Joint self-supervised and supervised contrastive learning for multimodal MRI data: Towards predicting abnormal neurodevelopment.

The integration of different imaging modalities, such as structural, diffusion tensor, and functional magnetic resonance imaging, with deep learning models has yielded promising outcomes in discerning phenotypic characteristics and enhancing disease diagnosis. The development of such a technique hinges on the efficient fusion of heterogeneous multimodal features, which initially reside within distinct representation spaces. Naively fusing the multimodal features does not adequately capture the complementary information and could even produce redundancy. In this work, we present a novel joint self-supervised and supervised contrastive learning method to learn the robust latent feature representation from multimodal MRI data, allowing the projection of heterogeneous features into a shared common space, and thereby amalgamating both complementary and analogous information across various modalities and among similar subjects. We performed a comparative analysis between our proposed method and alternative deep multimodal learning approaches. Through extensive experiments on two independent datasets, the results demonstrated that our method is significantly superior to several other deep multimodal learning methods in predicting abnormal neurodevelopment. Our method has the capability to facilitate computer-aided diagnosis within clinical practice, harnessing the power of multimodal data. The source code of the proposed model is publicly accessible on GitHub: https://github.com/leonzyzy/Contrastive-Network.

Cerebral injury and retinopathy as risk factors for blindness in extremely preterm infants.

OBJECTIVE: This study investigates whether and to what extent cerebral injury is associated with bilateral blindness in extremely preterm infants, which has been attributed mainly to retinopathy of prematurity (ROP). DESIGN: Multicentre analysis of children born from 1994 to 2021 at gestational age 22 0/7 to 28 6/7 weeks with follow-up at 18-26 months. Logistic regression examined the adjusted association of bilateral blindness with severe ROP and/or cerebral injury among extremely preterm infants. EXPOSURES: Severe ROP and cerebral injury, the latter defined as any of the following on cranial imaging: ventriculomegaly; blood/increased echogenicity in the parenchyma; cystic periventricular leukomalacia. MAIN OUTCOME MEASURES: Bilateral blindness, defined as a follow-up examination meeting criteria of 'blind-some functional vision' or 'blind-no useful vision' in both eyes. RESULTS: The 19 863 children included had a mean gestational age of 25.6±1.7 weeks, mean birth weight of 782±158 g and 213 (1%) had bilateral blindness. Multiplicative interaction between ROP and cerebral injury was statistically significant. For infants with only severe ROP (n=3130), odds of blindness were 8.14 times higher (95% CI 4.52 to 14.65), and for those with only cerebral injury (n=2836), odds were 8.38 times higher (95% CI 5.28 to 13.28), compared with the reference group without either condition. Risks were not synergistic for infants with both severe ROP and cerebral injury (n=1438, adjusted OR=28.7, 95% CI 16.0 to 51.7, p<0.0001). CONCLUSIONS: In a group of extremely preterm infants, severe ROP and cerebral injury were equally important risk factors for blindness. Besides ROP, clinicians should consider cerebral injury as a cause of blindness in children born extremely preterm. TRIAL REGISTRATION NUMBER: NCT00063063.

Effects of Early Enteral to Parenteral Protein Ratios on Brain Volume and Somatic Growth in Very Low Birth Weight Infants.

OBJECTIVE: To evaluate whether a higher proportion of enteral vs parenteral protein ratio (E:P ratio) in the first 28 days after birth is associated with increased brain volume and somatic growth in very low birth weight (VLBW; birth weight <1500 g) infants. STUDY DESIGN: This was a retrospective analysis of a subcohort of VLBW infants (n = 256, gestational age mean 28.07 [SD 2.17] weeks, birth weight 1038.80 [SD 262.95] grams) from the Cincinnati Infant Neurodevelopment Early Prediction Study, a regional prospective study of infants born at ≤32 weeks' gestation. Brain magnetic resonance imaging was obtained at term-equivalent age. Macronutrient intake and growth metrics for the first 28 days were collected retrospectively. The primary outcome was total brain tissue volume. The relationships between E:P ratio, total and regional brain tissue volumes, and somatic growth were analyzed by multivariable linear regression models; composite variables were used to adjust for potential confounders including pregnancy risk factors and initial severity of illness. RESULTS: Higher E:P ratio was associated with increased total brain tissue volume but was not associated with change in head circumference z score. In secondary analyses, higher E:P ratio was associated with increased weight velocity. There were no significant associations between E:P ratio and change in weight or length z scores or regional brain volumes. CONCLUSIONS: Higher E:P ratio in the first 28 days was positively associated with total brain volume and weight gain. Promoting the provision of enteral over parenteral protein may improve brain and somatic growth in VLBW infants.

Corpus Callosum Abnormalities at Term-Equivalent Age Are Associated with Language Development at 2 Years' Corrected Age in Infants Born Very Preterm.

We studied the effect of microstructural abnormalities in the corpus callosum on language development in 348 infants born very prematurely. We discovered that the fractional anisotropy of the corpus callosum anterior midbody was a significant predictor of standardized language scores at 2 years, independent of clinical and social risk factors.

Hypertensive Disorders of Pregnancy and Risk of Early Brain Abnormalities on Magnetic Resonance Imaging at Term among Infants Born at ≤32 Weeks' Gestational Age.

OBJECTIVE: To evaluate the proximal effects of hypertensive disorders of pregnancy (HDP) on a validated measure of brain abnormalities in infants born at ≤32 weeks' gestational age (GA) using magnetic resonance imaging at term-equivalent age. STUDY DESIGN: In a multisite prospective cohort study, 395 infants born at ≤32 weeks' GA, underwent 3T magnetic resonance imaging scan between 39 and 44 weeks' postmenstrual age. A single neuroradiologist, blinded to clinical history, evaluated the standardized Kidokoro global brain abnormality score as the primary outcome. We classified infants as HDP-exposed by maternal diagnosis of chronic hypertension, gestational hypertension, pre-eclampsia, or eclampsia. Linear regression analysis identified the independent effects of HDP on infant brain abnormalities, adjusting for histologic chorioamnionitis, maternal smoking, antenatal steroids, magnesium sulfate, and infant sex. Mediation analyses quantified the indirect effect of HDP mediated via impaired intrauterine growth and prematurity and remaining direct effects on brain abnormalities. RESULTS: A total of 170/395 infants (43%) were HDP-exposed. Adjusted multivariable analyses revealed HDP-exposed infants had 27% (95% CI 5%-53%) higher brain abnormality scores than those without HDP exposure (P = .02), primarily driven by increased white matter injury/abnormality scores (P = .01). Mediation analyses showed HDP-induced impaired intrauterine growth significantly (P = .02) contributed to brain abnormality scores (22% of the total effect). CONCLUSIONS: Maternal hypertension independently increased the risk for early brain injury and/or maturational delays in infants born at ≤32 weeks' GA with an indirect effect of 22% resulting from impaired intrauterine growth. Enhanced prevention/treatment of maternal hypertension may mitigate the risk of infant brain abnormalities and potential neurodevelopmental impairments.

Association of enlarged extra-axial spaces and subdural hemorrhage in preterm infants at term-equivalent age.

To determine the incidence of enlarged extra-axial space (EES) and its association with subdural hemorrhage (SDH) in a regional cohort of preterm infants. As part of a prospective cohort study of 395 preterm infants, brain magnetic resonance imaging (MRI) was collected on each infant at term-equivalent age. Six preterm infants showed evidence of SDH. We reviewed the MRIs to identify the incidence of EES in these 6 infants and the cohort broadly. We then completed a retrospective chart review of the 6 infants to identify any concerns for non-accidental trauma (NAT) since the MRI was obtained. The incidence of SDH in the cohort was 1.6%. The incidence of EES was 48.1% including all 6 infants with SDH. The incidence of SDH in infants with EES was 3.2%. The retrospective chart review of the 6 infants did not yield any evidence of NAT. The incidence of EES and SDH in our cohort was significantly higher than similar cohorts of term infants, demonstrating an increased risk in preterm infants. The incidence of SDH in infants with EES was greater than in the total cohort, suggesting that it is a risk factor for asymptomatic SDH in preterm infants.

A systematic review of automated methods to perform white matter tract segmentation.

White matter tract segmentation is a pivotal research area that leverages diffusion-weighted magnetic resonance imaging (dMRI) for the identification and mapping of individual white matter tracts and their trajectories. This study aims to provide a comprehensive systematic literature review on automated methods for white matter tract segmentation in brain dMRI scans. Articles on PubMed, ScienceDirect [NeuroImage, NeuroImage (Clinical), Medical Image Analysis], Scopus and IEEEXplore databases and Conference proceedings of Medical Imaging Computing and Computer Assisted Intervention Society (MICCAI) and International Symposium on Biomedical Imaging (ISBI), were searched in the range from January 2013 until September 2023. This systematic search and review identified 619 articles. Adhering to the specified search criteria using the query, "white matter tract segmentation OR fiber tract identification OR fiber bundle segmentation OR tractography dissection OR white matter parcellation OR tract segmentation," 59 published studies were selected. Among these, 27% employed direct voxel-based methods, 25% applied streamline-based clustering methods, 20% used streamline-based classification methods, 14% implemented atlas-based methods, and 14% utilized hybrid approaches. The paper delves into the research gaps and challenges associated with each of these categories. Additionally, this review paper illuminates the most frequently utilized public datasets for tract segmentation along with their specific characteristics. Furthermore, it presents evaluation strategies and their key attributes. The review concludes with a detailed discussion of the challenges and future directions in this field.

Structural connectivity at term equivalent age and language in preterm children at 2 years corrected.

We previously reported interhemispheric structural hyperconnectivity bypassing the corpus callosum in children born extremely preterm (<28 weeks) versus term children. This increased connectivity was positively associated with language performance at 4-6 years of age in our prior work. In the present study, we aim to investigate whether this extracallosal connectivity develops in extremely preterm infants at term equivalent age by leveraging a prospective cohort study of 350 very and extremely preterm infants followed longitudinally in the Cincinnati Infant Neurodevelopment Early Prediction Study. For this secondary analysis, we included only children born extremely preterm and without significant brain injury (n = 95). We use higher-order diffusion modelling to assess the degree to which extracallosal pathways are present in extremely preterm infants and predictive of later language scores at 22-26 months corrected age. We compare results obtained from two higher-order diffusion models: generalized q-sampling imaging and constrained spherical deconvolution. Advanced MRI was obtained at term equivalent age (39-44 weeks post-menstrual age). For structural connectometry analysis, we assessed the level of correlation between white matter connectivity at the whole-brain level at term equivalent age and language scores at 2 years corrected age, controlling for post-menstrual age, sex, brain abnormality score and social risk. For our constrained spherical deconvolution analyses, we performed connectivity-based fixel enhancement, using probabilistic tractography to inform statistical testing of the hypothesis that fibre metrics at term equivalent age relate to language scores at 2 years corrected age after adjusting for covariates. Ninety-five infants were extremely preterm with no significant brain injury. Of these, 53 had complete neurodevelopmental and imaging data sets that passed quality control. In the connectometry analyses adjusted for covariates and multiple comparisons (P < 0.05), the following tracks were inversely correlated with language: bilateral cerebellar white matter and middle cerebellar peduncles, bilateral corticospinal tracks, posterior commissure and the posterior inferior fronto-occipital fasciculus. No tracks from the constrained spherical deconvolution/connectivity-based fixel enhancement analyses remained significant after correction for multiple comparisons. Our findings provide critical information about the ontogeny of structural brain networks supporting language in extremely preterm children. Greater connectivity in more posterior tracks that include the cerebellum and connections to the regions of the temporal lobes at term equivalent age appears to be disadvantageous for language development.

Supervised contrastive learning enhances graph convolutional networks for predicting neurodevelopmental deficits in very preterm infants using brain structural connectome.

Very preterm (VPT) infants (born at less than 32 weeks gestational age) are at high risk for various adverse neurodevelopmental deficits. Unfortunately, most of these deficits cannot be accurately diagnosed until the age of 2-5 years old. Given the benefits of early interventions, accurate diagnosis and prediction soon after birth are urgently needed for VPT infants. Previous studies have applied deep learning models to learn the brain structural connectome (SC) to predict neurodevelopmental deficits in the preterm population. However, none of these models are specifically designed for graph-structured data, and thus may potentially miss certain topological information conveyed in the brain SC. In this study, we aim to develop deep learning models to learn the SC acquired at term-equivalent age for early prediction of neurodevelopmental deficits at 2 years corrected age in VPT infants. We directly treated the brain SC as a graph, and applied graph convolutional network (GCN) models to capture complex topological information of the SC. In addition, we applied the supervised contrastive learning (SCL) technique to mitigate the effects of the data scarcity problem, and enable robust training of GCN models. We hypothesize that SCL will enhance GCN models for early prediction of neurodevelopmental deficits in VPT infants using the SC. We used a regional prospective cohort of ∼280 VPT infants who underwent MRI examinations at term-equivalent age from the Cincinnati Infant Neurodevelopment Early Prediction Study (CINEPS). These VPT infants completed neurodevelopmental assessment at 2 years corrected age to evaluate cognition, language, and motor skills. Using the SCL technique, the GCN model achieved mean areas under the receiver operating characteristic curve (AUCs) in the range of 0.72∼0.75 for predicting three neurodevelopmental deficits, outperforming several competing models. Our results support our hypothesis that the SCL technique is able to enhance the GCN model in our prediction tasks.

Development and Evaluation of I-PASS-to-PICU: A Standard Electronic Template to Improve Referral Communication for Interfacility Transfers to the Pediatric ICU.

BACKGROUND: Miscommunication during interfacility handoffs to a higher level of care can harm critically ill children. Adapting evidence-based handoff interventions to interfacility referral communication may prevent adverse events. The objective of this project was to develop and evaluate a standard electronic referral template (I-PASS-to-PICU) to improve communication for interfacility pediatric ICU (PICU) transfers. METHODS: I-PASS-to-PICU was iteratively developed in a single PICU. A core PICU stakeholder group collaboratively designed an electronic health record (EHR)-supported clinical note template by adapting elements from I-PASS, an evidence-based handoff program, to support information exchange between referring clinicians and receiving PICU physicians. I-PASS-to-PICU is a receiver-driven tool used by PICU physicians to guide verbal communication and electronic documentation during PICU transfer calls. The template underwent three cycles of iterative evaluation and redesign informed by individual and group interviews of multidisciplinary PICU staff, usability testing using simulated and actual referral calls, and debriefing with PICU physicians. RESULTS: Individual and group interviews with 21 PICU staff members revealed that relevant, accurate, and concise information was needed for adequate admission preparedness. Time constraints and secondhand information transmission were identified as barriers. Usability testing with six receiving PICU physicians using simulated and actual calls revealed good usability on the validated System Usability Scale (SUS), with a mean score of 77.5 (standard deviation 10.9). Fellows indicated that most fields were relevant and that the template was feasible to use. CONCLUSION: I-PASS-to-PICU was technically feasible, usable, and relevant. The authors plan to further evaluate its effectiveness in improving information exchange during real-time PICU practice.

Prenatal tobacco smoke exposure and risk for cognitive delays in infants born very premature.

Prenatal tobacco smoke exposure (TSE) and prematurity are independent risk factors for abnormal neurodevelopment. The objectives were to compare differences in Bayley-III cognitive, language, and motor scores at 2 years corrected age (CA) in 395 infants born very preterm (≤ 32 weeks gestation) with and without prenatal TSE. We performed multivariable linear regression analyses to examine associations between prenatal TSE and neurodevelopmental outcomes and a mediation analysis to estimate direct effects of prenatal TSE on outcomes and indirect effects through preterm birth. In total, 50 (12.6%) infants had prenatal TSE. Infants with prenatal TSE had lower mean [95% CI] Cognitive score (82.8 [78.6, 87.1]) vs. nonexposed infants (91.7 [90.1, 93.4]). In children with and without prenatal TSE, there were significant differences in mean [95% CI] Language scores (81.7 [76.0, 87.4] vs. 92.4 [90.2, 94.6], respectively) and mean [95% CI] Motor scores (86.5 [82.2, 90.7] vs. 93.4 [91.8, 95.0], respectively); scores remained significant after controlling for confounders. Preterm birth indirectly mediated 9.0% of the total effect of prenatal TSE on Cognitive score (P = NS). However, 91% of the remaining total effect was significant and attributable to TSE's direct harmful effects on cognitive development (ß = - 5.17 [95% CI - 9.97, - 0.38]). The significant association is largely due to TSE's direct effect on cognitive development and not primarily due to TSE's indirect effect on preterm birth.

Prenatal Opioid Exposure and Risk for Adverse Brain and Motor Outcomes in Infants Born Premature.

OBJECTIVE: To compare brain magnetic resonance imaging (MRI) biomarkers and neurodevelopmental test scores in infants born preterm with and without prenatal opioid exposure (POE). STUDY DESIGN: We examined 395 preterm infants (≤32 weeks gestational age) who had term-equivalent brain MRIs, composite scores from the Bayley Scales of Infant and Toddler Development-III at 2 years corrected age, and POE data. MRI parameters included total/regional brain volumes and severe punctate white matter lesions (PWMLs). We conducted bivariable analysis and multivariable logistic regression analyses. RESULTS: The mean ± SD gestational age was 29.3 ± 2.5 weeks; 35 (8.9%) had POE and 20 (5.1%) had severe PWML. Compared with unexposed infants, those with POE exhibited higher rates of severe PWML (17.1% vs 3.9%, respectively; P = .002); findings remained significant with an OR of 4.16 (95% CI, 1.26-13.68) after adjusting for confounders. On mediation analysis, the significant relationship between POE and severe PWML was not indirectly mediated through preterm birth/gestational age (OR, 0.93; 95% CI, 0.78-1.10), thus suggesting the association was largely driven by a direct adverse effect of POE on white matter. In multivariable analyses, POE was associated with a significantly lower score by -6.2 (95% CI, -11.8 to -0.6) points on the Bayley Scales of Infant and Toddler Development-III Motor subscale compared with unexposed infants. CONCLUSIONS: POE was associated with severe PWML; this outcome may be a direct effect of POE rather than being mediated by premature birth. POE was also associated with worse motor development. Continued follow-up to understand the long-term effects of POE is warranted.

Mathematics abilities associated with adaptive functioning in preschool children born preterm.

It is well known that infants born very preterm (VPT) often demonstrate deficits in mathematical abilities in early childhood which are associated with poorer academic outcomes. Mathematic skills are also critical for other areas of functioning. However, it is not known whether mathematics skills are associated with adaptive functioning in children born preterm. Infants born at ≤31 weeks gestation and full term were recruited at birth and followed over time. At the 36-month corrected age assessment, children were administered the Early Number Concepts subtest of the Differential Abilities Scale, Second Edition, and caregivers completed the Adaptive Behavior Assessment System, Third Edition. After controlling for age, sex, cognitive abilities, and caregiver education, performance on the mathematics measure was uniquely and positively associated with adaptive behavior for preschool children in the VPT group only. Exploratory analyses revealed this association to be specifically related to the Practical and Social composites. Knowledge of concepts of number and quantity were associated with better adaptive functioning, particularly for behaviors related to functioning at home and in the community and play/social functioning, for children born VPT. Although replication is warranted, it would appear that mathematical skills may be an important early intervention target for children born VPT.

Feasibility and acceptability of an online parenting intervention to address behaviour problems in moderately to extremely preterm pre-school and school-age children.

BACKGROUND: Preterm birth is associated with adverse mental health outcomes, including internalizing problems, social difficulties and inattention. Interventions are needed beyond infancy and toddlerhood to support children and their families. We examined the feasibility and acceptability of the I-InTERACT Preterm pilot study, an online parenting intervention for preterm children ages 3-8. METHOD: Families participated in a weekly intervention comprised of seven sessions with online modules followed by videoconference coaching sessions with a therapist. Following completion of the study, caregivers completed a survey to assess their satisfaction and were asked to participate in a voluntary semi-structured interview to provide feedback. We anticipated greater than a 50% participation rate (enrollment feasibility) and 75% completion rate (adherence feasibility). We also hypothesized that at least 80% of participants would be satisfied with the intervention (acceptability). RESULTS: Nineteen of 32 families (59%) enrolled in the study, suggesting adequate enrollment feasibility. Feasibility of programme completion (adherence) was lower than anticipated (59%). Regarding satisfaction, all caregivers agreed that the programme's information was relevant to them and their family. Nearly all participants (92%) indicated that they had a better understanding of the effects of preterm birth on behaviour, that they enjoyed the programme, that it met their expectations and that they recommend the programme to others. In qualitative interviews, caregivers expressed satisfaction with the content, skills they learned, and receiving direct coaching. Caregivers suggested improvements to increase intervention feasibility and skill implementation, including offering biweekly sessions and more hands-on coaching. CONCLUSION: Our largely satisfactory acceptability rates suggest the value of and need for a parenting intervention for children born preterm past the initial period of early development. Future directions include modifying the intervention in response to caregiver feedback to improve recruitment, engagement and adherence.

Corpus callosum abnormalities at term-equivalent age are associated with language development at two years corrected age in infants born very preterm.

We studied the impact of microstructural abnormalities in the corpus callosum on language development in 348 infants born very prematurely. We discovered that the fractional anisotropy of the corpus callosum anterior midbody was a significant predictor of standardized language scores at two years, independent of clinical and social risk factors.

Language Outcomes of Children Born Very Preterm in Relation to Early Maternal Depression and Anxiety.

Unaddressed maternal psychological distress within the first year postpartum is known to have numerous negative consequences on the child's developmental outcomes, including language acquisition. This study examined the relationship between early maternal psychosocial factors and the language outcomes of children born very preterm (VPT; ≤32 weeks gestational age). It used data from the Cincinnati Infant Neurodevelopment Early Prediction Study, an ongoing National-Institutes-of-Health-funded prospective, multicenter cohort investigation of VPT infants. A total of 243 (125 boys; 118 girls) children born VPT (M = 29.03 weeks of gestation; SD = 2.47) and their corresponding 207 mothers (34 with multiple infants) were included in this study. We did not find an association between maternal depression or anxiety and Bayley-III (M = 92.3, SD = 18.9) language scores. Additionally, maternal grit and self-efficacy did not modify the relationship between depression and anxiety and language scores. A higher level of maternal education and infant female sex were significantly associated with higher language scores. While preterm birth typically results in higher rates of depression and anxiety for parents, the findings suggest that maternal depression, anxiety, and grit and the self-efficacy of the mothers in this sample did not relate to the language development of their children, independent of maternal education and infant female sex.

Discovery and Characterization of ZL-2201, a Potent, Highly Selective, and Orally Bioavailable Small-molecule DNA-PK Inhibitor.

DNA-dependent protein kinase (DNA-PK), a driver of the non-homologous end-joining (NHEJ) DNA damage response pathway, plays an instrumental role in repairing double-strand breaks (DSB) induced by DNA-damaging poisons. We evaluate ZL-2201, an orally bioavailable, highly potent, and selective pharmacologic inhibitor of DNA-PK activity, for the treatment of human cancerous malignancies. ZL-2201 demonstrated greater selectivity for DNA-PK and effectively inhibited DNA-PK autophosphorylation in a concentration- and time-dependent manner. Initial data suggested a potential correlation between ataxia-telangiectasia mutated (ATM) deficiency and ZL-2201 sensitivity. More so, ZL-2201 showed strong synergy with topoisomerase II inhibitors independent of ATM status in vitro. In vivo oral administration of ZL-2201 demonstrated dose-dependent antitumor activity in the NCI-H1703 xenograft model and significantly enhanced the activity of approved DNA-damaging agents in A549 and FaDu models. From a phosphoproteomic mass spectrometry screen, we identified and validated that ZL-2201 and PRKDC siRNA decreased Ser108 phosphorylation of MCM2, a key DNA replication factor. Collectively, we have characterized a potent and selective DNA-PK inhibitor with promising monotherapy and combinatory therapeutic potential with approved DNA-damaging agents. More importantly, we identified phospho-MCM2 (Ser108) as a potential proximal biomarker of DNA-PK inhibition that warrants further preclinical and clinical evaluation. Significance: ZL-2201, a potent and selective DNA-PK inhibitor, can target tumor models in combination with DNA DSB-inducing agents such as radiation or doxorubicin, with potential to improve recurrent therapies in the clinic.

Correlation of NICU anthropometry in extremely preterm infants with brain development and language scores at early school age.

Growth in preterm infants in the neonatal intensive care unit (NICU) is associated with increased global and regional brain volumes at term, and increased postnatal linear growth is associated with higher language scores at age 2. It is unknown whether these relationships persist to school age or if an association between growth and cortical metrics exists. Using regression analyses, we investigated relationships between the growth of 42 children born extremely preterm (< 28 weeks gestation) from their NICU hospitalization, standardized neurodevelopmental/language assessments at 2 and 4-6 years, and multiple neuroimaging biomarkers obtained from T1-weighted images at 4-6 years. We found length at birth and 36 weeks post-menstrual age had positive associations with language scores at 2 years in multivariable linear regression. No growth metric correlated with 4-6 year assessments. Weight and head circumference at 36 weeks post-menstrual age positively correlated with total brain volume and negatively with global cortical thickness at 4-6 years of age. Head circumference relationships remained significant after adjusting for age, sex, and socioeconomic status. Right temporal cortical thickness was related to receptive language at 4-6 years in the multivariable model. Results suggest growth in the NICU may have lasting effects on brain development in extremely preterm children.

A novel collaborative self-supervised learning method for radiomic data.

The computer-aided disease diagnosis from radiomic data is important in many medical applications. However, developing such a technique relies on labeling radiological images, which is a time-consuming, labor-intensive, and expensive process. In this work, we present the first novel collaborative self-supervised learning method to solve the challenge of insufficient labeled radiomic data, whose characteristics are different from text and image data. To achieve this, we present two collaborative pretext tasks that explore the latent pathological or biological relationships between regions of interest and the similarity and dissimilarity of information between subjects. Our method collaboratively learns the robust latent feature representations from radiomic data in a self-supervised manner to reduce human annotation efforts, which benefits the disease diagnosis. We compared our proposed method with other state-of-the-art self-supervised learning methods on a simulation study and two independent datasets. Extensive experimental results demonstrated that our method outperforms other self-supervised learning methods on both classification and regression tasks. With further refinement, our method will have the potential advantage in automatic disease diagnosis with large-scale unlabeled data available.