MRI Scans Research Articles

Brain tumor classification from MRI scans using fine-tuned CNN models

A rotator cuff tear (RCT) refers to a partial or complete rupture of the rotator cuff (the tendons of the supraspinatus, infraspinatus, teres minor, and subscapularis muscles). It is commonly caused by degenerative changes or trauma and manifests as shoulder pain, weakness, and limited range of motion.This study aimed to explore the correlation between inferior position of coracoid process and risk factors for rotator cuff injury. (1) Does the inferior position of the coracoid process increase the probability of risk factors leading to RCT? (2) Does a lower positioning of the coracoid process increase the probability of subcoracoid impact? (3) Does a lower positioning of the coracoid process correlate with the shape of the acromion? From May 2018 to Dec 2019, 93 patients (38 males and 55 females) with rotator cuff injuries were analyzed, with an average age of 57.7 ± 6.2 years. All patients received X-ray examination with Anterior-posterior view and supraspinatus outlet view to distinguish acromion morphology before surgery. MRI scan was used to assess the rotator cuff injury. Three-dimensional computerized tomography examination measured the position of coracoid process. The location of rotator cuff injury and the subcoracoacromial arch impingement were evaluated by arthroscopy. Pearson chi-square test was used for statistical analysis of the position of coracoid process. 76.3% (71/93) of the patients had rotator cuff tear (RCT) within 1.5 cm of the biceps tendon groove lateral, and the rates were higher in Type III/IV coracoids. Total 60 cases had subacromial impact, while 28 cases had subcoracoid impacts. Type III/IV coracoids had a higher probability of impact. The prevalence of Type III/IV coracoid processes was 22 (78.6%) in the 28 cases with type III acromion, which was significantly higher than that in cases with Type I/II acromion (P = 0.034). The inferior position of the coracoid process increases the probability of RCT, which may be related to the high incidence of type III acromion in such patients and the reduction of the subcoracoid gap. Level III, therapeutic study.

Background and Objectives: The pineal gland is a neuroendocrine structure whose function can be disrupted in patients with malignancies. This study examines the differences in pineal gland volume between oncology patients and healthy controls, as well as the relationship between volume and the duration of chemotherapy. Materials and Methods: A retrospective study included 400 participants, divided into two groups: 200 oncology patients and 200 healthy controls. The pineal gland volume was measured using MRI scans, utilizing T1-sagittal, T2-coronal/axial sections, and post-contrast 3D T1W MPRAGE tomograms. The volume was calculated based on the ellipse approximation formula: V = (L × H × W)/2. The study analyzed the relationships between pineal gland volume and factors such as age, sex, primary tumor origin, and the duration of chemotherapy. Results: The pineal gland volume was significantly smaller in oncology patients in comparison with the healthy controls (p < 0.001). The average volume in oncology patients was 32.41 ± 16.79 mm3, whereas in healthy controls, it was 59.26 ± 29.99 mm3. A significantly smaller pineal gland volume was observed in patients with malignancies, regardless of sex, with no notable differences between groups. Age also did not significantly influence gland volume (p > 0.05). The primary tumor site did not significantly influence gland volume (p > 0.05). A moderate positive correlation was observed between the duration of chemotherapy and pineal gland volume (ρ = 0.322; p = 0.007). Conclusions: The pineal gland showed reduced volume in oncology patients compared to controls. The observed positive correlation with chemotherapy duration should be interpreted cautiously and may reflect survivorship bias rather than direct treatment effects.

Visual hallucinations (VH) and mild cognitive impairment (MCI) often co-occur in Parkinson's Disease (PD). Each symptom in isolation is associated with cholinergic basal forebrain (BF) atrophy. However, it is unclear whether BF integrity relates to one symptom preferentially or to their co-occurrence, and whether this relationship is specific to the nucleus basalis of Meynert (Ch4 & Ch4p) or extends to other BF nuclei (Ch1-2 & Ch3). To bridge the literature on MCI, VH, and the BF by examining associations between total and subregional BF volumes with VH, MCI, and concurrent symptomology. We used clinical and structural MRI data in PD patients (n = 296), evaluated prior to deep brain stimulation. PD-MCI was identified in 61% of patients, VH in 25%, and symptoms co-occurred in 21% of the sample. We conducted logistic regressions to examine the relationships between BF volumes (total BF, Ch1-2, Ch3, Ch4, and Ch4p) with MCI, VH, and their combination, accounting for potential confounders (age, sex, LEDD, dopamine agonist, and anticholinergic medications). Total BF, Ch1-2, and Ch4 volumes are negatively associated with the likelihood of PD-MCI, but not VH when the other symptom type is not considered. When both symptoms are considered, Total BF and Ch4 volumes are negatively related to the co-occurrence of MCI and VH relative to neither symptom. Our data support the "cholinergic phenotype" in that BF atrophy, including Ch4, is associated with concomitant MCI and VH. Ch1-2 volume is further associated with MCI, which is interesting given its temporal lobe projections.

Surgery plays a key role in treating neuroblastoma. To assist surgical planning, anatomical 3D models derived from the segmentation of anatomical structures on MRI scans are often used. Automation using deep learning can make segmentations less time-consuming and more reliable. We organized the Surgical Planning in PedIatric Neuroblastoma (SPPIN) challenge, to stimulate developments and benchmarking of automatic segmentation of neuroblastoma on MRI. SPPIN is the first segmentation challenge in extracranial pediatric oncology. Nine teams provided a valid submission. Evaluation was based on the Dice similarity coefficient (Dice score), the 95th percentile of the Hausdorff distance (HD95), and the volumetric similarity (VS). A combination of these scores determined the ranking of the teams. The spread in the median evaluation scores per team was large (Dice: 0.21–0.82; HD95: 63.31–7.69; VS: 0.31–0.91). The top-performing team achieved a median Dice score of 0.82 (with an HD95 of 7.69 mm and a VS of 0.91) using a large, pre-trained model. However, in the pre-operative segmentations, significantly lower evaluation scores were observed. Our results indicate that pre-training might be useful in small, pediatric datasets. Although the general results of the winning team were high, they were insufficient to use for surgical planning in small, pre-operative tumors.

Systematic and random errors in radiation dose delivery necessitate the use of planning target volume (PTV) margins to ensure adequate clinical target volume (CTV) treatment. Advances in magnetic resonance-guided radiation therapy (MRgRT) have enabled improved imaging with possible margin reduction; however, the optimal PTV margins remain uncertain. This study aimed to evaluate the adaptive radiotherapy component of intra-fractional prostate movement in MRgRT for prostate cancer (PCa) patients and determine appropriate PTV margins. This study retrospectively analyzed 18 PCa patients treated using a 1.5 T MR-Linac. The initial fusion MR and verification MR scans were registered offline to assess prostate displacement between the two scans in the anterior-posterior (AP), left-right (LR) and superior-inferior (SI) directions. Random and systematic errors were calculated, and the PTV margins were determined using the Van Herk formula. The average time between MR scans was 22 min (range 9-54 min) compared to an average beam-on time of 6 min (range 2-11 min). Mean and standard deviation of translational displacement was 1.2 ± 0.9 mm in the AP, 0.6 ± 0.5 mm in the LR, and 1.1 ± 0.8 mm in the SI directions. The calculated PTV margin was 3.2 mm in AP, 1.7 mm in LR, and 3.2 mm in SI directions. There was an observed trend of increased prostate motion with increased treatment duration. MRgRT facilitates PTV margin reduction for PCa; however, our findings suggest that increased on-couch time may be associated with greater prostate motion. Future studies with larger patient cohorts and real-time motion monitoring are recommended to optimise margin strategies.

Aims: The relationship between bicipital groove morphology and biceps tendon pathology remains controversial, and few studies have evaluated volumetric parameters. To investigate whether patients with biceps tendinitis have distinct morphometric and volumetric features of the bicipital groove compared with asymptomatic individuals. Methods: This retrospective study included 36 patients who underwent surgical treatment for biceps tendon pathology (tenodesis or tenotomy) and 66 asymptomatic controls with available shoulder MRI and CT scans. Exclusion criteria were prior shoulder surgery or fracture, labral lesions, adhesive capsulitis, inflammatory arthritis, advanced osteoarthritis, inadequate imaging quality, and prior corticosteroid injection. Radiological parameters included total opening angle (TOA), medial wall angle (MWA), groove width, groove depth, and groove volume measured on CT and MRI. Volumetric analysis was performed using OsiriX software. Inter- and intra-observer reliability was tested with intraclass correlation coefficients (ICCs). Results: Compared with controls, patients with biceps tendon pathology had significantly narrower groove width (10.26 vs 11.62 mm, p=0.010), greater depth (6.50 vs 5.52 mm, p=0.001), smaller TOA (75.73° vs 90.98°, p=0.001), larger MWA (51.23° vs 44.42°, p=0.001), and lower groove volumes on both CT (0.704 vs 0.918 mm³, p=0.001) and MRI (0.623 vs 0.794 mm³, p=0.001). No significant differences were found between tenodesis and tenotomy subgroups, or between male and female patients. ICC values indicated excellent inter- and intra-observer reliability (0.84–0.91). Conclusion: Patients with biceps tendinitis exhibit a narrower, deeper, and smaller bicipital groove compared to asymptomatic individuals, suggesting that groove morphology may represent a potential risk factor for tendon pathology. Volumetric analysis provides a novel and reliable tool for characterizing this anatomy. However, due to the limited sample size, group heterogeneity, and retrospective design, these results should be interpreted as preliminary and hypothesis-generating.

ABSTRACT An 81‐year‐old man with a 30‐year history of taking sodium valproate (800 mg/day) for epilepsy experienced an acute onset of involuntary hyperkinetic movements. This progressive condition kept him bedridden for 3 months before he visited our hospital. Upon admission, he exhibited generalized choreiform and athetotic movements and senile hearing loss. Both brain MRI and CT scans revealed no acute or chronic specific lesions. Notably, his valproate serum concentration was 8.0 mg/L, markedly below the therapeutic range of 50–100 mg/L. Since the COVID‐19 pandemic began, he has limitedly visited his primary care physician for medication refills. No epileptiform activity was identified on the electroencephalography. Given the potential link between his choreo‐athetosis and withdrawal of chronic valproate treatment, which can act as a D 2 antagonist, he was started on 50 mg of tiapride hydrochloride (a D 2 antagonist) taken thrice daily, which significantly improved his choreo‐athetosis within a week.

Central nervous system (CNS) Lyme neuroborreliosis (LNB) rarely occurs. Clinical and radiological findings are heterogenous and considered unspecific, often delaying the time to diagnosis and treatment. Recently, specific clinical courses and MRI patterns have been proposed for patients presenting with brainstem encephalitis, but their relevance remains unclear. To describe radio-clinical features of patients with CNS-LNB encephalomyelitis, and to further investigate potential specific imaging patterns that could facilitate the diagnosis. We report a case of late LNB encephalomyelitis with MRI findings compatible with a recently suggested distinct pattern. We provide a systematic review conducted using Embase and PubMed databases. The clinical, radiological and CSF features, as well as treatment outcome are reported. Thirty-six cases of confirmed LNB encephalomyelitis were analyzed. The median time to diagnosis was 3months. One third of patients had a history of tick bite. Common symptoms and signs included gait disorder (74%), headache (71%), fatigue (48%), muscle wasting (48%), and brainstem signs (45%). Common MRI findings were symmetric pontomesencephalic lesions (56%) with patterns compatible with Tarsier-sign (19%) and/or M-sign (33%), and leptomeningeal basal and/or perimedullary enhancement (31%). Twenty-eight percent of the patients recovered completely after treatment, while the remaining had mild neurological deficits (mRS 1-2). Leptomeningeal basal and perimedullary enhancement, along with pontomesencephalic symmetric lesions-sometimes compatible with Tarsier-sign and M-sign-could serve as potential MRI markers suggestive of CNS late LNB encephalomyelitis, aiding in earlier diagnosis and treatment.

Objectives: This exploratory preclinical study aimed to compare the correlations of apparent diffusion coefficient (ADC) and native T1 mapping histogram features with tumor cell proliferation, microvessel density (MVD), and extracellular matrix composition in neuroblastoma xenografts. Methods: Neuroblastoma xenografts (n = 42) were established by subcutaneously injecting three MYCN-amplified/non-amplified human neuroblastoma cell lines (IMR-32, SK-N-BE(2), and SH-SY5Y; n = 14 per group) into female immunodeficient BALB/c-nude mice. Once tumors reached a diameter within the range of 12–15 mm, native T1 mapping and diffusion-weighted imaging were performed using a 3.0T clinical MRI scanner. Tumor cell proliferation and MVD were assessed via immunohistochemical Ki-67 staining and CD31 staining, respectively. Collagen fibers were visualized using Masson staining to calculate the collagen volume fraction (CVF). Pearson correlation coefficients with false discovery rate (FDR) correction were used to evaluate their associations. Results: Significant negative correlations were observed between Ki-67 expression and multiple ADC values after FDR correction, including ADC10Percentile (r = −0.397, adjusted p = 0.032), ADC90Percentile (r = −0.394, adjusted p = 0.032), ADCmaximum (r = −0.362, adjusted p = 0.048), ADCmean (r = −0.421, adjusted p = 0.032), ADCmedian (r = −0.422, adjusted p = 0.032), ADCminimum (r = −0.390, adjusted p = 0.032), and ADCrootmeansquared (r = −0.419, adjusted p = 0.032). In contrast, multiple T1 mapping features showed significant positive correlations with CVF (adjusted p < 0.05). Conclusions: ADC and T1 mapping provide complementary insights into tumor proliferation and extracellular matrix composition in neuroblastoma. These preclinical findings support further research to validate their potential clinical utility.

This study aimed to (1) demonstrate the intra- and interrater reliability of quadriceps (QUADS) and hamstring (HAMS) mineral-free lean (MFL) mass measures using DXA scanning, (2) determine the association of total thigh MFL mass measures with MFL mass measures of the hamstrings and quadriceps combined and (3) analyze the association between total thigh MFL mass and total body MFL mass measures. A total of 80 young children (aged 5 to 11 yrs) participated and unique regions of interest were created using custom analysis software with manual tracing of the QUADS, HAMS, and total thigh MFL mass measures. Repeated-measure analysis of variance was used to determine if there were significant differences among the MFL measures while intraclass correlation coefficients (ICC), coefficients of variation (CV), and regression analysis were used to determine the intra- and interrater reliability and the explained variance in the association among MFL mass measures. The right interrater QUADS MFL mass was the only significant group mean difference, and ICCs between (≥0.961) and within (≥0.919) raters were high for all MFL measures with low variation across all MFL measures (≤6.13%). The explained variance was 92.5% and 96.3% for the between-investigator analyses of the right and left total thigh MFL mass measures, respectively. Furthermore, 97.5% of the variance in total body MFL mass was explained by the total thigh MFL mass. DXA MFL mass measures of the QUADS, HAMS and total thigh can be confidently used in young children and may provide an alternative to CT or MRI scanning when assessing changes in MFL cross-sectional area or volume measures due to disease progression, training and rehabilitative strategies.

High-resolution imaging is critical for the diagnosis, treatment planning, and postoperative monitoring of cerebral aneurysms, which affect up to 5% of the population and pose a significant risk of rupture and subarachnoid hemorrhage. Surgical clipping remains a definitive treatment option, but metallic clips can introduce substantial imaging artifacts, complicating posttreatment assessment. This review synthesizes current knowledge on the impact of aneurysm clip materials and designs on artifact generation and explores strategies for artifact mitigation. Conventional materials like titanium are favored for their biocompatibility and reduced ferromagnetism but still cause beam hardening, streak artifacts, and signal loss in CT and MRI scans. Emerging alternatives, including ceramics, composites, polymers, and bioresorbable clips, show promise in reducing artifacts while maintaining mechanical reliability. Innovations in clip design, such as fenestrated or low-profile models, further aid in minimizing imaging distortion. Advanced imaging methods, including dual-energy CT, iterative reconstruction algorithms, and metal artifact reduction software, demonstrate significant improvements in image quality but may introduce limitations such as increased processing demands or subtle anatomical distortions. Future directions emphasize the development of next-generation clip materials, robotic-assisted surgical approaches, and artificial intelligence-driven reconstruction techniques to further optimize visualization and patient safety. Continued research and multidisciplinary collaboration will be essential to translate these innovations into routine neurosurgical practice.

Abstract Background Ejection fraction (EF) and end-systolic volume (ESV) are prognostic markers in cardiovascular disease. While MRI provides accurate assessments, its cost limits widespread use. Non-contrast cardiac CT (NCCT), used for coronary artery disease screening, may offer additional functional information. Objective To evaluate the accuracy of AI-derived ventricular volumes and EF from NCCT compared to contrast cardiac CT (CCT) and MRI. Methods This single center study included 205 patients who underwent cardiac CT for valve planning, divided into retrospective and prospective cohorts. A validated AI algorithm was applied to low-dose NCCT images at end-diastole and end-systole. Right (RV) and left ventricles (LV) volumes and their EFs were compared with CCT and MRI. Results In the prospective cohort (49 women, 53 men; mean age 73.9 ± 10.3 years), NCCT correlated strongly with CCT for LVEDV (152 mL; -14.2% relative difference; r = 0.91) and LVESV (96 mL; +32.6%; r = 0.84), with similar correlations for RVEDV (163 mL; -8.4%; r = 0.82) and RVESV (121.4 mL; +33.1%; r = 0.85). NCCT predicted LVEF <40% with 98% negative predictive value and 87% accuracy. LVEDV correlated strongly with MRI (n=16) for CCT (240 mL; +4.2%; r = 0.99) and NCCT (197 mL; -14.3%; r = 0.97), as did LVESV for CCT (115 mL; -5%; r = 0.99) and NCCT (134 mL; +11%; r = 0.97). Conclusion AI-derived ventricular volumes from NCCT show moderate to strong correlations, but EF is underestimated. The derived EF can be a screening tool to rule out significant ventricular dysfunction.

Aims: This study aimed to evaluate the impact of atrial fibrillation (AF) on cognitive function and neuroradiological markers in patients with Alzheimer’s disease (AD). Methods: A total of 800 patients were screened at the Dementia Outpatient Clinic of Sancaktepe Şehit Prof. Dr. İlhan Varank Training and Research Hospital. Thirty-three AD patients with confirmed AF were included, alongside 31 age-, sex, and education-matched AD patients without AF as controls. Cognitive function was assessed using the Mini-Mental State Examination (MMSE), while mood symptoms were evaluated with the Geriatric Anxiety Scale and the Geriatric Depression Scale. Brain MRI scans were analyzed using Fazekas, Medial Temporal Atrophy (MTA), and Koedam scores. Due to the modest sample size, this study is best interpreted as exploratory in nature. Results: After six months, the AF-positive group exhibited a significant decline in MMSE scores, whereas the AF-negative group showed no significant change. Koedam scores were significantly higher in the AF-positive group, indicating more advanced cortical atrophy. Regression analysis revealed that AF had a borderline significant negative impact on cognitive performance after adjusting for baseline MMSE scores. Conclusion: AF may accelerate cognitive decline and is associated with greater cortical atrophy in patients with AD. These findings suggest that close cognitive monitoring and individualized care strategies may be beneficial in this population, though further research is needed to confirm this.

To assess the validity of information on the date of diagnosis of intracranial tumors in the Swedish National Cancer Register. The study was conducted in Region Stockholm, covering approximately 2.4 million inhabitants. Data from the Image and Function Service (BFT) archive, which contains a population-based database of all diagnostic radiology in the Stockholm Region, was used. The study included all primary cases of intracranial tumors (ICD10 codes C70 and C71) reported from Stockholm to the National Cancer Register from 2010 to 2020. Radiology reports from CT and MRI exams performed from 2000 to 2020 were reviewed. Reports from exams conducted more than one month before the Swedish Cancer Register diagnosis were manually examined to validate the date of diagnosis. For 98.8% of the brain tumor patients and 96.2% of the meningioma patients, the date of first radiological diagnosis was one month or less prior to date of diagnosis in the Swedish Cancer Register. Of the patients with radiological diagnoses more than one month prior to cancer register diagnosis, 30 patients (1.0%) had tumors reported to the register one month to one year after radiological diagnosis, and 36 patients (1.2%) had tumors reported more than one year later. The study found that for patients from the Stockholm Region with intracranial tumors reported to the Swedish Cancer Register during the period 2010 through 2020, date of diagnosis is highly accurate.

Diffuse intrinsic pontine glioma is a lethal disease with a median overall survival (OS) of less than one year. Currently, no curative treatment exists, and radiotherapy primarily aims for symptom palliation. Hypofractionated radiotherapy (HR) has been proposed to enhance quality of life through shorter treatment duration, fewer hospital visits, and reduced costs, particularly in low- and middle-income countries (LMICs). However, many LMIC institutions continue to use 54Gy in 30 fractions. The study aims to identify patients who may benefit from HR. We reviewed data from MAHAK patients treated between April 2010 and February 2020. Diagnosis was based on clinical symptoms and MRI imaging; biopsy was not mandatory. Among 108 identified patients, 34(32%) underwent a biopsy, with high-grade glioma or glioblastoma diagnosed in 18(53%). H3K27M mutation was found in two patients, while others were not evaluated. No seeding was detected at diagnosis. Thirty-one patients (29%) received 50-54Gy using normal fractionation, while the remainder received HR. Median OS was 9.6 months (range 1-67). No significant impact of fractionation was observed. Seventy-eight (72%) patients received chemotherapy, which showed no benefit. Eleven patients received re-irradiation at progression, showing a post-progression OS of 6 months. Poor-risk factors included age over 4 years, symptom duration < 3months, simultaneous triad signs, and absence of hydrocephalus (p < 0.05). Given their shorter life expectancy, poor-risk patients may benefit from HR. We propose a scoring system based on prognostic factors to guide decisions between HR and normal fractionation.

Preparation and characterization of chitosan/citric acid composite film and its application in blueberry preservation.

Deficits in cortical thickness have been a replicated finding in BD. Greater brain serotonin 1 A receptor (5HT1AR) binding has also been reported in BD, consistent with a disruption in serotonin signaling. Serotonin can cause a trophic effect on the cerebral cortex. 31 depressed participants with BD and 49 healthy volunteers (HVs) were included. PET imaging with [11C]WAY100635 was obtained with a metabolite-corrected arterial input function. Radiotracer binding potential (BPF) values were quantified within cortical regions. T1-weighted MRI scans were processed through Freesurfer 7.1.1, and cortical thickness values were quantified within a right caudal middle frontal region that had previously been shown to have deficits in BD. The previous results were replicated here, with lower cortical thickness in BD relative to HV’s in the region. Cortical thickness in that region was inversely associated with BPF of [11C]WAY100635 in BD (F = 4.77; df = 1, 28; p = 0.037), but no association was found in HVs (F = 0.62; df = 1, 46; p = 0.44). Whole brain exploratory analyses were performed to examine vertex-wise associations between BPF values and cortical thickness. Within BD participants, BPF of [11C]WAY100635 was inversely associated with right superior parietal cortical thickness, and positively associated in the right posterior cingulate cortex and the left entorhinal cortex. Mean cortical thickness values in the significant clusters did not differ between BD and HV groups. Within HVs, no clusters were associated between cortical thickness and BPF of [11C]WAY100635. These data are consistent with a model wherein disruptions in serotonin signaling lead to a loss of trophic effect on cortical tissue in BD. Associations between 5HT1AR binding and cortical thickness in BD are not limited to areas of cortical thickness deficits.

Accurate segmentation of brain tumors from multi-modal MRI scans is critical for diagnosis, treatment planning, and disease monitoring. Tumor heterogeneity and inter-image variability across MRI sequences pose challenging problems to state-of-the-art segmentation models. This paper presents a novel Multi-Modal Multi-Scale Contextual Aggregation with Attention Fusion (MM-MSCA-AF) framework that leverages multi-modal MRI images (T1, T2, FLAIR, and T1-CE) to enhance segmentation performance. The model employs multi-scale contextual aggregation to obtain global and fine-grained spatial features, and gated attention fusion for selectively refining effective feature representations and discarding noise. Evaluated on the BRATS 2020 dataset, MM-MSCA-AF achieves a Dice value of 0.8158 for necrotic tumor regions and 0.8589 in total, outperforming state-of-the-art architectures such as U-Net, nnU-Net, and Attention U-Net. These results demonstrate the effectiveness of MM-MSCA-AF in handling complex tumor shapes and improving segmentation accuracy. The proposed approach has significant clinical value, offering a more accurate and automatic brain tumor segmentation solution in medical imaging.

Neonates with complex congenital heart disease (CHD), in particular newborns with hypoplastic left heart syndrome (HLHS) or transposition of the great arteries (TGA) may show brain pathologies and altered brain growth after birth. Our prospective study investigates brain volumes and immature brain structures in these patient groups compared to healthy controls. Neonatal cerebral MRI scans were analysed by semi-automated segmentation (dHCP pipeline) in 51 children: 31 HLHS/HLHC/UVH (61%) at Stage I, in 18 neonates with TGA (35%) and in two with aortic arch obstruction and biventricular physiology (4%) at a mean GA 41.2 weeks, and in 209 controls at a mean GA 41.6 at time of the MRI. Newborns born premature were excluded. Brain volume comparisons used mixed models for imaging techniques and linear regression for CHD-control differences. Cerebral MRI was abnormal in 29 patients (57%), with multiple lesions in some patients: including liquor space enlargements (20%), small grey (20%) and white matter injuries (12%), stroke (8%), subdural haemorrhage (22%) and sinus venous thrombosis (8%). 69% of CHD neonates showed signs of brain immaturity in relation to GA. Intracranial volumes were reduced, while cerebrospinal fluid (CSF) volumes were enlarged compared to controls. Neonates with complex CHD show reduced cerebral growth, higher risk for brain injuries and impaired brain maturation, even before first surgery. This might constitute a higher perioperative risk in these patient groups than for normal developed brains. Identification of distinct patterns of brain volume loss might enable risk stratification for subsequent neuro-developmental impairment.