Accurate characterization of in-utero brain development is essential for understanding typical and atypical neurodevelopment. Building upon previous efforts to construct spatiotemporal fetal brain MRI atlases, we present the CRL-2025 fetal brain atlas, which is a spatiotemporal (4D) atlas of the developing fetal brain between 21 and 37 gestational weeks. This atlas is constructed from carefully processed MRI scans of 160 fetuses with typically-developing brains using a diffeomorphic deformable registration framework integrated with kernel regression on age. CRL-2025 uniquely includes detailed tissue segmentations, transient white matter compartments, and parcellation into 126 anatomical regions. This atlas offers significantly enhanced anatomical details over the CRL-2017 atlas, and is released along with the CRL diffusion MRI atlas with its newly created tissue segmentation and labels as well as deep learning-based multiclass segmentation models for fine-grained fetal brain MRI segmentation. The CRL-2025 atlas and its associated tools provide a robust and scalable platform for fetal brain MRI segmentation, groupwise analysis, and early neurodevelopmental research, and these materials are publicly released to support the broader research community.
Journal Papers
M
Bagheri, Mahdi, Clemente Velasco-Annis, Jian Wang, Razieh Faghihpirayesh, Shadab Khan, Camilo Calixto, Camilo Jaimes, et al. (2025) 2025. “An MRI Atlas of the Human Fetal Brain: Reference and Segmentation Tools for Fetal Brain MRI Analysis.”. ArXiv.
L
Coll-Font, Jaume, Onur Afacan, Jeanne Chow, and Sila Kurugol. (2019) 2019. “Linear Time Invariant Model Based Motion Correction (LiMo-MoCo) of Dynamic Radial Contrast Enhanced MRI”. Med Image Comput Comput Assist Interv 11765: 430-37. https://doi.org/10.1007/978-3-030-32245-8_48.
Early identification of kidney function deterioration is essential to determine which newborn patients with dilation of the renal pelvis (hydronephrosis) should undergo surgery. Kidney function can be measured by fitting a tracer kinetic (TK) model onto a series of Dynamic Contrast Enhanced (DCE) MR images and deriving the glomerular filtration rate (GFR) from the TK model. Unfortunately, heavy breathing and large bulk motion events create outliers and misalignments that introduce large errors in the TK estimates. Moreover, aligning the series of DCE images is not trivial due to the contrast differences between them and the undersampling artifacts due to fast imaging. We present a bulk motion detection and a linear time invariant (LTI) model-based motion correction approach for DCE-MRI alignment that leverages the temporal dynamics of the DCE data at each voxel. We evaluate our approach on 10 newborn patients that underwent DCE imaging without sedation. For each patient, we reconstructed the sequence of DCE images, detected and removed the volumes corrupted by motion using a self navigation approach, aligned the sequence using our approach and fitted the TK model to compute GFR. The results show that our approach correctly aligned all volumes and improved the TK model fit and, on average, reducing the normalized root-mean-squared error by 0.17.
Koçanaoğullari, Aziz, Cemre Ariyurek, Onur Afacan, and Sila Kurugol. (2022) 2022. “Learning the Regularization in DCE-MR Image Reconstruction for Functional Imaging of Kidneys”. IEEE Access 10: 4102-11. https://doi.org/10.1109/access.2021.3139854.
Kidney DCE-MRI aims at both qualitative assessment of kidney anatomy and quantitative assessment of kidney function by estimating the tracer kinetic (TK) model parameters. Accurate estimation of TK model parameters requires an accurate measurement of the arterial input function (AIF) with high temporal resolution. Accelerated imaging is used to achieve high temporal resolution, which yields under-sampling artifacts in the reconstructed images. Compressed sensing (CS) methods offer a variety of reconstruction options. Most commonly, sparsity of temporal differences is encouraged for regularization to reduce artifacts. Increasing regularization in CS methods removes the ambient artifacts but also over-smooths the signal temporally which reduces the parameter estimation accuracy. In this work, we propose a single image trained deep neural network to reduce MRI under-sampling artifacts without reducing the accuracy of functional imaging markers. Instead of regularizing with a penalty term in optimization, we promote regularization by generating images from a lower dimensional representation. In this manuscript we motivate and explain the lower dimensional input design. We compare our approach to CS reconstructions with multiple regularization weights. Proposed approach results in kidney biomarkers that are highly correlated with the ground truth markers estimated using the CS reconstruction which was optimized for functional analysis. At the same time, the proposed approach reduces the artifacts in the reconstructed images.
Sui, Yao, Onur Afacan, Ali Gholipour, and Simon Warfield. (2020) 2020. “Learning a Gradient Guidance for Spatially Isotropic MRI Super-Resolution Reconstruction”. Med Image Comput Comput Assist Interv 12262: 136-46. https://doi.org/10.1007/978-3-030-59713-9_14.
In MRI practice, it is inevitable to appropriately balance between image resolution, signal-to-noise ratio (SNR), and scan time. It has been shown that super-resolution reconstruction (SRR) is effective to achieve such a balance, and has obtained better results than direct high-resolution (HR) acquisition, for certain contrasts and sequences. The focus of this work was on constructing images with spatial resolution higher than can be practically obtained by direct Fourier encoding. A novel learning approach was developed, which was able to provide an estimate of the spatial gradient prior from the low-resolution (LR) inputs for the HR reconstruction. By incorporating the anisotropic acquisition schemes, the learning model was trained over the LR images themselves only. The learned gradients were integrated as prior knowledge into a gradient-guided SRR model. A closed-form solution to the SRR model was developed to obtain the HR reconstruction. Our approach was assessed on the simulated data as well as the data acquired on a Siemens 3T MRI scanner containing 45 MRI scans from 15 subjects. The experimental results demonstrated that our approach led to superior SRR over state-of-the-art methods, and obtained better images at lower or the same cost in scan time than direct HR acquisition.
I
Sui, Yao, Onur Afacan, Ali Gholipour, and Simon Warfield. (2019) 2019. “Isotropic MRI Super-Resolution Reconstruction With Multi-Scale Gradient Field Prior”. Med Image Comput Comput Assist Interv 11766: 3-11. https://doi.org/10.1007/978-3-030-32248-9_1.
In this work, we proposed a novel image-based MRI super-resolution reconstruction (SRR) approach based on anisotropic acquisition schemes. We achieved superior reconstruction to state-of-the-art work by introducing a new multi-scale gradient field prior that guides the reconstruction of the high-resolution (HR) image. The prior improves both spatial smoothness and edge preservation. The inverse of the forward model of image formation is used to propagate the gradient guidance from the low-resolution (LR) images to the HR image space. The gradient fields over multiple scales were exploited for more accurate edge localization in the reconstruction. The proposed SRR allows inter-volume motion during the MRI scans and can incorporate with the LR images with arbitrary orientations and displacements in the frequency space, such as orthogonal and origin-shifted scans. The proposed approach was evaluated on the synthetic data as well as the data acquired on a Siemens 3T MRI scanner containing 45 MRI scans from 14 subjects. The evaluation results demonstrate that our proposed prior leads to improved SRR as compared to state-of-the-art priors, and that the proposed SRR obtains better results at lower or the same cost in scan time than direct HR acquisition. In particular, the anatomical structures of hippocampus can be clearly shown in our reconstructed images. This is a significant improvement for the in vivo studies of the hippocampus.
Asaturyan, Hykoush, Barbara Villarini, Karen Sarao, Jeanne Chow, Onur Afacan, and Sila Kurugol. 2021. “Improving Automatic Renal Segmentation in Clinically Normal and Abnormal Paediatric DCE-MRI via Contrast Maximisation and Convolutional Networks for Computing Markers of Kidney Function”. Sensors (Basel) 21 (23). https://doi.org/10.3390/s21237942.
There is a growing demand for fast, accurate computation of clinical markers to improve renal function and anatomy assessment with a single study. However, conventional techniques have limitations leading to overestimations of kidney function or failure to provide sufficient spatial resolution to target the disease location. In contrast, the computer-aided analysis of dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) could generate significant markers, including the glomerular filtration rate (GFR) and time-intensity curves of the cortex and medulla for determining obstruction in the urinary tract. This paper presents a dual-stage fully modular framework for automatic renal compartment segmentation in 4D DCE-MRI volumes. (1) Memory-efficient 3D deep learning is integrated to localise each kidney by harnessing residual convolutional neural networks for improved convergence; segmentation is performed by efficiently learning spatial-temporal information coupled with boundary-preserving fully convolutional dense nets. (2) Renal contextual information is enhanced via non-linear transformation to segment the cortex and medulla. The proposed framework is evaluated on a paediatric dataset containing 60 4D DCE-MRI volumes exhibiting varying conditions affecting kidney function. Our technique outperforms a state-of-the-art approach based on a GrabCut and support vector machine classifier in mean dice similarity (DSC) by 3.8% and demonstrates higher statistical stability with lower standard deviation by 12.4% and 15.7% for cortex and medulla segmentation, respectively.
Freiman, Moti, Onur Afacan, Robert Mulkern, and Simon Warfield. (2013) 2013. “Improved Multi B-Value Diffusion-Weighted MRI of the Body by Simultaneous Model Estimation and Image Reconstruction (SMEIR)”. Med Image Comput Comput Assist Interv 16 (Pt 3): 1-8. https://doi.org/10.1007/978-3-642-40760-4_1.
Diffusion-weighted MRI images acquired with multiple b-values have the potential to improve diagnostic accuracy by increasing the conspicuity of lesions and inflammatory activity with background suppression. Unfortunately, the inherently low signal-to-noise ratio (SNR) of DW-MRI reduces enthusiasm for using these images for diagnostic purposes. Moreover, lengthy acquisition times limit our ability to improve the quality of multi b-value DW-MRI images by multiple excitations acquisition and signal averaging at each b-value. To offset these limitations, we propose the Simultaneous Model Estimation and Image Reconstruction (SMEIR) for DW-MRI, which substantially improves the quality of multi b-value DW-MRI images without increasing acquisition times. Our model introduces the physiological signal decay model of DW-MRI as a constraint in the reconstruction of the DW-MRI images. An in-vivo experiment using 6 low-quality DW-MRI datasets of a healthy subject showed that SMEIR reconstruction of low-quality data improved SNR by 55% in the liver and by 41% in the kidney without increasing acquisition times. We also demonstrated the clinical impact of our SMEIR reconstruction by increasing the conspicuity of inflamed bowel regions in DW-MRI of 12 patients with Crohn's disease. The contrast-to-noise ratio (CNR) of the inflamed regions in the SMEIR images was higher by 12.6% relative to CNR in the original DW-MRI images.
Ariyurek, Cemre, Lina Lu, Georgios Antonios Sideris, Valentina Valencia Ferrer, Liam Timms, Serge Didenko Vasylechko, Onur Afacan, and Sila Kurugol. (2025) 2025. “Improved IVIM Imaging in Adolescent Crohn’s Disease Using Dual-Echo EPI Distortion and Motion Correction.”. NMR in Biomedicine 38 (9): e70117. https://doi.org/10.1002/nbm.70117.
Diffusion-weighted imaging (DWI) during MR enterography helps identify bowel inflammation in Crohn's disease (CD). However, image quality is compromised by geometric distortions from B0 field variations and physiological motion, making it challenging for radiologists to correlate findings between DWI and structural images. Traditional correction methods using reversed polarity scans are ineffective due to motion between acquisitions, which limits accurate estimation of intravoxel incoherent motion (IVIM) parameters. We propose a dual-echo echo-planar imaging (EPI) method that retrospectively corrects both geometric distortions and motion in 3T bowel DWI by accounting for field changes during peristalsis and breathing. We added a 5- to 7-min dual-echo EPI DW sequence (eight b-values, six directions) to the clinical MR enterography protocol of 21 patients with suspected CD at 3T MRI. Distortion correction was applied based on dynamically estimated fields from dual-echo DWI, followed by intra-volume registration between odd-even slices and inter-volume registration for motion correction. Two experienced board-certified radiologists evaluated the severity of the disease using simplified magnetic resonance index of activity (MaRIA) scores. Based on their consensus scores, patients were categorized into three groups: no active disease (MaRIA score = 0), active disease (MaRIA score = 1-2), and severe disease (MaRIA score = 3-5). The proposed DWI correction pipeline improved DWI/T2-weighted image Dice similarity from 0.73 to 0.89, enabling better correlation of findings between structural and DW-MR images and enhancing DWI's clinical value. Corrected IVIM parameters showed stronger correlations with MaRIA scores (D: ρ = -0.93; f: ρ = -0.94, p < 0.001) compared to uncorrected parameters (D: ρ = -0.68, p = 0.001; f: ρ = -0.35, p = 0.118). Diagnostic sensitivity increased from 0.44 to 0.89, while parameter uncertainty decreased from 35.58% to 19.31% for D and 63.48% to 40.40% for f (p < 0.001). These improvements strengthen quantitative IVIM imaging for CD assessment, potentially reducing reliance on contrast imaging while offering enhanced tissue perfusion and diffusion insights.
Ahtam, Banu, Susan Waisbren, Vera Anastasoaie, Gerard Berry, Matthew Brown, Stephanie Petrides, Onur Afacan, et al. 2020. “Identification of Neuronal Structures and Pathways Corresponding to Clinical Functioning in Galactosemia”. J Inherit Metab Dis 43 (6): 1205-18. https://doi.org/10.1002/jimd.12279.
Classic galactosemia (OMIM# 230400) is an autosomal recessive disorder due to galactose-1-phosphate uridyltransferase deficiency. Newborn screening and prompt treatment with a galactose-free diet prevent the severe consequences of galactosemia, but clinical outcomes remain suboptimal. Five men and five women with classic galactosemia (mean age = 27.2 ± 5.47 years) received comprehensive neurological and neuropsychological evaluations, electroencephalogram (EEG) and magnetic resonance imaging (MRI). MRI data from nine healthy controls (mean age = 30.22 ± 3.52 years) were used for comparison measures. Galactosemia subjects experienced impaired memory, language processing, visual-motor skills, and increased anxiety. Neurological examinations revealed tremor and dysarthria in six subjects. In addition, there was ataxia in three subjects and six subjects had abnormal gait. Mean full scale IQ was 80.4 ± 17.3. EEG evaluations revealed right-sided abnormalities in five subjects and bilateral abnormalities in one subject. Compared to age- and gender-matched controls, subjects with galactosemia had reduced volume in left cerebellum white matter, bilateral putamen, and left superior temporal sulcus. Galactosemia patients also had lower fractional anisotropy and higher radial diffusivity values in the dorsal and ventral language networks compared to the controls. Furthermore, there were significant correlations between neuropsychological test results and the T1 volume and diffusivity scalars. Our findings help to identify anatomic correlates to motor control, learning and memory, and language in subjects with galactosemia. The results from this preliminary assessment may provide insights into the pathophysiology of this inborn error of metabolism.
H
Wallace, Tess, Onur Afacan, Maryna Waszak, Tobias Kober, and Simon Warfield. 2019. “Head Motion Measurement and Correction Using FID Navigators”. Magn Reson Med 81 (1): 258-74. https://doi.org/10.1002/mrm.27381.
PURPOSE: To develop a novel framework for rapid, intrinsic head motion measurement in MRI using FID navigators (FIDnavs) from a multichannel head coil array.
METHODS: FIDnavs encode substantial rigid-body motion information; however, current implementations require patient-specific training with external tracking data to extract quantitative positional changes. In this work, a forward model of FIDnav signals was calibrated using simulated movement of a reference image within a model of the spatial coil sensitivities. A FIDnav module was inserted into a nonselective 3D FLASH sequence, and rigid-body motion parameters were retrospectively estimated every readout time using nonlinear optimization to solve the inverse problem posed by the measured FIDnavs. This approach was tested in simulated data and in 7 volunteers, scanned at 3T with a 32-channel head coil array, performing a series of directed motion paradigms.
RESULTS: FIDnav motion estimates achieved mean absolute errors of 0.34 ± 0.49 mm and 0.52 ± 0.61° across all subjects and scans, relative to ground-truth motion measurements provided by an electromagnetic tracking system. Retrospective correction with FIDnav motion estimates resulted in substantial improvements in quantitative image quality metrics across all scans with intentional head motion.
CONCLUSIONS: Quantitative rigid-body motion information can be effectively estimated using the proposed FIDnav-based approach, which represents a practical method for retrospective motion compensation in less cooperative patient populations.