ObjectiveDiscriminating atypical hepatocellular carcinoma (HCC) from other malignancies in liver nodules classified as Liver Imaging Reporting and Data System category M (LR-M) remains a significant diagnostic challenge on conventional ultrasound examination. The LR-M category, originally intended to capture non-HCC malignancies, paradoxically contains up to 63% of atypical HCCs that deviate from classic enhancement patterns, leading to potential misdiagnosis and suboptimal treatment planning. While deep learning has shown promise in HCC diagnosis, most existing models rely exclusively on single-modality ultrasound, overlooking the diagnostic benefits of integrating complementary information from multiple imaging sources. To address this gap, we propose a novel attention-weighted tri-modal ultrasound network (TUS-Net) that integrates contrast-enhanced ultrasound (CEUS), B-mode ultrasound (BUS), and time-intensity curves (TICs) to improve diagnostic accuracy for these clinically challenging lesions. MethodsOur framework incorporates a three-dimensional convolutional neural network (C3D) backbone to extract spatiotemporal features from CEUS videos, capturing dynamic vascular patterns critical for lesion characterization. To effectively fuse complementary modalities, we introduce a dual-channel feature fusion module (DCFFM) that adaptively combines features from CEUS and BUS through channel-wise attention mechanisms, allowing the model to dynamically weigh the contribution of each modality based on diagnostic relevance. Additionally, we propose a temporal intensity feature fusion module (TIFFM) that leverages quantitative hemodynamic information from TICs to guide the model’s attention toward diagnostically critical temporal phases, such as arterial wash-in and portal venous washout. The model is further enhanced by automated lesion localization using YOLOX and class activation mapping for interpretability, ensuring that predictions align with clinically meaningful imaging features. ResultsEvaluated on a tri-modal ultrasound dataset comprising 161 patients with pathologically confirmed LR-M nodules (131 atypical HCC and 30 non-HCC malignancies), our model achieved an accuracy of 86.83%, a sensitivity of 92.50%, a specificity of 75.50%, and an AUC of 89.32% in screening atypical HCC. Compared to single-modality baselines, TUS-Net demonstrated superior specificity, a clinically critical metric given the higher risk associated with misclassifying non-HCC malignancies. Ablation studies confirmed the contribution of each module, with the full model outperforming both standard C3D and 3D ResNet backbones integrated with attention mechanisms. A reader study involving junior and senior radiologists further validated the clinical utility of AI assistance, showing consistent improvements in specificity and inter-reader consistency, particularly for less experienced clinicians. ConclusionThese results surpass existing benchmark models and demonstrate the potential of our approach to enhance diagnostic precision in clinically specific cases. By intelligently fusing multi-modal ultrasound data with attention-guided mechanisms, TUS-Net offers a reliable and interpretable tool that holds promise for improving the non-invasive diagnosis of atypical HCC in challenging LR-M liver nodules.

ObjectiveDiscriminating atypical hepatocellular carcinoma (HCC) from other malignancies in liver nodules classified as Liver Imaging Reporting and Data System category M (LR-M) remains a significant diagnostic challenge on conventional ultrasound examination. The LR-M category, originally intended to capture non-HCC malignancies, paradoxically contains up to 63% of atypical HCCs that deviate from classic enhancement patterns, leading to potential misdiagnosis and suboptimal treatment planning. While deep learning has shown promise in HCC diagnosis, most existing models rely exclusively on single-modality ultrasound, overlooking the diagnostic benefits of integrating complementary information from multiple imaging sources. To address this gap, we propose a novel attention-weighted tri-modal ultrasound network (TUS-Net) that integrates contrast-enhanced ultrasound (CEUS), B-mode ultrasound (BUS), and time-intensity curves (TICs) to improve diagnostic accuracy for these clinically challenging lesions. MethodsOur framework incorporates a three-dimensional convolutional neural network (C3D) backbone to extract spatiotemporal features from CEUS videos, capturing dynamic vascular patterns critical for lesion characterization. To effectively fuse complementary modalities, we introduce a dual-channel feature fusion module (DCFFM) that adaptively combines features from CEUS and BUS through channel-wise attention mechanisms, allowing the model to dynamically weigh the contribution of each modality based on diagnostic relevance. Additionally, we propose a temporal intensity feature fusion module (TIFFM) that leverages quantitative hemodynamic information from TICs to guide the model’s attention toward diagnostically critical temporal phases, such as arterial wash-in and portal venous washout. The model is further enhanced by automated lesion localization using YOLOX and class activation mapping for interpretability, ensuring that predictions align with clinically meaningful imaging features. ResultsEvaluated on a tri-modal ultrasound dataset comprising 161 patients with pathologically confirmed LR-M nodules (131 atypical HCC and 30 non-HCC malignancies), our model achieved an accuracy of 86.83%, a sensitivity of 92.50%, a specificity of 75.50%, and an AUC of 89.32% in screening atypical HCC. Compared to single-modality baselines, TUS-Net demonstrated superior specificity, a clinically critical metric given the higher risk associated with misclassifying non-HCC malignancies. Ablation studies confirmed the contribution of each module, with the full model outperforming both standard C3D and 3D ResNet backbones integrated with attention mechanisms. A reader study involving junior and senior radiologists further validated the clinical utility of AI assistance, showing consistent improvements in specificity and inter-reader consistency, particularly for less experienced clinicians. ConclusionThese results surpass existing benchmark models and demonstrate the potential of our approach to enhance diagnostic precision in clinically specific cases. By intelligently fusing multi-modal ultrasound data with attention-guided mechanisms, TUS-Net offers a reliable and interpretable tool that holds promise for improving the non-invasive diagnosis of atypical HCC in challenging LR-M liver nodules.

ObjectiveThis paper aims to observe the syndrome improvement and negative antigen conversion rate of Qingxuan Daozhi formula in the treatment of influenza in children （heat accumulation in the lung and stomach syndrome）. MethodsThrough a multi-center randomized controlled methodology design，confirmed influenza cases were collected from October 2022 to April 2023 in the pediatrics department of eight hospitals，such as Dongfang Hospital of Beijing University of Chinese Medicine. A total of 180 children with influenza and heat accumulation in the lung and stomach syndrome conforming to the standard were recruited through the clinic. The sick children meeting the inclusion criteria were randomly divided into groups by a block-randomized method. The children in the experimental group were treated with Qingxuan Daozhi formula for five days，and those in the control group were treated with Oseltamivir Phosphate Granules for five days. The primary efficacy indicator was the negative conversion rate of influenza antigen detection. Secondary efficacy indicators were the Canadian acute respiratory illness and flu scale （CARIFS） and the incidence of complications，severe cases， and critical cases. Follow-up observation was conducted on the day of enrollment，48 hours after medication，72 hours after medication， and （6+1） d after medication. ResultsOne hundred and eighty participants were randomly assigned to the experimental group （90 cases） or the control group （90 cases）. All participants were followed up during the study. Comparison of influenza antigen detection results in the primary efficacy indicators showed that the average time of negative influenza antigen conversion in the experimental group was （5.29±1.25） d，and that in the control group was （5.40±1.68） d，without a statistically significant difference. After five days of intervention，52 cases in the experimental group and 51 cases in the control group converted to negative，without a statistically significant difference. CARIFS score results in the secondary efficacy indicators showed that during 72 hours after intervention，there were statistically significant differences between the experimental group and the control group in three dimensions， including headache，muscle soreness， and the need for extra care （P<0.05）. On the （6+1） days after the intervention，the differences in both the experimental group and the control group were statistically significant in 10 dimensions， including sore throat，bad sleep，uncomfortable feeling，poor spirit and fatigue，crying more than usual，the need for extra care，symptom，function，influence on parents，and total score （P<0.05）. The comparison results within the group in the dimensional scores of symptom， function， and influence on parents，as well as the CARIFS total score showed that with the delay of follow-up time，scores of both groups decreased significantly，with a statistically significant difference （P<0.01）. Inter-group comparison results showed that the mean score of the experimental group was higher than that of the control group at the time of enrollment. With the progress of intervention，the score of the experimental group was significantly decreased compared with that of the control group. At the end of follow-up，the mean score of the experimental group was lower than that of the control group，with no statistically significant difference. In terms of the incidence of complications，severe cases， and critical cases， there were no complications，severe cases， and critical cases in the two groups，without a statistically significant difference. ConclusionThe symptom improvement effect and negative antigen conversion rate of Qingxuan Daozhi formula in the treatment of influenza in children （heat accumulation in the lung and stomach syndrome） are not inferior to Oseltamivir Phosphate granules， and children's acceptance is better. It can be more widely used in clinical treatment of influenza in children （heat accumulation in the lung and stomach syndrome）.

Chaihu Guizhi Ganjiangtang was first recorded in the Treatise on Cold Damage （Shang Han Lun）. This prescription is composed of Bupleuri Radix， Scutellariae Radix， Cinnamomi Ramulus， Zingiberis Rhizoma， Trichosanthis Radix， Ostreae Concha， and Glycyrrhizae Radix et Rhizoma. It has the effects of soothing Lesser Yang， warming the spleen， and stimulating the generation of body fluid. It is mainly used to treat digestive tract diseases such as chronic cholecystitis （CC）， irritable bowel syndrome， and non-alcoholic fatty liver disease. Dyspepsia caused by CC presents a variety of gastrointestinal symptoms such as abdominal pain， poor appetite， postprandial fullness， aversion to greasy food， soft stool， and bitter mouth， being a type of biliary dyspepsia. In modern medicine， dyspepsia caused by CC is mainly managed by medical treatment and surgical treatment. Internal medicine mainly focuses on reducing inflammation， promoting the function of gallbladder， resolving stones， alleviating spasms， and relieving the pain for CC， demonstrating definite short-term efficacy but suffering from single effects， high recurrence rate， and poor compliance. Although surgical treatment can cure cholecystitis， it is accompanied by the increased incidence of adverse events such as abdominal pain， diarrhea， and dyspepsia. Modern clinical studies have confirmed that Chaihu Guizhi Ganjiangtang can significantly alleviate the symptoms such as abdominal pain and dyspepsia of CC patients. Pharmacological studies have found that Chaihu Guizhi Ganjiangtang mainly contains active ingredients such as Bupleuri Radix saponins， baicalin， cinnamaldehyde， gingerol， Trichosanthis Radix polysaccharide， Ostreae Concha polysaccharide， and Glycyrrhizae Radix et Rhizoma total flavonoids. Chaihu Guizhi Ganjiangtang can ameliorate the symptoms of dyspepsia caused by CC by inhibiting inflammatory responses， improving gallbladder contraction and gastrointestinal motility， regulating the bile acid-intestinal flora axis and the brain-gut axis， and modulating blood lipids through multiple targets. By reviewing the previous literature， this article summarizes the research progress in the treatment of dyspepsia caused by CC with Chaihu Guizhi Ganjiangtang and its main active ingredients as well as the pathogenesis of this disease and puts forward the shortcomings and improvement strategies for the current research. The review aims to provide a reference for the further research on Chaihu Guizhi Ganjiangtang in the treatment of dyspepsia caused by CC.

AIM: To investigate the diagnostic and prognostic value of differential expression of Cyclin D1 and p53 in eyelid tumors.METHODS: This retrospective study enrolled patients who underwent surgical resection for eyelid tumors at our hospital between March 2018 and March 2023. Participants were categorized into benign and malignant groups based on tumor characteristics. Clinical data were collected. Genetic data for eyelid tumors were obtained from the GEO database, and differential gene analysis, including volcano plot visualization and KEGG pathway enrichment analysis, was performed using the Sangerbox 3.0 platform. Immunohistochemistry was used to detect the expression levels of Cyclin D1, p53, and BAX in tissue samples. Correlations with clinical features were analyzed using Spearman analysis, and prognostic factors were identified via Logistic regression analysis.RESULTS: This study included 69 patients with eyelid tumors(78 eyes), categorized into a benign group(37 patients, 41 eyes)and a malignant group(32 patients, 37 eyes)based on tumor characteristics. There were significant differences between the two groups in histological subtype, TNM staging, vascular invasion, differentiation status, and local infiltration(all P<0.05). Among benign tumors: pigmented nevi in 11 eyes(27%), hemangiomas in 9 eyes(22%), squamous cell papillomas in 5 eyes(12%), epidermoid cysts in 5 eyes(12%), seborrheic keratoses in 4 eyes(10%), neurofibromas in 3 eyes(7%), and both calcifying epithelioma and xanthelasma in 2 eyes each(5%); among malignant tumors: basal cell carcinoma in 18 eyes(49%), meibomian gland carcinoma in 8 eyes(22%), squamous cell carcinoma in 5 eyes(14%), sebaceous gland carcinoma in 4 eyes(11%), lymphoma and malignant melanoma each in 1 eye(3%). At the follow-up cutoff date of March 2025, the 2-year survival rate in the benign group(95%)was significantly higher than that in the malignant group(78%; P<0.05). Bioinformatics analysis identified 4 103 differentially expressed genes, including Cyclin D1, p53, and BAX, which were predominantly involved in pathways such as the p53 signaling pathway and calcium-related signaling. Spearman analysis revealed that local invasion(rs=0.71, P<0.05)and TNM stage(rs=0.73, P<0.05)correlated with Cyclin D1 expression; local invasion(rs=0.76, P<0.05)and histological subtype(rs=0.65, P<0.05)correlated with p53 expression. Logistic regression results indicated that Cyclin D1, p53, TNM staging, and local invasion were prognostic risk factors. ROC curve analysis demonstrated that the combined detection of these four indicators had the highest predictive value for prognosis(AUC=0.83).CONCLUSION: High expression of cyclin D1 and p53 serves as molecular markers for distinguishing benign from malignant eyelid tumors and assessing prognosis. Combined detection of these markers with TNM staging and local invasion demonstrates high predictive value for prognosis.

ObjectiveTo analyze the volatile components in raw Coptidis Rhizoma， Euodiae Fructus juice-processed Coptidis Rhizoma （Yuhuanglian）， and the processing adjuvant Euodiae Fructus juice， identify marker volatile components in the processed product， and elucidate the material basis changes during the processing with Euodiae Fructus juice. MethodsHeadspace solid-phase microextraction coupled with gas chromatography-quadrupole-time-of-flight mass spectrometry （HS-SPME-GC-Q-TOF-MS） was employed to analyze the volatile components in raw Coptidis Rhizoma， Yuhuanglian， and Euodiae Fructus juice. Chemometric analysis was performed to compare volatile components among samples and screen the marker components. Finally， the relative odor activity value （ROAV） was calculated to investigate odor characteristics. ResultsA total of 127 components were detected， with 27， 75， 84 volatile components identified in Euodiae Fructus juice， raw Coptidis Rhizoma， and Yuhuanglian， respectively. Processing with Euodiae Fructus juice significantly increased the number and content of volatile components. Chemometric analysis revealed significant changes in volatile components among the three sample types， allowing clear discrimination. After processing， 39 components such as methyl octanoate and allyl propionate were introduced， while 34 components including furfural and hexanoic acid increased in content， and 11 components including butyric acid and octanal decreased. Additionally， eight components， including zingiberene， 6-methyl-6-hepten-2-one， and 5-methyl-2-hexanone， were screened as marker components of Yuhuanglian. Moreover， processing enhanced the ROAV values of 2，5-dimethylpyrazine （roasted， nutty， coffee-like aroma）， 3-ethylphenol， and guaiacol （smoky， woody aroma）. Additionally， it incorporated 1-hepten-3-one from Euodiae Fructus juice and the newly generated 1-hexen-3-one， contributing mushroom and cucumber-like odors. The sensory profile of Yuhuanglian integrated nutty， roasted， smoky， and woody notes with the citrus and fruity aroma originally contributed by octanal. ConclusionThe processing with Euodiae Fructus juice introduced components from adjuvant， significantly altering the volatile component profile and enhancing the sensory odor characteristics of Yuhuanglian. This study provides a valuable reference for exploring the material basis of volatile components and establishing processing specifications for Yuhuanglian.

ObjectiveTo analyze the volatile components in raw Coptidis Rhizoma， Euodiae Fructus juice-processed Coptidis Rhizoma （Yuhuanglian）， and the processing adjuvant Euodiae Fructus juice， identify marker volatile components in the processed product， and elucidate the material basis changes during the processing with Euodiae Fructus juice. MethodsHeadspace solid-phase microextraction coupled with gas chromatography-quadrupole-time-of-flight mass spectrometry （HS-SPME-GC-Q-TOF-MS） was employed to analyze the volatile components in raw Coptidis Rhizoma， Yuhuanglian， and Euodiae Fructus juice. Chemometric analysis was performed to compare volatile components among samples and screen the marker components. Finally， the relative odor activity value （ROAV） was calculated to investigate odor characteristics. ResultsA total of 127 components were detected， with 27， 75， 84 volatile components identified in Euodiae Fructus juice， raw Coptidis Rhizoma， and Yuhuanglian， respectively. Processing with Euodiae Fructus juice significantly increased the number and content of volatile components. Chemometric analysis revealed significant changes in volatile components among the three sample types， allowing clear discrimination. After processing， 39 components such as methyl octanoate and allyl propionate were introduced， while 34 components including furfural and hexanoic acid increased in content， and 11 components including butyric acid and octanal decreased. Additionally， eight components， including zingiberene， 6-methyl-6-hepten-2-one， and 5-methyl-2-hexanone， were screened as marker components of Yuhuanglian. Moreover， processing enhanced the ROAV values of 2，5-dimethylpyrazine （roasted， nutty， coffee-like aroma）， 3-ethylphenol， and guaiacol （smoky， woody aroma）. Additionally， it incorporated 1-hepten-3-one from Euodiae Fructus juice and the newly generated 1-hexen-3-one， contributing mushroom and cucumber-like odors. The sensory profile of Yuhuanglian integrated nutty， roasted， smoky， and woody notes with the citrus and fruity aroma originally contributed by octanal. ConclusionThe processing with Euodiae Fructus juice introduced components from adjuvant， significantly altering the volatile component profile and enhancing the sensory odor characteristics of Yuhuanglian. This study provides a valuable reference for exploring the material basis of volatile components and establishing processing specifications for Yuhuanglian.

ObjectiveTo explore the potential mechanism by which Huoxue Wentong prescription （HXWT） ameliorates post-myocardial infarction （MI） ventricular remodeling by regulating endothelial-to-mesenchymal transition （EndMT） via the hypoxia-inducible factor-1α （HIF-1α）/transforming growth factor-β₁ （TGF-β₁）/periostin （POSTN） signaling pathway. MethodsEighty SPF-grade male Sprague-Dawley rats were randomly assigned to a sham-operated group （n=10） and a modeling group （n=70）. The MI model was established by left anterior descending coronary artery ligation. Successfully modeled rats were further randomized into a model group， a sacubitril/valsartan （Entresto， 0.021 g·kg^-1） group， and low， medium， and high-dose （4.15， 8.30， and 16.60 g·kg^-1， respectively） HXWTF groups （n=10 per group）. From postoperative day 2， rats received HXWTF or sacubitril/valsartan by gavage once daily for 4 weeks， and those in sham and model groups received an equal volume of normal saline. After intervention， general conditions were assessed. Echocardiography was performed to measure the left ventricular ejection fraction （LVEF）， left ventricular fractional shortening （LVFS）， left ventricular end diastolic diameter （LVEDD）， and left ventricular end systolic diameter （LVESD）. Hematoxylin and eosin （HE） staining was employed to evaluate myocardial histopathology， and Masson's trichrome staining to assess myocardial fibrosis. Serum levels of B-type natriuretic peptide （BNP）， interleukin （IL）-6， IL-10， and matrix metalloproteinase-9 （MMP-9） were determined by enzyme-linked immunosorbent assay （ELISA）. Collagen type Ⅰ （ColⅠ） expression was assessed by immunohistochemistry. Immunofluorescence double staining was adopted to detect the expression of platelet endothelial cell adhesion molecule-1 （CD31） and α-smooth muscle actin （α-SMA）. Western blot was performed to quantify the expression of collagen type Ⅰ α2 （Col1A2）， VE-cadherin， CD31， α-SMA， HIF-1α， TGF-β₁， and POSTN. ResultsCompared with the sham group， the model group exhibited thinning and collapse of the left ventricular wall， decreased LVEF and LVFS and increased LVEDD and LVESD （P<0.01）， disrupted myocardial fibers with marked inflammatory cell infiltration， increased collagen deposition， enhanced ColⅠ immunoreactivity， elevate serum BNP， IL-6， and MMP-9 levels， a reduced IL-10 level （P<0.01）， downregulated expression of CD31 and VE-cadherin， and upregulated expression of α-SMA， Col1A2， HIF-1α， TGF-β₁， and POSTN （P<0.05，P<0.01）， accompanied by increased CD31/α-SMA co-localization. Compared with the model group， the sacubitril/valsartan group and all HXWTF groups showed improved ventricular wall structure， increased LVEF and LVFS， decreased LVEDD and LVESD， attenuated myocardial injury with reduced inflammatory infiltration and collagen deposition， declined serum BNP， IL-6， and MMP-9 levels， an elevated serum IL-10 level， upregulated expression of CD31 and VE-cadherin， downregulated expression of α-SMA， Col1A2， HIF-1α， TGF-β₁， and POSTN， and reduced CD31/α-SMA co-localization （P<0.05，P<0.01）. No significant differences were observed between the sacubitril/valsartan group and the medium- and high-dose HXWTF groups. ConclusionHXWTF may regulate the HIF-1α/TGF-β₁/POSTN pathway to inhibit EndMT， thereby ameliorating post-MI ventricular remodeling.

Objective:To evaluate the sample preparation proficiency and storage proficiency of 11 clinical biobanks in Beijing through simulated experiments, and to establish an assessment method for the quality comparability of biological samples.Methods:An exploratory research design was adopted. In November 2023, artificial composite serum quality control materials containing six recombinant human protein markers—recombinant human alanine aminotransferase (rhALT), recombinant human aspartate aminotransferase (rhAST), recombinant human creatine kinase (rhCK), recombinant human creatine kinase-MB (rhCK-MB), recombinant human B-type natriuretic peptide (rhBNP), and recombinant human troponin I (rhTNI)—were distributed to 11 clinical biobanks in Beijing City. Sample preparation and storage followed the standardized operating procedures. Proficiency differences were assessed through statistical analysis.Results:Three-way repeated measures ANOVA revealed all six protein markers showed a declining trend over storage time in ultra-low-temperature environments ( F values 11.68-4 179.66, all P<0.01). However, neither long-term/temporary refrigerator types ( F values 0.01-1.23, all P>0.05)nor placement locations within refrigerators significantly affected the stability of these six proteins ( F valus 0.03-1.47, all P>0.05). The biases in detection results for rhALT, rhAST, rhTNI, and rhBNP at different storage time points were within the allowable bias limits for each item, supporting their use as markers for protein stability in biobank samples. All 11 institutions passed the storage proficiency assessment. In the preparation proficiency assessment, deviations were observed in post-preparation sample results, with a notably high out-of-control rate for rhCK (36.36%). Conclusion:Sample preparation proficiency can serve as a quality control metric for clinical biobanks. Future external quality assessment systems for biobanks should focus on sample preparation rather than storage processes.

Objective To observe brain structural changes in preterm infants and to analyze associated clinical risk factors based on 3D T1 structural MRI.Methods Brain 3D T1 structural MRI data of 82 preterm infants(preterm group)and 50 term infants(term group)were analyzed.Cortical morphology,including cortical thickness,surface area,sulcal depth and gyrification index were compared between groups.Spearman partial correlation analysis was used to explore the correlations of cortical structural changes and perinatal clinical variables.Results Compared with those in term group,increased cortical thickness of the right caudal middle frontal gyrus,reduced surface area of the left inferior parietal lobule,left precuneus and bilateral supramarginal gyrus,as well as decreased gyrification index in the right superior temporal gyrus,right lateral occipital gyrus,left inferior parietal lobule and left parahippocampal gyrus were observed in preterm group(all FDR corrected P＜0.05).No significant difference of sulcal depth was found between groups(all P＞0.05).Cortical surface area in bilateral supramarginal gyrus of preterm infant lowly-weakly negatively(rs=-0.327,-0.267,both P＜0.05)correlated,while the gyrification index in left parahippocampal gyrus of preterm infant weakly and positively(rs=0.221,P=0.045)correlated with maternal gestational diabetes mellitus.The surface area of left inferior parietal lobule,left precuneus,left supramarginal gyrus and right supramarginal gyrus in preterm infant weakly and negatively correlated with maternal infection during pregnancy(rs=-0.284—-0.224,all P＜0.05).Meanwhile,cortical thickness of the right caudal middle frontal gyrus and surface area of the right supramarginal gyrus in preterm infant lowly and negatively correlated with premature rupture of membranes(rs=-0.311,-0.301,both P＜0.05).Conclusion 3D T1 structural MRI was useful for detecting abnormal cortical morphology of preterm infants.Maternal gestational diabetes,infection during pregnancy and premature rupture of membranes might be risk factors for abnormal brain structure in newborns.