MRI Inputs
Automated lesion segmentation can support rapid and reproducible quantitative assessment and long-term monitoring of multiple sclerosis. Ultra-high-field 7T MRI provides increased spatial resolution and lesion visibility but introduces challenges for automated analysis, including field inhomogeneity, susceptibility-related artifacts, and spatially varying image contrast. Although deep learning methods perform well at conventional field strengths, their application to 7T MRI remains underexplored. We evaluated nnUNet using FLAIR alone, MP2RAGE alone, and combined MP2RAGE+FLAIR. The dataset included 115 longitudinal 7T MRI scans from 57 patients with MS, with 20 scans from 11 independent patients reserved for testing. DeepMedic was included as a comparative architecture. On the test set, nnUNet achieved mean DSC values of 0.72, 0.71, and 0.66 for combined, MP2RAGE-only, and FLAIR-only inputs, respectively. No significant DSC differences were observed between nnUNet and DeepMedic. For both architectures, adding FLAIR to MP2RAGE did not show a statistically significant additional benefit in the present test cohort, whereas MP2RAGE-containing inputs significantly outperformed FLAIR-only input. The findings of the present study provide preliminary evidence suggesting that the observed voxel-level segmentation performance was primarily associated with the inclusion of MP2RAGE. However, lesion-level performance depended on lesion size, indicating further investigation of contrast-specific effects on small-lesion detection.
- 10 March 2026
- Published by American Society of Neuroradiology (ASNR) in American Journal of Neuroradiology
- p. ajnr.A9282
Microcephaly with Simplified Gyral Pattern in Children: Quantitative Morphometric Assessment on Brain MRI and Correlation with Clinical Outcome
Microcephaly with simplified gyral pattern (MSGP) represents an often subtle subtype of malformation of cortical development that is commonly accompanied by epilepsy and neurodevelopmental delay. However, whether gyral simpflication scales to reduced brain volume, and if MRI features predict clinical outcome still remains unclear. The aims of this study are therefore to apply a quantitative morphometric approach to distinguish simplified gyral pattern from microcephaly-related effects, and to correlate quantitative MRI-derived features with clinical outcome. From a single-center cohort of 34 patients with microcephaly and visually confirmed simplified gyral pattern (SGP) (shallow sulci, normal cortical thickness), 18 subjects (mean age 7.1±6.5 years) that fulfilled the eligibility criteria were retrospectively included for final analysis. 26 age- and sex-matched healthy controls were included. Volumetric gray/white-matter volumes (GMV/WMV) and surface-based measures were derived using FreeSurfer. Scaling corrections based on log-log robust regression and cortical folding analyses were subsequently analyzed. Principal component analyses clustered along morphometric phenotypes. ANCOVA, Kendall τ correlations, and logistic regression were applied to evaluated group differences and clinical outcome predictors. Compared with healthy controls, patients showed significantly reduced WMV, cortical surface area, local gyrification index (LGI) (all q<0.003), and mildly increased cortical thickness. Allometric modeling indicated that while WMV and LGI scaled proportionally to total brain volume, cortical surface area was disproportionately reduced, suggesting deviation beyond microcephalic scaling. Spectral folding analysis showed loss of specifically macro-scale (primary) folding power. Morphometric clustering separated four subtypes differing in cortical thickness and area, of which the cluster with lowest surface area and LGI exhibited universal developmental delay and included both early deaths. Across the cohort, neurodevelopmental delay correlated negatively with surface area. Logistic regression suggested lower surface area, WMV, and LGI as predictors of adverse clinical outcome (q<0.02). The presence of SGP may reflect additional disturbance of primary sulcation and tangential cortical expansion beyond scaling to microcephaly alone. Our results suggest reduced cortical surface area as a key morphometric marker and strong MRI feature predictive of neurodevelopmental delay, underscoring the clinical value of quantitative morphometry for stratifying disease severity in microcephaly with simplified gyral pattern.
- 10 March 2026
- Published by American Society of Neuroradiology (ASNR) in American Journal of Neuroradiology
- p. ajnr.A9282
Microcephaly with Simplified Gyral Pattern in Children: Quantitative Morphometric Assessment on Brain MRI and Correlation with Clinical Outcome
Microcephaly with simplified gyral pattern (MSGP) represents an often subtle subtype of malformation of cortical development that is commonly accompanied by epilepsy and neurodevelopmental delay. However, whether gyral simpflication scales to reduced brain volume, and if MRI features predict clinical outcome still remains unclear. The aims of this study are therefore to apply a quantitative morphometric approach to distinguish simplified gyral pattern from microcephaly-related effects, and to correlate quantitative MRI-derived features with clinical outcome. From a single-center cohort of 34 patients with microcephaly and visually confirmed simplified gyral pattern (SGP) (shallow sulci, normal cortical thickness), 18 subjects (mean age 7.1±6.5 years) that fulfilled the eligibility criteria were retrospectively included for final analysis. 26 age- and sex-matched healthy controls were included. Volumetric gray/white-matter volumes (GMV/WMV) and surface-based measures were derived using FreeSurfer. Scaling corrections based on log-log robust regression and cortical folding analyses were subsequently analyzed. Principal component analyses clustered along morphometric phenotypes. ANCOVA, Kendall τ correlations, and logistic regression were applied to evaluated group differences and clinical outcome predictors. Compared with healthy controls, patients showed significantly reduced WMV, cortical surface area, local gyrification index (LGI) (all q<0.003), and mildly increased cortical thickness. Allometric modeling indicated that while WMV and LGI scaled proportionally to total brain volume, cortical surface area was disproportionately reduced, suggesting deviation beyond microcephalic scaling. Spectral folding analysis showed loss of specifically macro-scale (primary) folding power. Morphometric clustering separated four subtypes differing in cortical thickness and area, of which the cluster with lowest surface area and LGI exhibited universal developmental delay and included both early deaths. Across the cohort, neurodevelopmental delay correlated negatively with surface area. Logistic regression suggested lower surface area, WMV, and LGI as predictors of adverse clinical outcome (q<0.02). The presence of SGP may reflect additional disturbance of primary sulcation and tangential cortical expansion beyond scaling to microcephaly alone. Our results suggest reduced cortical surface area as a key morphometric marker and strong MRI feature predictive of neurodevelopmental delay, underscoring the clinical value of quantitative morphometry for stratifying disease severity in microcephaly with simplified gyral pattern.
- Open Access
Image Quality Evaluation of Neonatal Brain MRI Using a Deep Learning Reconstruction Algorithm: A Quantitative and Multireader Study Using Variable Denoising Levels at 3 Tesla
Purpose: Neonatal imaging is particularly challenging because newborns have a high likelihood of head motion, which can degrade image quality and complicate interpretation. Improving MRI brain image quality may help reduce diagnostic uncertainty and facilitate the nuanced assessment of early myelinating structures in the neonatal brain. Although deep learning reconstruction algorithms designed to improve MRI image quality have been evaluated in pediatric imaging, they have not been specifically studied in exclusively neonatal populations. We sought to evaluate image quality improvement through the employment of a deep learning reconstruction algorithm in neonatal brain imaging. Methods: 3D T1-weighted brain MRIs were obtained in 15 neonates. A deep-learning reconstruction algorithm was applied to the image sets using low, medium, and high levels of denoising. Three radiologists qualitatively rated image quality (signal-to-noise ratio, presence of artifacts, and overall clarity) on a 4-point scale of eight early myelinating structures. Objective apparent signal-to-noise ratio (aSNR) and apparent contrast-to-noise ratio (aCNR), based on signal intensities of white-and gray-matter, was measured across all three denoising levels. Results: Evaluation by radiologists indicated an overall increase in all image quality categories and increased conspicuity of the early myelinating structures as the level of denoising increased. Objective aSNR and aCNR values also increased progressively with denoising, with significant differences observed for nearly all pairwise comparisons. Conclusion: Our findings suggest that the use of the proposed deep learning reconstruction algorithm improves image quality in 3D T1-weighted neonatal brain MRIs at 3T.
Open AccessImage Quality Evaluation of Neonatal Brain MRI Using a Deep Learning Reconstruction Algorithm: A Quantitative and Multireader Study Using Variable Denoising Levels at 3 Tesla
Purpose: Neonatal imaging is particularly challenging because newborns have a high likelihood of head motion, which can degrade image quality and complicate interpretation. Improving MRI brain image quality may help reduce diagnostic uncertainty and facilitate the nuanced assessment of early myelinating structures in the neonatal brain. Although deep learning reconstruction algorithms designed to improve MRI image quality have been evaluated in pediatric imaging, they have not been specifically studied in exclusively neonatal populations. We sought to evaluate image quality improvement through the employment of a deep learning reconstruction algorithm in neonatal brain imaging. Methods: 3D T1-weighted brain MRIs were obtained in 15 neonates. A deep-learning reconstruction algorithm was applied to the image sets using low, medium, and high levels of denoising. Three radiologists qualitatively rated image quality (signal-to-noise ratio, presence of artifacts, and overall clarity) on a 4-point scale of eight early myelinating structures. Objective apparent signal-to-noise ratio (aSNR) and apparent contrast-to-noise ratio (aCNR), based on signal intensities of white-and gray-matter, was measured across all three denoising levels. Results: Evaluation by radiologists indicated an overall increase in all image quality categories and increased conspicuity of the early myelinating structures as the level of denoising increased. Objective aSNR and aCNR values also increased progressively with denoising, with significant differences observed for nearly all pairwise comparisons. Conclusion: Our findings suggest that the use of the proposed deep learning reconstruction algorithm improves image quality in 3D T1-weighted neonatal brain MRIs at 3T.
- 26 January 2026
- Published by IOP Publishing in Biomedical Physics & Engineering Express
- Vol. 12 (1)
Comprehensive segmentation of focal cortical dysplasia by combining surface-based and whole-brain MRI deep learning algorithms: a proof-of-concept study
Introduction: Focal cortical dysplasia type II (FCD II) is a significant cause of drug-resistant epilepsy, and the full surgical resection of the lesion is linked with excellent disease-free outcomes. Its imaging hallmark is the white matter hyperintense funnel-shaped transmantle sign on T2-FLAIR magnetic resonance imaging (MRI). Manual delineation of this abnormality is challenging and inconsistent. Most current artificial intelligence (AI) segmentation tools focus on cortical features and do not fully evaluate the white matter component. We tested whether integrating an algorithm trained on white matter lesions may improve FCD II segmentation. Methods: We evaluated the combination of two AI algorithms, MELD Graph (surface-based FCD segmentation) and MindGlide (whole-brain/white-matter lesion segmentation tool) in 49 FCD cases with a radiologically confirmed transmantle sign. Segmentation accuracy was assessed against expert manual annotations using the Dice similarity coefficient and segmentation volumes. Results: MELD Graph detected the lesion in 31 cases, 22 of which had the transmantle sign included in the expert lesion mask. Among these, MindGlide detected the transmantle sign in eight cases (36%). The mean added Dice score was 0.033 (95% CI, 0.013–0.056). Overall Dice values of MELD Graph were 0.321 and increased to 0.354 with the addition of MindGlide. It also contributed additional lesion volume in these eight cases, ranging from 0.028 to 4.18 cm³, with a mean added volume of 0.77 cm³. Discussion: Despite not being trained on FCD data, MindGlide, when combined with MELD Graph, provided a modest improvement in FCD II segmentation, including the deep white matter component of the lesion that is not captured by MELD Graph. Conclusion: These findings provide preliminary evidence supporting the consideration of a sequential cortical and white matter segmentation approach in FCD II, which may guide further epilepsy-specific AI model development.
- 26 January 2026
- Published by IOP Publishing in Biomedical Physics & Engineering Express
- Vol. 12 (1)
Comprehensive segmentation of focal cortical dysplasia by combining surface-based and whole-brain MRI deep learning algorithms: a proof-of-concept study
Introduction: Focal cortical dysplasia type II (FCD II) is a significant cause of drug-resistant epilepsy, and the full surgical resection of the lesion is linked with excellent disease-free outcomes. Its imaging hallmark is the white matter hyperintense funnel-shaped transmantle sign on T2-FLAIR magnetic resonance imaging (MRI). Manual delineation of this abnormality is challenging and inconsistent. Most current artificial intelligence (AI) segmentation tools focus on cortical features and do not fully evaluate the white matter component. We tested whether integrating an algorithm trained on white matter lesions may improve FCD II segmentation. Methods: We evaluated the combination of two AI algorithms, MELD Graph (surface-based FCD segmentation) and MindGlide (whole-brain/white-matter lesion segmentation tool) in 49 FCD cases with a radiologically confirmed transmantle sign. Segmentation accuracy was assessed against expert manual annotations using the Dice similarity coefficient and segmentation volumes. Results: MELD Graph detected the lesion in 31 cases, 22 of which had the transmantle sign included in the expert lesion mask. Among these, MindGlide detected the transmantle sign in eight cases (36%). The mean added Dice score was 0.033 (95% CI, 0.013–0.056). Overall Dice values of MELD Graph were 0.321 and increased to 0.354 with the addition of MindGlide. It also contributed additional lesion volume in these eight cases, ranging from 0.028 to 4.18 cm³, with a mean added volume of 0.77 cm³. Discussion: Despite not being trained on FCD data, MindGlide, when combined with MELD Graph, provided a modest improvement in FCD II segmentation, including the deep white matter component of the lesion that is not captured by MELD Graph. Conclusion: These findings provide preliminary evidence supporting the consideration of a sequential cortical and white matter segmentation approach in FCD II, which may guide further epilepsy-specific AI model development.
- Open Access
Pipeline Evaluation of a State-of-the-Art AI Algorithm for Detection of Focal Cortical Dysplasia: Insights into Potential Failure Sources
Purpose: MELD Graph is a state-of-the-art artificial intelligence (AI) model for automated detection of focal cortical dysplasia (FCD), but its performance remains limited, highlighting the need to investigate which aspects of the pipeline affect its accuracy. Methods: A retrospective failure-mode analysis of the MELD Graph pipeline was performed in 242 subjects, with model predictions and FreeSurfer segmentations reviewed to classify errors as segmentation-associated or algorithm-related. FCD imaging features salient to humans were quantified, with statistical associations examined for both MELD Graph detection and focal FreeSurfer segmentation failure. Results: MELD Graph demonstrated overall performance similar to previously published non-harmonized results, achieving a sensitivity of 69%, specificity of 44%, and positive predictive value (PPV) of 75%. Focal FreeSurfer segmentation failures were associated with 21% of false negative patients, 25% of false positive clusters in patients, and 16% of false positive clusters in controls. Higher conspicuity on T1-weighted images was associated with MELD Graph detection, whereas greater conspicuity on T2-FLAIR images relative to T1 was associated with detection failure. Bottom-of-sulcus dysplasia (BOSD) and presence of transmantle sign were not associated with detection. Non-BOSD lesions, higher human conspicuity measures, and low T1 image quality were positively associated with focal FreeSurfer segmentation failures. Conclusion: FreeSurfer segmentation failures are a significant potential source of error in the MELD Graph pipeline. FCD imaging features salient to humans and image quality were also associated with variability in the algorithm performance. Robust cortical segmentation and stronger integration of T2-FLAIR imaging features may be beneficial for automated FCD detection tools.
Open AccessPipeline Evaluation of a State-of-the-Art AI Algorithm for Detection of Focal Cortical Dysplasia: Insights into Potential Failure Sources
Purpose: MELD Graph is a state-of-the-art artificial intelligence (AI) model for automated detection of focal cortical dysplasia (FCD), but its performance remains limited, highlighting the need to investigate which aspects of the pipeline affect its accuracy. Methods: A retrospective failure-mode analysis of the MELD Graph pipeline was performed in 242 subjects, with model predictions and FreeSurfer segmentations reviewed to classify errors as segmentation-associated or algorithm-related. FCD imaging features salient to humans were quantified, with statistical associations examined for both MELD Graph detection and focal FreeSurfer segmentation failure. Results: MELD Graph demonstrated overall performance similar to previously published non-harmonized results, achieving a sensitivity of 69%, specificity of 44%, and positive predictive value (PPV) of 75%. Focal FreeSurfer segmentation failures were associated with 21% of false negative patients, 25% of false positive clusters in patients, and 16% of false positive clusters in controls. Higher conspicuity on T1-weighted images was associated with MELD Graph detection, whereas greater conspicuity on T2-FLAIR images relative to T1 was associated with detection failure. Bottom-of-sulcus dysplasia (BOSD) and presence of transmantle sign were not associated with detection. Non-BOSD lesions, higher human conspicuity measures, and low T1 image quality were positively associated with focal FreeSurfer segmentation failures. Conclusion: FreeSurfer segmentation failures are a significant potential source of error in the MELD Graph pipeline. FCD imaging features salient to humans and image quality were also associated with variability in the algorithm performance. Robust cortical segmentation and stronger integration of T2-FLAIR imaging features may be beneficial for automated FCD detection tools.