Combined model of radiomics and clinical features for predicting prognosis of term neonatal hypoxic-ischemic encephalopathy after one year: an exploratory study

Combined model of radiomics and clinical features for predicting prognosis of term neonatal hypoxic-ischemic encephalopathy after one year: an exploratory study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Combined model of radiomics and clinical features for predicting prognosis of term neonatal hypoxic-ischemic encephalopathy after one year: an exploratory study Jing Tang, Si-Ping He, Yu-Qing Liu, Yong-Hua Xiang, Ting Yi, Ke Jin This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6097732/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 29 Dec, 2025 Read the published version in BMC Medical Imaging → Version 1 posted 14 You are reading this latest preprint version Abstract Purpose The purpose of this study was to establish a combined model based on T1WI, T2WI, FLAIR images and clinical parameters to predict the prognosis of hypoxic-ischemic encephalopathy (HIE) in full-term newborns. Methods Based on the results of cognitive scores and motor function scores at 12 months post-birth, the patients were classified into two groups: Group B for those with good prognosis (n = 84) and Group W for those with poor prognosis (n = 96). A total of 180 patients were retrospectively evaluated and assigned to either the training data set (n = 126) or the testing data set (n = 54). The clinical characteristics of both groups were compared first. Then, a clinical model, a radiomics model, and a combined model were developed. Finally, we evaluated the performance of the three constructed models using the receiver operating characteristic (ROC) curve and the area under the curve (AUC). Results The Apgar scores at 1 minute, 5 minutes, and 10 minutes were all higher in Group B compared to Group W, with P < 0.05 indicating a statistically significant difference. The clinical model showed that the Apgar score at 10 minutes was the most effective factor, with an AUC of 0.857 in the training data set and an AUC of 0.737 in the testing data set. For the radiomics model, 9 radiomics features were found to be significantly related to predicting the prognosis of HIE, with AUCs of 0.916 and 0.770 in the training and testing data sets, respectively. For the combined model, 7 radiomics features, Apgar scores at 5 minutes, and Apgar scores at 10 minutes were independent predictors for predicting the prognosis of HIE, with AUCs of 0.952 and 0.823 in the training and testing data sets, respectively. The combined model demonstrates better performance than both the clinical and radiomics models. Conclusions The combined model, which incorporates MR-based radiomics signatures, and clinical factors, is effective in predicting the prognosis of HIE. hypoxic-ischemic encephalopathy radiomics Magnetic Resonance predict prognosis Figures Figure 1 Figure 2 Figure 3 1. Introduction Hypoxic-ischemic brain damage caused by perinatal asphyxia is known as neonatal hypoxic-ischemic encephalopathy(HIE)( 1 ). HIE is one of the primary reasons for acquired neonatal brain injury, which isalso the most common cause of neonatal encephalopathy ( 2 ).Theincidence rate of HIE is approximately 1–8 cases per 1000 total live births ( 3 , 4 ).The pathogenesis of HIE primarily involves the induction of cerebral cell necrosis and apoptosis. Severe and prolonged injury can result in diffuse and pronounced neuronal necrosis, causing significant and irreversible brain tissue damage ( 5 ).The prognosis of HIE depends on the severity of the condition. For patients with mild HIE, the prognosis for recovery is generally good, with most able to fully recover without long-term neurological issues. Moderate HIE patients also have a higher cure rate when receiving prompt and appropriate treatment, but some may still face sequelae such as slow intellectual development or limited motor function. In contrast, the mortality rate for severe HIE patients is relatively high, and those who survive often have severe neurological sequelae, such as cerebral palsy, intellectual disabilities, and epileptic seizures ( 6 ). Accurate short-term and long-term prognostic predictions indeed play a crucial role in the treatment and care of HIE. These predictions not only assist clinicians in formulating more precise treatment strategies but also provide essential information to the families of the affected infants regarding their prognosis, thereby helping them make informed decisions ( 7 ).Therefore, accurate prognostic predictors have been the focus of recent research in the field ofHIE Although conventional magnetic resonance imaging (MRI) is widely used in HIE, the information provided by human observation alone is limited. Radiomics is one of the most promising emerging methods for advancing imaging assessment, which enables high-throughput computational extraction and analysis of features from digital medical images and converts this information into mineable data. These data are then analyzed through feature selection to construct models for disease prediction and diagnosis ( 8 ).Currently, radiomics has been relatively widely applied in clinical practice, but there is still limited assessment of ischemic changes in full-term neonates. No study has yet utilized conventional MRI-based radiomics to evaluate the long-term prognosis of full-term HIE. This study aims to develop and validate clinical and radiomic models to predict the long-term prognosis of full-term HIE. 2. Materials and methods 2.1 Patients and data collection A retrospective study was conducted to screen 834 infants with HIE admitted to Hunan Children's Hospital from January 2013 to May 2023 for enrollment. Inclusion criteria were as follows: a. History of perinatal asphyxia; b. Clinical diagnosis of HIE or a clinical need for MRI to exclude or assess brain injury; c. Neonates with a gestational age of 37 weeks to less than 42 weeks who underwent MRI within 10 days after birth; d. Complete imaging data (T1WI, T2WI, FLAIR) with an MRI diagnosis of HIE; e. Clinical follow-up for intellectual and motor assessment at 12 months post-birth. Exclusion criteria were as follows: a. Possible brain injury caused by neonatal hyperbilirubinemia, intrauterine or intracranial infections, neonatal hypoglycemia, intracranial hemorrhage, congenital diseases, and metabolic diseases; b. Poor image quality unsuitable for analysis. After the enrollment screening, clinical and imaging data were collected from 198 infants. Following post-processing with radiomics software, 180 images met the requirements and were used for feature analysis and data modeling. This retrospective study was approved by the Medical Ethics Committee of the Hunan Children’s Hospital of SouthChina University.The subject selection processis shown in Fig. 1 Clinical data were collected from the infants, including variables related to their birth history such as gestational age at birth, mode of delivery, gender, birth weight, head circumference, and Apgar scores, as well as the age at which the MRI scan was performed. At 12 months of age, the infants' cognitive abilities were assessed using the Bayley Scales of Infant Development (BSID), and their motor functions were evaluated using the Gross Motor Function Measure-88 (GMFM-88). Based on the results of the cognitive and motor function assessments, the infants were classified into two groups:Group B for those with a good prognosis and Group W for those with a poor prognosis. The clinical features, including gestational age at birth, mode of delivery, age, gender, birth weight, head circumference, and Apgar scores, were compared between Group B and Group W. 2.2 MR image acquisition The MRI examinations were conducted when the infants' conditions had reached a basically stable state. We used the Siemens Skyra or Prisma 3.0T MRI scanner and an eight-channel head coil with the same MRI parameters for the examinations. The specific scanning sequences included axial T1WI, T2WI, FLAIR, and sagittal T1WI. The parameters for the sagittal T1WI were as follows: repetition time (TR) of 400 ms and echo time (TE) of 8.1 ms. For the axial T1WI, TR was 468 ms and TE was 11 ms; for the axial T2WI, TR was 4000 ms and TE was 101 ms; for the axial FLAIR, TR was 9000 ms and TE was 98 ms. The slice thickness was set to 4 mm with an inter-slice gap of 0.32 mm. Prior to segmentation and feature extraction, the MR images were preprocessed to eliminate any potential differences that might exist between images acquired from the two different scanners. 2.3 Establishment of clinical model The clinical model was constructed by incorporating clinical features that significantly differed between Group B and Group W. Feature selection was performed using the Kruskal-Wallis test, and Support Vector Machine (SVM) was used as the classification tool. The hyperparameters of the SVM model were determined using Leave-One-Out Cross-Validation (LOOCV) on the training dataset. 2.4 Establishment of radiomics model 2.5 ROI drawing and image analysis Two radiologists with over ten years of experience delineated the region of interest (ROI) using ITK-SNAP software (available from http://www.itksnap.org ). The ROIs were all placed in the basal ganglia region with the largest diameter, avoiding the venous sinuses, veins, and cerebrospinal fluid as much as possible. Please refer to Fig. 2. 2.6 Radiomics feature-extraction and model evaluation We selected 126 cases as the training dataset (67 in Group W and 59 in Group B). Additionally, we chose another 54 cases as the independent testing dataset (29 in Group W and 25 in Group B). To address the imbalance in the training dataset, we utilized the Synthetic Minority Oversampling Technique (SMOTE) to balance the samples between Group W and Group B. Subsequently, we normalized the feature matrix by subtracting the mean value from each feature vector and dividing it by its length.Due to the high dimensionality of the feature space, we compared the similarity between each pair of features. If the Pearson Correlation Coefficient (PCC) value between a pair of features exceeded 0.990, we removed one of them. Through this process, the dimensionality of the feature space was reduced, and each feature became independent of the others.Prior to model construction, we employed Recursive Feature Elimination (RFE) for feature selection. The goal of RFE is to select features based on a classifier by recursively considering smaller subsets of features. We chose the SVM as the classifier, given its effectiveness and robustness in model building. The kernel function has the capability to map features into a higher dimension to find the hyperplane that separates samples with different labels. In this case, we used a linear kernel function because it facilitates the interpretation of feature coefficients in the final model.Todetermine the model's hyperparameters (such as the number of features), we applied LOOCV in the training dataset. In the training date set with 126 samples, using the leave-one-out method, we train the model with 125 samples each time and test it with the remaining 1 sample. By performing cross-validation 126 times with mutually exclusive sets, we can obtain the cv training date set and thevalidation date set. The hyperparameters were set based on the model's performance on the validation dataset. All of the above processes were implemented using FeAture Explorer Pro (FAE, version 0.5.8) software on Python (version 3.7.6). 2.7 Establishment of combined model A combined model for predicting poor and good prognosis was developed using the same method as the established radiomics model, which integrates MR-based radiomic features and clinical factors. 2.8 Statistical analysis All statistical analyses were performed using SPSS 25.0 and R statistical software (version 3.6.1). For the continuous variables, the Kolmogorov-Smirnov and Levene tests were first performed to verify the normality and homogeneity of the samples. Variables conforming to the normal distribution were expressed as Mean ± SD. Group comparisons were made using the t-test. Categorical variables were expressed as frequencies and percentages, and comparisons between groups were made using the χ² test. We plotted the Receiver Operating Characteristic (ROC) curve and comprehensively evaluated the model's performance in predicting the prognosis of HIE by calculating indicators including the area under the curve (AUC), accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). We used the bootstrap method to estimate the 95% confidence intervals. A two-sided P < 0.05 indicated a statistically significant difference. 3. Results 3.1 Patient characteristics A total of 180 full-term newborns were included in this study, with 84 in Group B and 96 in Group W. The detailed clinical parameters of patients in Group B and Group W are presented in Table 1 . The results in Table 1 revealed that there were no statistically significant differences in age, gender, weight, head circumference, mode of delivery, or gestational age of the mothers between the two groups. The Apgar scores at 1 minute, 5 minutes, and 10 minutes were all higher in Group B compared to Group W, with P < 0.05 indicating a statistically significant difference. Table 1 Comparison of clinical data in Group B and Group W Variables Group B(N = 84) No. of patient (%) GroupW(N = 96) No. of patient (%) P value Age (days) Mean ± SD 6.75 ± 2.8 6.59 ± 3.14 0.71 Gender 0.26 Male 64(76.19) 66(68.75) Female 20(23.81) 30(31.25) Delivery method 0.32 Vaginal delivery 50(59.52) 64(66.7) Cesarean section 34(40.48) 32(33.3) Gestational week 39.06 ± 1.28 39.54 ± 1.29 0.06 Weight(kg) 3.23 ± 0.54 3.33 ± 0.48 0.17 Head circumference 33.48 ± 1.25 33.82 ± 1.31 0.08 Apgar scores at 1minite 5.66 ± 2.72 3.25 ± 2.01 <0.01 Apgar scores at 5minite 7.88 ± 2.02 5.91 ± 1.88 <0.01 Apgar scores at 10minite 9.06 ± 1.29 7.20 ± 1.56 <0.01 3.2 Performance of clinical model The clinical model showed that the Apgar score at 10 minutes was the most effective factor for differentiating Group B from Group W, with an AUC of 0.857 and an accuracy of 0.833 in the training dataset. At this point, the AUC, accuracy, sensitivity, and specificity of the clinical model achieved 0.737, 0.778, 0.965, and 0.56 in the testing dataset, respectively (Table 3 , Fig. 3 .A). 3.3 Performance of radiomics model A total of 4,227 radiomic features were extracted from T1WI, T2WI, and FLAIR images. These features included shape, first-order statistics, texture, and higher-order statistical features. After ranking these features, the nine best-performing significant radiomic features were selected. The AUC and accuracy could achieve 0.916 and 0.841 in the training dataset, respectively. At this point, the AUC, accuracy, sensitivity, and specificity of the radiomics model achieved 0.770, 0.778, 0.862, and 0.680 in the testing dataset(Fig. 3 .B). The final filtered radiomic features and corresponding model coefficients are presented in Table 2 . 3.4 Performance of combined model The combined model indicated that Apgar scores at 10 minutes, Apgar scores at 5 minutes, and seven best-performing significant radiomic features were independent predictors for differentiating Group W from Group B. The seven best-performing significant radiomic features were wavelet LLL glcm imc2, logarithm glcm inverse variance, wavelet HLL first-order mean, wavelet HLL first-order skewness, gradient glszm large area high gray level emphasis, logarithm gldm large dependence low gray level emphasis, and logarithm glrlm low gray level run emphasis. The AUC and accuracy could achieve 0.952 and 0.900 in the training dataset, respectively. At this point, the AUC, accuracy, sensitivity, and specificity of the combined model achieved 0.823, 0.796, 0.862, and 0.720 in the testing dataset (Table 3 , Fig. 3 .C). 3.5 Comparation of the assessment indicatorsof three models in the testing date set For the testing dataset, the AUC of the combined model was larger than that of the clinical and radiomics models. The clinical model had the highest sensitivity (0.966) and NPV (0.933), and the combined model had the highest specificity (0.931) and PPV (0.925). Table 2 ༎The coefficients of features in the model Features Coef in model Clinical model Apgar scores at 10 minute -5.667 Radiomics model FLAIR_wavelet-LLL_glcm_Imc2 T1WI_log-sigma-1-mm-3D_gldm_LargeDependenceLowGrayLevelEmphasis T1WI_logarithm_glcm_InverseVariance T1WI_wavelet-HLL_firstorder_Mean T1WI_wavelet-LLH_firstorder_Skewness T2WI_gradient_glszm_LargeAreaHighGrayLevelEmphasis T2WI_logarithm_gldm_LargeDependenceLowGrayLevelEmphasis T2WI_logarithm_glrlm_LowGrayLevelRunEmphasis T2WI_squareroot_gldm_LargeDependenceLowGrayLevelEmphasis -1.111 -0.014 1.129 -2.172 2.743 -2.164 -0.825 -1.914 -0.461 Combined model Apgar scores at 10 minute Apgar scores at 5 minute FLAIR_wavelet-LLL_glcm_Imc2 T1WI_logarithm_glcm_InverseVariance T1WI_wavelet-HLL_firstorder_Mean T1WI_wavelet-LLH_firstorder_Skewness T2WI_gradient_glszm_LargeAreaHighGrayLevelEmphasis T2WI_logarithm_gldm_LargeDependenceLowGrayLevelEmphasis T2WI_logarithm_glrlm_LowGrayLevelRunEmphasis -2.566 -0.632 -0.875 0.935 –1.787 2.205 -1.362 -0.768 -1.969 Table 3 Performance of the clinical, radiomics, and combined models Group AUC (95% CI) Cut-off Accuracy Sensitivity Specificity PPV NPV Training data set Clinical model 0.857(0.789–0.925) 0.602 0.833 0.881 0.780 0.820 0.852 Radiomics model 0.916(0.870–0.962) 0.349 0.841 0.925 0.746 0.805 0.898 Combined model 0.952(0.919–0.985) 0.650 0.900 0.866 0.932 0.936 0.860 CV training date set Clinical model 0.856(0.850–0.862) 0.573 0.828 0.881 0.776 0.800 0.867 Radiomics model 0.912(0.908–0.917) 0.517 0.833 0.822 0.845 0.841 0.826 Combined model 0.951(0.948–0.953) 0.642 0.891 0.851 0.931 0.925 0.862 validation data set Clinical model 0.843(0.765–0.921) 0.610 0.865 0.881 0.848 0.868 0.862 Radiomics model 0.867(0.804–0.930) 0.389 0.825 0.881 0.763 0.808 0.850 Combined model 0.931(0.890–0.973) 0.660 0.881 0.851 0.915 0.919 0.844 testing data set Clinical model 0.737(0.595–0.879) 0.602 0.778 0.966 0.560 0.718 0.933 Radiomics model 0.770(0.641–0.898) 0.297 0.778 0.862 0.680 0.758 0.810 Combined model 0.823(0.705–0.942) 0.380 0.796 0.851 0.931 0.925 0.862 4. Discussion The development of accurate prognostic indicators has been a focus of research in the field of neonatal hypoxic-ischemic encephalopathy (HIE) in recent years. Currently known prognostic indicators include serum proteins ( 9 – 11 ), neurophysiological examinations ( 12 , 13 ), and neuroimaging studies. As a non-invasive technique, brain MRI plays a crucial role in precisely locating and assessing brain lesions. One of the biggest challenges in prediction by traditional anatomic MRI (such as T1- and T2-weighted images) is the low sensitivity. Children with HIE who experience severe neurological outcomes may lack apparent MRI findings during the neonatal period (the first 28 days after birth) ( 14 – 16 ). Radiomics, which extracts sub-visual but quantitative imaging features from radiological images ( 17 , 18 ), can compensate for this limitation. In this study, there were no statistically significant differences in age, gender, weight, head circumference, mode of delivery, or gestational age of the mothers between Group B and Group W. However, the Apgar scores at 1 minute, 5 minutes, and 10 minutes were all higher in Group B compared to Group W, with statistically significant differences. The clinical model showed that the Apgar score at 10 minutes was the most effective factor for differentiating Group B from Group W, with an AUC of 0.843 and an accuracy of 0.865 on the validation dataset, and an AUC of 0.737 and an accuracy of 0.778 on the testing dataset, respectively. Some studies have shown that the risk of adverse outcomes increases when the Apgar score remains low at 10 minutes and beyond ( 19 – 22 ). Vivek V. Shukla's research suggests that a standalone 10-minute Apgar score of 0 does not accurately predict the risk of death or moderate to severe disability, but the predictive accuracy improves when the 10-minute Apgar score is combined with other risk variables available during resuscitation using a classification and regression tree analysis ( 23 ). In our study, we also found that the clinical model had the lowest AUC and accuracy among the three models. The combined model incorporating the Apgar score and radiomic features significantly improved the AUC and accuracy. For establishing the radiomics model, we selected the bilateral basal ganglia and thalamus as the ROI, primarily because these areas are more susceptible to damage in neonates with HIE ( 24 ). HIE can result in a series of biochemical changes, and patients undergo a complex pathophysiological process that includes hypoxia, ischemia, and subsequent reperfusion injury. Some scholars believe that during this process, when ischemic tissues regain blood supply (i.e., reperfusion), the resulting damage may be more severe than that caused by ischemia alone. This is because during reperfusion, the cells and tissues that were initially damaged by ischemia may suffer further injury, leading to a deterioration of the condition. Multiple studies utilizing ASL to assess changes in cerebral hemodynamics in neonates with HIE have consistently found early hyperperfusion of the basal ganglia and thalamus ( 25 – 27 ). Additionally, the thalamus, as one of the most metabolically active regions in the human body, has frequent cellular activity and high energy demands. Therefore, even mild HIE can trigger a certain degree of cellular apoptosis in the thalamus. In cases of mild HIE, conventional MRI examinations often fail to directly observe white matter damage and injuries to deep brain nuclei with the naked eye, but radiomics can analyze changes in their internal characteristics. In this study, we established a radiomic model for predicting the prognosis of HIE based on T1WI, T2WI, and FLAIR. This radiomic model was constructed using 9 non-zero coefficient features extracted from the original images, as well as images processed with Gaussian filters and wavelet transforms. These features include 2 first-order statistical features and 7 texture features. The most predictive feature was T1WI_wavelet-LLH_firstorder_Skewness. The radiomics results revealed the ability of radiomics to predict the prognosis of HIE in the training dataset (AUC = 0.916), the validation dataset (AUC = 0.867), and the testing dataset (AUC = 0.77), thereby indicating that radiomics can predict the prognosis of HIE. Youwon Shin et al. ( 28 ) enrolled 46 preterm neonates who underwent brain MRI near or at term-equivalent age, and neurodevelopment was assessed at a corrected age of 12 months. A prediction model for the binary classification of the psychomotor developmental index was developed. The AUCs of prediction models based on T1WI, T2WI, and both T1WI and T2WI were 0.925, 0.834, and 0.902, respectively. This finding was consistent with our results, indicating that it is feasible to establish a prognostic prediction model for neonatal brain injury using conventional MRI sequences. Through feature selection, this study identified 2 clinical features and 7 radiomic features to establish a combined model. The clinical features were the Apgar scores at 5 minutes and the Apgar scores at 10 minutes. Among the 7 radiomic features, 1 was obtained from FLAIR images, 3 were derived from T1WI images, and 3 were from T2WI images. The two most critical feature parameters in the combined model were the Apgar scores at 10 minutes and T1WI_wavelet-LLH_firstorder_Skewness. The combined model showed a great ability to predict the prognosis of HIE, and the highest AUCs for the training dataset, validation dataset, and testing dataset were 0.952, 0.931, and 0.823, respectively, which were higher than those of the radiomics model (training dataset: AUC = 0.916; validation dataset: AUC = 0.867; testing dataset: AUC = 0.770) and the clinical model (training dataset: AUC = 0.857; validation dataset: AUC = 0.843; testing dataset: AUC = 0.737). The clinical model had the highest sensitivity (0.966) and NPV (0.933), while the combined model had the highest specificity (0.931) and PPV (0.925). This means that the clinical model was better at evaluating children with poor prognosis, while the combined model was better at evaluating children with good prognosis. There were still some limitations to this study. Firstly, it was a single-center retrospective analysis, lacking multi-center validation of the model with external data. Future research will require a larger sample size to evaluate the model. Secondly, in this study, the image segmentation was conducted through manual delineation, which was time-consuming and inevitably subject to human error. We plan to explore semi-automatic or deep learning-based automatic segmentation methods to enhance the objectivity of our research methodology. Thirdly, we utilized conventional MRI sequences; therefore, in the future, we will incorporate more sensitive MRI sequences such as Diffusion Weighted Imaging (DWI) and Susceptibility Weighted Imaging (SWI) to improve the efficacy of the prognostic prediction model. In conclusion, the clinical model and radiomics model constructed in this study performed well in predicting the prognosis of HIE. Based on this, the combined model, which integrates clinical factors and MR radiomics, further improved the accuracy of predicting the prognosis of HIE. This model is objective, non-invasive, and reproducible. Declarations 5.1 Ethics This study constitutes a non-interventional investigation that does not involve human clinical trials or experimental interventions. In accordance with international research ethics guidelines, formal trial registration is not applicable. The research protocol adheres to the ethical principles outlined in the Declaration of Helsinki for non-clinical studies, particularly regarding data integrity and analytical rigor. 5.2 Consent to Participate Written informed consent was obtained from all individual participants (or their legal guardians in cases involving minors) prior to their inclusion in the study. Participants were informed about the research purpose, methodology, potential risks, and benefits, and their right to withdraw at any stage without consequences. 5.3 Consent to Publish For studies involving identifiable personal data (e.g., clinical images, case details), written consent for publication was secured from all participants or their legal representatives. Anonymity was maintained for non-essential identifying information. 5.4 Competing Interests The authors declare no competing financial or non-financial interests relevant to this work. 5.5 Author Contributions Guarantor of integrity of the entire study: Ke Jin and Ting Yi. Study concepts and design: Jing Tang, Ting Yi and Ke Jin. Literature research: Jing Tang and Ting Yi. Clinical studies: Jing Tang and Siping He; Experimental studies/data analysis: Ting Yi, Jing Tang and Yonghua Xiang. Statistical analysis: Ting Yia and Yuqing Liu. Manuscript preparation: Jing Tang. Manuscript editing: Jing Tang and Ting Yi. 5.6 Data Availability The data that support the findings of this study are available from the corresponding author upon reasonable request. 6. Acknowledgments This work was supported by the Hunan Provincial Science and Technology Innovation Plan (Natural Science Foundation of Hunan Province, Clinical Medical Technology Innovation Guidance Project, Grant No. 2021SK50511). References Parmentier CEJ, de Vries LS, Groenendaal F. Magnetic Resonance Imaging in (Near-)Term Infants with Hypoxic-Ischemic Encephalopathy. Diagnostics(Basel). 2022;12(3):645. Allen KA, Brandon DH. Hypoxic Ischemic Encephalopathy: Pathophysiology and Experimental Treatments. Newborn Infant Nurs Rev. 2011;11(3):125–33. Kurinczuk JJ, White-Koning M, Badawi N. Epidemiology of neonatal encephalopathy and hypoxicischaemic encephalopathy. Early Hum Dev. 2010;86(6):329–38. Chau V, Poskitt KJ, Dunham CP, et al. Magnetic resonance imaging in the encephalopathic term newborn. Curr Pediatr Rev. 2014;10(1):28–36. Onda K, Chavez-Valdez R. GrahamEM,etal.Quantification of diffusion MRI for prognostic prediction of neonatal hypoxic-ischemic encephalopathy. Dev Neurosci. 2024;46(1):55–68. Yildiz EP, EkiciB,Tatil B. Neonatal hypoxic ischemic encephalopathy:an update on disease pathogenesis and treatment. Expert Rev Neurother. 2017;17(5):449–59. Robertson CM, Perlman M. Follow-up of the term infant after hypoxic-ischemic encephalopathy. Paediatrics& Child Health. 2006;11(5):278–82. Rizzo S, Botta F, Raimondi S, et al. Radiomics: the facts and the challenges of image analysis. EurRadiol Exp. 2018;2(1):36. Douglas-Escobar M, Weiss MD. Biomarkers of hypoxic-ischemic encephalopathy in newborns. Front Neurol. 2012;3:144. Einspieler C, Prechtl HF, Ferrari F, et al. The qualitative assessment of general movements in preterm, term and young infants—review of the methodology. Early Hum Dev. 1997;50(1):47–60. Haataja L, Mercuri E, Regev R, et al. Optimality score for the neurologic examination of the infant at 12 and 18 months of age. J Pediatr. 1999;135(2):153–61. Lori S, Bertini G, Molesti E, et al. The prognostic role of evoked potentials in neonatal hypoxic-ischemic insult. J Maternal-Fetal Neonatal Med. 2011;24(sup1):69–71. Stewart AM, Chapman KE. Prognostication in Pediatrics. In: Husain AM, Sinha SR, editors. Continuous EEG Monitoring: Principles and Practice. Cham: Springer International Publishing; 2017. pp. 465–81. Martinez-Biarge M, Diez-Sebastian J, Kapellou O, et al. Predicting motor outcome and death in term hypoxic-ischemic encephalopathy. Neurology. 2011;76(24):2055–61. van Laerhoven H, de Haan TR, Offringa M, et al. Prognostic tests in term neonates with hypoxic-ischemic encephalopathy: a systematic review. Pediatrics. 2013;131(1):88–98. Utsunomiya H. Diffusion MRI abnormalities in pediatric neurological disorders. Brain Develop. 2011;33(3):235–42. Parmar C, Leijenaar RT, Grossmann P, et al. Radiomic feature clusters and prognostic signatures specific for lung and head & neck cancer. Sci Rep. 2015;5:11044. Gillies RJ, Kinahan PE, Hricak H. Radiomics: images are more than pictures they are data. Radiology. 2016;278:563–77. American Academy of Pediatrics, Committee on Fetus and Newborn. American College of Obstetricians and Gynecologists and Committee on Obstetric Practice. The Apgar score. Pediatrics. 2006;117:1444–7. Casalaz DM, Marlow N, Speidel BD. Outcome of resuscitation following unexpected apparent stillbirth. Arch Dis Child Fetal Neonatal Ed. 1998;78:F112–5. Natarajan G, Shankaran S, LaptookAR, et al. Apgar scores at 10 min and outcomes at 6–7 years following hypoxic-ischaemic encephalopathy. Arch Dis Child Fetal Neonatal Ed. 2013;98(6):F473–479. Pavel AM, O'Toole JM, Proietti J, et al. Machine learning for the early prediction of infants with electrographic seizures in neonatal hypoxic-ischemic encephalopathy. Epilepsia. 2023;64(2):456–68. Shukla VV, Bann CM, Ramani M, et al. Predictive Ability of 10-Minute Apgar Scores for Mortality and Neurodevelopmental Disability. Pediatrics. 2022;149(4):e2021054992. Wu TW. TamraziB,HsuKH,et al. Cerebral lactate concentration inneonatal hypoxic-ischemic encephalopathy:in relation to time,characteristic of injury,and serum lactate concentration. Front Neurol. 2018;9:293. Wang J, Li J, Yin X, et al. Cerebral hemodynamics of hypoxic-ischemic encephalopathy neonates at different ages detected by arterial spin labeling imaging. Clin HemorheolMicrocirc. 2022;81(4):271–9. Wang J, Li J, Yin X, et al. The Value of Arterial Spin Labeling Imaging in the Classification and Prognostic Evaluation of Neonatal Hypoxic-ischemic Encephalopathy. Curr Neurovasc Res. 2021;18(3):307–13. Proisy M, Corouge I, LegouhyA, et al. Changes in brain perfusion in successive arterial spin labeling MRI scans in neonates with hypoxic-ischemic encephalopathy. Neuroimage Clin. 2019;24:101939. YouwonShin. Brain MRI radiomics analysis may predict poor psychomotor outcome in preterm neonates. EurRadiol. 2021;31(8):6147–55. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 29 Dec, 2025 Read the published version in BMC Medical Imaging → Version 1 posted Editorial decision: Revision requested 30 Sep, 2025 Reviewers agreed at journal 29 Sep, 2025 Reviews received at journal 29 Sep, 2025 Reviews received at journal 29 Sep, 2025 Reviewers agreed at journal 24 Sep, 2025 Reviewers agreed at journal 24 Sep, 2025 Reviews received at journal 23 Sep, 2025 Reviewers agreed at journal 21 Sep, 2025 Reviewers agreed at journal 16 Jul, 2025 Reviewers agreed at journal 16 Jul, 2025 Reviewers invited by journal 06 Jul, 2025 Editor assigned by journal 04 Apr, 2025 Submission checks completed at journal 04 Apr, 2025 First submitted to journal 24 Feb, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6097732","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":481816123,"identity":"3099ef4b-c98d-413c-a285-ec5e1e524e72","order_by":0,"name":"Jing Tang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAy0lEQVRIie3PMQrCQBCF4QmBpBmwHVHIFSYIYhHIQWw2CJtGIWBrERurXEA8RW4QCFh5gLWLCKnXXtCAjV02neD+9XwwD8Bm+8FGQZM+NEcY+LkhGeeidSmT07CoDAlXSQuo6wiUMP2sqmVD7KJzvJcKdtGyVzj7Q8oZe+hP5HYBZ7nJ+4jroiBiROe0npOT1/3E80aakAnhejEkiCA7wggKDQkRrGbEAsOi2yJMtsQKkpt+vuLAr0uld1E/+Y5JDDn/kKHCZrPZ/qM3YNU64qjasZcAAAAASUVORK5CYII=","orcid":"","institution":"Central South University","correspondingAuthor":true,"prefix":"","firstName":"Jing","middleName":"","lastName":"Tang","suffix":""},{"id":481816124,"identity":"028dc588-add8-4cc5-a82d-31ccb520c76c","order_by":1,"name":"Si-Ping He","email":"","orcid":"","institution":"Central South University","correspondingAuthor":false,"prefix":"","firstName":"Si-Ping","middleName":"","lastName":"He","suffix":""},{"id":481816125,"identity":"058440e0-aadc-4648-9677-288fbdeb4fd2","order_by":2,"name":"Yu-Qing Liu","email":"","orcid":"","institution":"Central South University","correspondingAuthor":false,"prefix":"","firstName":"Yu-Qing","middleName":"","lastName":"Liu","suffix":""},{"id":481816126,"identity":"a36c9a01-c73a-4599-b701-1dfbbb1508a6","order_by":3,"name":"Yong-Hua Xiang","email":"","orcid":"","institution":"Central South University","correspondingAuthor":false,"prefix":"","firstName":"Yong-Hua","middleName":"","lastName":"Xiang","suffix":""},{"id":481816127,"identity":"f0df0e52-6d76-4025-9971-db7aa76084ea","order_by":4,"name":"Ting Yi","email":"","orcid":"","institution":"Central South University","correspondingAuthor":false,"prefix":"","firstName":"Ting","middleName":"","lastName":"Yi","suffix":""},{"id":481816128,"identity":"5e7b2fd3-0115-4a30-8fa4-20dc4a5495c1","order_by":5,"name":"Ke Jin","email":"","orcid":"","institution":"Central South University","correspondingAuthor":false,"prefix":"","firstName":"Ke","middleName":"","lastName":"Jin","suffix":""}],"badges":[],"createdAt":"2025-02-24 14:23:10","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6097732/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6097732/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12880-025-02020-5","type":"published","date":"2025-12-29T15:57:47+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":86625725,"identity":"5a2f985b-f706-41bc-82c2-297014ea4a1b","added_by":"auto","created_at":"2025-07-14 05:13:14","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":18122,"visible":true,"origin":"","legend":"\u003cp\u003eFlow chart illustrating enrolment of the study population.\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6097732/v1/3ba9a3cc97f3434880ff1443.jpg"},{"id":86626544,"identity":"da944893-da2f-49d3-93b5-800205fa28cb","added_by":"auto","created_at":"2025-07-14 05:30:09","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":21376,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart illustrating the process of clinical、radiomics and combined model.\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6097732/v1/19bd3ec864fccdf2ffc6a9a3.jpg"},{"id":86626550,"identity":"650ceb25-978c-4cd2-bd94-866c82544c47","added_by":"auto","created_at":"2025-07-14 05:30:09","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":130975,"visible":true,"origin":"","legend":"\u003cp\u003eThe receiver operating characteristic curve analyses in the clinical model (A), radiomicsmodel (B), and combined model (C)\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6097732/v1/f8e57c70efe7bec73b3d1257.jpg"},{"id":99545282,"identity":"13c9c220-0921-48d6-8f97-fadbb6a68242","added_by":"auto","created_at":"2026-01-05 16:05:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1108615,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6097732/v1/1f27b911-43f5-4800-bb19-32b81bd70a61.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Combined model of radiomics and clinical features for predicting prognosis of term neonatal hypoxic-ischemic encephalopathy after one year: an exploratory study","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eHypoxic-ischemic brain damage caused by perinatal asphyxia is known as neonatal hypoxic-ischemic encephalopathy(HIE)(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). HIE is one of the primary reasons for acquired neonatal brain injury, which isalso the most common cause of neonatal encephalopathy (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e).Theincidence rate of HIE is approximately 1\u0026ndash;8 cases per 1000 total live births (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e).The pathogenesis of HIE primarily involves the induction of cerebral cell necrosis and apoptosis. Severe and prolonged injury can result in diffuse and pronounced neuronal necrosis, causing significant and irreversible brain tissue damage (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e).The prognosis of HIE depends on the severity of the condition. For patients with mild HIE, the prognosis for recovery is generally good, with most able to fully recover without long-term neurological issues. Moderate HIE patients also have a higher cure rate when receiving prompt and appropriate treatment, but some may still face sequelae such as slow intellectual development or limited motor function. In contrast, the mortality rate for severe HIE patients is relatively high, and those who survive often have severe neurological sequelae, such as cerebral palsy, intellectual disabilities, and epileptic seizures (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eAccurate short-term and long-term prognostic predictions indeed play a crucial role in the treatment and care of HIE. These predictions not only assist clinicians in formulating more precise treatment strategies but also provide essential information to the families of the affected infants regarding their prognosis, thereby helping them make informed decisions (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e).Therefore, accurate prognostic predictors have been the focus of recent research in the field ofHIE\u003c/p\u003e\u003cp\u003eAlthough conventional magnetic resonance imaging (MRI) is widely used in HIE, the information provided by human observation alone is limited. Radiomics is one of the most promising emerging methods for advancing imaging assessment, which enables high-throughput computational extraction and analysis of features from digital medical images and converts this information into mineable data. These data are then analyzed through feature selection to construct models for disease prediction and diagnosis (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e).Currently, radiomics has been relatively widely applied in clinical practice, but there is still limited assessment of ischemic changes in full-term neonates. No study has yet utilized conventional MRI-based radiomics to evaluate the long-term prognosis of full-term HIE. This study aims to develop and validate clinical and radiomic models to predict the long-term prognosis of full-term HIE.\u003c/p\u003e"},{"header":"2. Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 Patients and data collection\u003c/h2\u003e\u003cp\u003eA retrospective study was conducted to screen 834 infants with HIE admitted to Hunan Children's Hospital from January 2013 to May 2023 for enrollment. Inclusion criteria were as follows: a. History of perinatal asphyxia; b. Clinical diagnosis of HIE or a clinical need for MRI to exclude or assess brain injury; c. Neonates with a gestational age of 37 weeks to less than 42 weeks who underwent MRI within 10 days after birth; d. Complete imaging data (T1WI, T2WI, FLAIR) with an MRI diagnosis of HIE; e. Clinical follow-up for intellectual and motor assessment at 12 months post-birth. Exclusion criteria were as follows: a. Possible brain injury caused by neonatal hyperbilirubinemia, intrauterine or intracranial infections, neonatal hypoglycemia, intracranial hemorrhage, congenital diseases, and metabolic diseases; b. Poor image quality unsuitable for analysis. After the enrollment screening, clinical and imaging data were collected from 198 infants. Following post-processing with radiomics software, 180 images met the requirements and were used for feature analysis and data modeling. This retrospective study was approved by the Medical Ethics Committee of the Hunan Children\u0026rsquo;s Hospital of SouthChina University.The subject selection processis shown in Fig.\u0026nbsp;1\u003c/p\u003e\u003cp\u003eClinical data were collected from the infants, including variables related to their birth history such as gestational age at birth, mode of delivery, gender, birth weight, head circumference, and Apgar scores, as well as the age at which the MRI scan was performed. At 12 months of age, the infants' cognitive abilities were assessed using the Bayley Scales of Infant Development (BSID), and their motor functions were evaluated using the Gross Motor Function Measure-88 (GMFM-88). Based on the results of the cognitive and motor function assessments, the infants were classified into two groups:Group B for those with a good prognosis and Group W for those with a poor prognosis.\u003c/p\u003e\u003cp\u003eThe clinical features, including gestational age at birth, mode of delivery, age, gender, birth weight, head circumference, and Apgar scores, were compared between Group B and Group W.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2 MR image acquisition\u003c/h2\u003e\u003cp\u003eThe MRI examinations were conducted when the infants' conditions had reached a basically stable state. We used the Siemens Skyra or Prisma 3.0T MRI scanner and an eight-channel head coil with the same MRI parameters for the examinations. The specific scanning sequences included axial T1WI, T2WI, FLAIR, and sagittal T1WI. The parameters for the sagittal T1WI were as follows: repetition time (TR) of 400 ms and echo time (TE) of 8.1 ms. For the axial T1WI, TR was 468 ms and TE was 11 ms; for the axial T2WI, TR was 4000 ms and TE was 101 ms; for the axial FLAIR, TR was 9000 ms and TE was 98 ms. The slice thickness was set to 4 mm with an inter-slice gap of 0.32 mm. Prior to segmentation and feature extraction, the MR images were preprocessed to eliminate any potential differences that might exist between images acquired from the two different scanners.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Establishment of clinical model\u003c/h2\u003e\u003cp\u003eThe clinical model was constructed by incorporating clinical features that significantly differed between Group B and Group W. Feature selection was performed using the Kruskal-Wallis test, and Support Vector Machine (SVM) was used as the classification tool. The hyperparameters of the SVM model were determined using Leave-One-Out Cross-Validation (LOOCV) on the training dataset.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4 Establishment of radiomics model\u003c/h2\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.5 ROI drawing and image analysis\u003c/h2\u003e\u003cp\u003eTwo radiologists with over ten years of experience delineated the region of interest (ROI) using ITK-SNAP software (available from \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.itksnap.org\u003c/span\u003e\u003cspan address=\"http://www.itksnap.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The ROIs were all placed in the basal ganglia region with the largest diameter, avoiding the venous sinuses, veins, and cerebrospinal fluid as much as possible. Please refer to Fig.\u0026nbsp;2.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e2.6 Radiomics feature-extraction and model evaluation\u003c/h2\u003e\u003cp\u003eWe selected 126 cases as the training dataset (67 in Group W and 59 in Group B). Additionally, we chose another 54 cases as the independent testing dataset (29 in Group W and 25 in Group B). To address the imbalance in the training dataset, we utilized the Synthetic Minority Oversampling Technique (SMOTE) to balance the samples between Group W and Group B. Subsequently, we normalized the feature matrix by subtracting the mean value from each feature vector and dividing it by its length.Due to the high dimensionality of the feature space, we compared the similarity between each pair of features. If the Pearson Correlation Coefficient (PCC) value between a pair of features exceeded 0.990, we removed one of them. Through this process, the dimensionality of the feature space was reduced, and each feature became independent of the others.Prior to model construction, we employed Recursive Feature Elimination (RFE) for feature selection. The goal of RFE is to select features based on a classifier by recursively considering smaller subsets of features. We chose the SVM as the classifier, given its effectiveness and robustness in model building. The kernel function has the capability to map features into a higher dimension to find the hyperplane that separates samples with different labels. In this case, we used a linear kernel function because it facilitates the interpretation of feature coefficients in the final model.Todetermine the model's hyperparameters (such as the number of features), we applied LOOCV in the training dataset. In the training date set with 126 samples, using the leave-one-out method, we train the model with 125 samples each time and test it with the remaining 1 sample. By performing cross-validation 126 times with mutually exclusive sets, we can obtain the cv training date set and thevalidation date set. The hyperparameters were set based on the model's performance on the validation dataset. All of the above processes were implemented using FeAture Explorer Pro (FAE, version 0.5.8) software on Python (version 3.7.6).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003e2.7 Establishment of combined model\u003c/h2\u003e\u003cp\u003eA combined model for predicting poor and good prognosis was developed using the same method as the established radiomics model, which integrates MR-based radiomic features and clinical factors.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e2.8 Statistical analysis\u003c/h2\u003e\u003cp\u003eAll statistical analyses were performed using SPSS 25.0 and R statistical software (version 3.6.1). For the continuous variables, the Kolmogorov-Smirnov and Levene tests were first performed to verify the normality and homogeneity of the samples. Variables conforming to the normal distribution were expressed as Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD. Group comparisons were made using the t-test. Categorical variables were expressed as frequencies and percentages, and comparisons between groups were made using the χ\u0026sup2; test. We plotted the Receiver Operating Characteristic (ROC) curve and comprehensively evaluated the model's performance in predicting the prognosis of HIE by calculating indicators including the area under the curve (AUC), accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). We used the bootstrap method to estimate the 95% confidence intervals. A two-sided P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 indicated a statistically significant difference.\u003c/p\u003e\u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.1 Patient characteristics\u003c/h2\u003e\u003cp\u003eA total of 180 full-term newborns were included in this study, with 84 in Group B and 96 in Group W. The detailed clinical parameters of patients in Group B and Group W are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The results in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e revealed that there were no statistically significant differences in age, gender, weight, head circumference, mode of delivery, or gestational age of the mothers between the two groups. The Apgar scores at 1 minute, 5 minutes, and 10 minutes were all higher in Group B compared to Group W, with P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 indicating a statistically significant difference.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eComparison of clinical data in Group B and Group W\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGroup B(N\u0026thinsp;=\u0026thinsp;84) No. of patient (%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eGroupW(N\u0026thinsp;=\u0026thinsp;96) No. of patient (%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eP value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge (days) \u003c/p\u003e\u003cp\u003e\u0026emsp;Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e6.75\u0026thinsp;\u0026plusmn;\u0026thinsp;2.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6.59\u0026thinsp;\u0026plusmn;\u0026thinsp;3.14\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.71\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGender\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.26\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e64(76.19)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e66(68.75)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFemale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e20(23.81)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e30(31.25)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDelivery method\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.32\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVaginal delivery\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e50(59.52)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e64(66.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCesarean section\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e34(40.48)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e32(33.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGestational week\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e39.06\u0026thinsp;\u0026plusmn;\u0026thinsp;1.28\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e39.54\u0026thinsp;\u0026plusmn;\u0026thinsp;1.29\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.06\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWeight(kg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.23\u0026thinsp;\u0026plusmn;\u0026thinsp;0.54\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.33\u0026thinsp;\u0026plusmn;\u0026thinsp;0.48\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.17\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHead circumference\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e33.48\u0026thinsp;\u0026plusmn;\u0026thinsp;1.25\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e33.82\u0026thinsp;\u0026plusmn;\u0026thinsp;1.31\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.08\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eApgar scores at 1minite\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e5.66\u0026thinsp;\u0026plusmn;\u0026thinsp;2.72\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.25\u0026thinsp;\u0026plusmn;\u0026thinsp;2.01\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;0.01\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eApgar scores at 5minite\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e7.88\u0026thinsp;\u0026plusmn;\u0026thinsp;2.02\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5.91\u0026thinsp;\u0026plusmn;\u0026thinsp;1.88\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;0.01\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eApgar scores at 10minite\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e9.06\u0026thinsp;\u0026plusmn;\u0026thinsp;1.29\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e7.20\u0026thinsp;\u0026plusmn;\u0026thinsp;1.56\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026lt;0.01\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Performance of clinical model\u003c/h2\u003e\u003cp\u003eThe clinical model showed that the Apgar score at 10 minutes was the most effective factor for differentiating Group B from Group W, with an AUC of 0.857 and an accuracy of 0.833 in the training dataset. At this point, the AUC, accuracy, sensitivity, and specificity of the clinical model achieved 0.737, 0.778, 0.965, and 0.56 in the testing dataset, respectively (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e3\u003c/span\u003e.A).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003e3.3 Performance of radiomics model\u003c/h2\u003e\u003cp\u003eA total of 4,227 radiomic features were extracted from T1WI, T2WI, and FLAIR images. These features included shape, first-order statistics, texture, and higher-order statistical features. After ranking these features, the nine best-performing significant radiomic features were selected. The AUC and accuracy could achieve 0.916 and 0.841 in the training dataset, respectively. At this point, the AUC, accuracy, sensitivity, and specificity of the radiomics model achieved 0.770, 0.778, 0.862, and 0.680 in the testing dataset(Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e3\u003c/span\u003e.B). The final filtered radiomic features and corresponding model coefficients are presented in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003e3.4 Performance of combined model\u003c/h2\u003e\u003cp\u003eThe combined model indicated that Apgar scores at 10 minutes, Apgar scores at 5 minutes, and seven best-performing significant radiomic features were independent predictors for differentiating Group W from Group B. The seven best-performing significant radiomic features were wavelet LLL glcm imc2, logarithm glcm inverse variance, wavelet HLL first-order mean, wavelet HLL first-order skewness, gradient glszm large area high gray level emphasis, logarithm gldm large dependence low gray level emphasis, and logarithm glrlm low gray level run emphasis. The AUC and accuracy could achieve 0.952 and 0.900 in the training dataset, respectively. At this point, the AUC, accuracy, sensitivity, and specificity of the combined model achieved 0.823, 0.796, 0.862, and 0.720 in the testing dataset (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e3\u003c/span\u003e.C).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003e3.5 Comparation of the assessment indicatorsof three models in the testing date set\u003c/h2\u003e\u003cp\u003eFor the testing dataset, the AUC of the combined model was larger than that of the clinical and radiomics models. The clinical model had the highest sensitivity (0.966) and NPV (0.933), and the combined model had the highest specificity (0.931) and PPV (0.925).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e༎The coefficients of features in the model\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFeatures\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCoef in model\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClinical model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eApgar scores at 10 minute\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e-5.667\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRadiomics model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFLAIR_wavelet-LLL_glcm_Imc2 \u003c/p\u003e\u003cp\u003eT1WI_log-sigma-1-mm-3D_gldm_LargeDependenceLowGrayLevelEmphasis \u003c/p\u003e\u003cp\u003eT1WI_logarithm_glcm_InverseVariance \u003c/p\u003e\u003cp\u003eT1WI_wavelet-HLL_firstorder_Mean \u003c/p\u003e\u003cp\u003eT1WI_wavelet-LLH_firstorder_Skewness \u003c/p\u003e\u003cp\u003eT2WI_gradient_glszm_LargeAreaHighGrayLevelEmphasis \u003c/p\u003e\u003cp\u003eT2WI_logarithm_gldm_LargeDependenceLowGrayLevelEmphasis \u003c/p\u003e\u003cp\u003eT2WI_logarithm_glrlm_LowGrayLevelRunEmphasis \u003c/p\u003e\u003cp\u003eT2WI_squareroot_gldm_LargeDependenceLowGrayLevelEmphasis\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e-1.111\u003c/p\u003e\u003cp\u003e-0.014\u003c/p\u003e\u003cp\u003e1.129\u003c/p\u003e\u003cp\u003e-2.172\u003c/p\u003e\u003cp\u003e2.743\u003c/p\u003e\u003cp\u003e-2.164\u003c/p\u003e\u003cp\u003e-0.825\u003c/p\u003e\u003cp\u003e-1.914\u003c/p\u003e\u003cp\u003e-0.461\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCombined model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eApgar scores at 10 minute\u003c/p\u003e\u003cp\u003eApgar scores at 5 minute\u003c/p\u003e\u003cp\u003eFLAIR_wavelet-LLL_glcm_Imc2 \u003c/p\u003e\u003cp\u003eT1WI_logarithm_glcm_InverseVariance \u003c/p\u003e\u003cp\u003eT1WI_wavelet-HLL_firstorder_Mean \u003c/p\u003e\u003cp\u003eT1WI_wavelet-LLH_firstorder_Skewness \u003c/p\u003e\u003cp\u003eT2WI_gradient_glszm_LargeAreaHighGrayLevelEmphasis \u003c/p\u003e\u003cp\u003eT2WI_logarithm_gldm_LargeDependenceLowGrayLevelEmphasis \u003c/p\u003e\u003cp\u003eT2WI_logarithm_glrlm_LowGrayLevelRunEmphasis\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e-2.566\u003c/p\u003e\u003cp\u003e-0.632\u003c/p\u003e\u003cp\u003e-0.875\u003c/p\u003e\u003cp\u003e0.935\u003c/p\u003e\u003cp\u003e\u0026ndash;1.787\u003c/p\u003e\u003cp\u003e2.205\u003c/p\u003e\u003cp\u003e-1.362\u003c/p\u003e\u003cp\u003e-0.768\u003c/p\u003e\u003cp\u003e-1.969\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003ePerformance of the clinical, radiomics, and combined models\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"8\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGroup\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAUC (95% CI)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCut-off\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eSensitivity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eSpecificity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003ePPV\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c8\"\u003e\u003cp\u003eNPV\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eTraining data set\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClinical model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.857(0.789\u0026ndash;0.925)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.602\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.833\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.881\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.780\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.820\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.852\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRadiomics model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.916(0.870\u0026ndash;0.962)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.349\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.841\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.925\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.746\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.805\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.898\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCombined model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.952(0.919\u0026ndash;0.985)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.650\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.900\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.866\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.932\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.936\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.860\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eCV training date set\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClinical model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.856(0.850\u0026ndash;0.862)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.573\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.828\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.881\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.776\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.800\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.867\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRadiomics model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.912(0.908\u0026ndash;0.917)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.517\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.833\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.822\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.845\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.841\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.826\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCombined model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.951(0.948\u0026ndash;0.953)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.642\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.891\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.851\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.931\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.925\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.862\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003evalidation data set\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClinical model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.843(0.765\u0026ndash;0.921)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.610\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.865\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.881\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.848\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.868\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.862\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRadiomics model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.867(0.804\u0026ndash;0.930)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.389\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.825\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.881\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.763\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.808\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.850\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCombined model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.931(0.890\u0026ndash;0.973)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.660\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.881\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.851\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.915\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.919\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.844\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003etesting data set\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClinical model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003e0.737(0.595\u0026ndash;0.879)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.602\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.778\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.966\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.560\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.718\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.933\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRadiomics model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003e0.770(0.641\u0026ndash;0.898)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.297\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.778\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.862\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.680\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.758\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.810\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCombined model\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003e0.823(0.705\u0026ndash;0.942)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.380\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.796\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.851\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.931\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.925\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.862\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThe development of accurate prognostic indicators has been a focus of research in the field of neonatal hypoxic-ischemic encephalopathy (HIE) in recent years. Currently known prognostic indicators include serum proteins (\u003cspan additionalcitationids=\"CR10\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e), neurophysiological examinations (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e), and neuroimaging studies. As a non-invasive technique, brain MRI plays a crucial role in precisely locating and assessing brain lesions. One of the biggest challenges in prediction by traditional anatomic MRI (such as T1- and T2-weighted images) is the low sensitivity. Children with HIE who experience severe neurological outcomes may lack apparent MRI findings during the neonatal period (the first 28 days after birth) (\u003cspan additionalcitationids=\"CR15\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). Radiomics, which extracts sub-visual but quantitative imaging features from radiological images (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e), can compensate for this limitation.\u003c/p\u003e\u003cp\u003eIn this study, there were no statistically significant differences in age, gender, weight, head circumference, mode of delivery, or gestational age of the mothers between Group B and Group W. However, the Apgar scores at 1 minute, 5 minutes, and 10 minutes were all higher in Group B compared to Group W, with statistically significant differences. The clinical model showed that the Apgar score at 10 minutes was the most effective factor for differentiating Group B from Group W, with an AUC of 0.843 and an accuracy of 0.865 on the validation dataset, and an AUC of 0.737 and an accuracy of 0.778 on the testing dataset, respectively. Some studies have shown that the risk of adverse outcomes increases when the Apgar score remains low at 10 minutes and beyond (\u003cspan additionalcitationids=\"CR20 CR21\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). Vivek V. Shukla's research suggests that a standalone 10-minute Apgar score of 0 does not accurately predict the risk of death or moderate to severe disability, but the predictive accuracy improves when the 10-minute Apgar score is combined with other risk variables available during resuscitation using a classification and regression tree analysis (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). In our study, we also found that the clinical model had the lowest AUC and accuracy among the three models. The combined model incorporating the Apgar score and radiomic features significantly improved the AUC and accuracy.\u003c/p\u003e\u003cp\u003eFor establishing the radiomics model, we selected the bilateral basal ganglia and thalamus as the ROI, primarily because these areas are more susceptible to damage in neonates with HIE (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). HIE can result in a series of biochemical changes, and patients undergo a complex pathophysiological process that includes hypoxia, ischemia, and subsequent reperfusion injury. Some scholars believe that during this process, when ischemic tissues regain blood supply (i.e., reperfusion), the resulting damage may be more severe than that caused by ischemia alone. This is because during reperfusion, the cells and tissues that were initially damaged by ischemia may suffer further injury, leading to a deterioration of the condition. Multiple studies utilizing ASL to assess changes in cerebral hemodynamics in neonates with HIE have consistently found early hyperperfusion of the basal ganglia and thalamus (\u003cspan additionalcitationids=\"CR26\" citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). Additionally, the thalamus, as one of the most metabolically active regions in the human body, has frequent cellular activity and high energy demands. Therefore, even mild HIE can trigger a certain degree of cellular apoptosis in the thalamus. In cases of mild HIE, conventional MRI examinations often fail to directly observe white matter damage and injuries to deep brain nuclei with the naked eye, but radiomics can analyze changes in their internal characteristics.\u003c/p\u003e\u003cp\u003eIn this study, we established a radiomic model for predicting the prognosis of HIE based on T1WI, T2WI, and FLAIR. This radiomic model was constructed using 9 non-zero coefficient features extracted from the original images, as well as images processed with Gaussian filters and wavelet transforms. These features include 2 first-order statistical features and 7 texture features. The most predictive feature was T1WI_wavelet-LLH_firstorder_Skewness. The radiomics results revealed the ability of radiomics to predict the prognosis of HIE in the training dataset (AUC\u0026thinsp;=\u0026thinsp;0.916), the validation dataset (AUC\u0026thinsp;=\u0026thinsp;0.867), and the testing dataset (AUC\u0026thinsp;=\u0026thinsp;0.77), thereby indicating that radiomics can predict the prognosis of HIE. Youwon Shin et al. (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e) enrolled 46 preterm neonates who underwent brain MRI near or at term-equivalent age, and neurodevelopment was assessed at a corrected age of 12 months. A prediction model for the binary classification of the psychomotor developmental index was developed. The AUCs of prediction models based on T1WI, T2WI, and both T1WI and T2WI were 0.925, 0.834, and 0.902, respectively. This finding was consistent with our results, indicating that it is feasible to establish a prognostic prediction model for neonatal brain injury using conventional MRI sequences.\u003c/p\u003e\u003cp\u003eThrough feature selection, this study identified 2 clinical features and 7 radiomic features to establish a combined model. The clinical features were the Apgar scores at 5 minutes and the Apgar scores at 10 minutes. Among the 7 radiomic features, 1 was obtained from FLAIR images, 3 were derived from T1WI images, and 3 were from T2WI images. The two most critical feature parameters in the combined model were the Apgar scores at 10 minutes and T1WI_wavelet-LLH_firstorder_Skewness. The combined model showed a great ability to predict the prognosis of HIE, and the highest AUCs for the training dataset, validation dataset, and testing dataset were 0.952, 0.931, and 0.823, respectively, which were higher than those of the radiomics model (training dataset: AUC\u0026thinsp;=\u0026thinsp;0.916; validation dataset: AUC\u0026thinsp;=\u0026thinsp;0.867; testing dataset: AUC\u0026thinsp;=\u0026thinsp;0.770) and the clinical model (training dataset: AUC\u0026thinsp;=\u0026thinsp;0.857; validation dataset: AUC\u0026thinsp;=\u0026thinsp;0.843; testing dataset: AUC\u0026thinsp;=\u0026thinsp;0.737). The clinical model had the highest sensitivity (0.966) and NPV (0.933), while the combined model had the highest specificity (0.931) and PPV (0.925). This means that the clinical model was better at evaluating children with poor prognosis, while the combined model was better at evaluating children with good prognosis.\u003c/p\u003e\u003cp\u003eThere were still some limitations to this study. Firstly, it was a single-center retrospective analysis, lacking multi-center validation of the model with external data. Future research will require a larger sample size to evaluate the model. Secondly, in this study, the image segmentation was conducted through manual delineation, which was time-consuming and inevitably subject to human error. We plan to explore semi-automatic or deep learning-based automatic segmentation methods to enhance the objectivity of our research methodology. Thirdly, we utilized conventional MRI sequences; therefore, in the future, we will incorporate more sensitive MRI sequences such as Diffusion Weighted Imaging (DWI) and Susceptibility Weighted Imaging (SWI) to improve the efficacy of the prognostic prediction model.\u003c/p\u003e\u003cp\u003eIn conclusion, the clinical model and radiomics model constructed in this study performed well in predicting the prognosis of HIE. Based on this, the combined model, which integrates clinical factors and MR radiomics, further improved the accuracy of predicting the prognosis of HIE. This model is objective, non-invasive, and reproducible.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003e5.1 Ethics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study constitutes a non-interventional investigation that does not involve human clinical trials or experimental interventions. In accordance with international research ethics guidelines, formal trial registration is not applicable. The research protocol adheres to the ethical principles outlined in the Declaration of Helsinki for non-clinical studies, particularly regarding data integrity and analytical rigor.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.2 Consent to Participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWritten informed consent was obtained from all individual participants (or their legal guardians in cases involving minors) prior to their inclusion in the study. Participants were informed about the research purpose, methodology, potential risks, and benefits, and their right to withdraw at any stage without consequences.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.3 Consent to Publish\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor studies involving identifiable personal data (e.g., clinical images, case details), written consent for publication was secured from all participants or their legal representatives. Anonymity was maintained for non-essential identifying information.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.4 Competing Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing financial or non-financial interests relevant to this work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.5 Author Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGuarantor of integrity of the entire study: Ke Jin and Ting Yi.\u003c/p\u003e\n\u003cp\u003eStudy concepts and design: Jing Tang, Ting Yi and Ke Jin.\u003c/p\u003e\n\u003cp\u003eLiterature research: Jing Tang and Ting Yi.\u003c/p\u003e\n\u003cp\u003eClinical studies: Jing Tang and Siping He;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eExperimental studies/data analysis: Ting Yi, Jing Tang and Yonghua Xiang.\u003c/p\u003e\n\u003cp\u003eStatistical analysis: Ting Yia and Yuqing Liu.\u003c/p\u003e\n\u003cp\u003eManuscript preparation: Jing Tang.\u003c/p\u003e\n\u003cp\u003eManuscript editing: Jing Tang and Ting Yi.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.6 Data Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data that support the findings of this study are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e6. Acknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Hunan Provincial Science and Technology Innovation Plan (Natural Science Foundation of Hunan Province, Clinical Medical Technology Innovation Guidance Project, Grant No. 2021SK50511).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eParmentier CEJ, de Vries LS, Groenendaal F. Magnetic Resonance Imaging in (Near-)Term Infants with Hypoxic-Ischemic Encephalopathy. Diagnostics(Basel). 2022;12(3):645.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAllen KA, Brandon DH. Hypoxic Ischemic Encephalopathy: Pathophysiology and Experimental Treatments. Newborn Infant Nurs Rev. 2011;11(3):125\u0026ndash;33.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKurinczuk JJ, White-Koning M, Badawi N. Epidemiology of neonatal encephalopathy and hypoxicischaemic encephalopathy. Early Hum Dev. 2010;86(6):329\u0026ndash;38.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChau V, Poskitt KJ, Dunham CP, et al. Magnetic resonance imaging in the encephalopathic term newborn. Curr Pediatr Rev. 2014;10(1):28\u0026ndash;36.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOnda K, Chavez-Valdez R. GrahamEM,etal.Quantification of diffusion MRI for prognostic prediction of neonatal hypoxic-ischemic encephalopathy. Dev Neurosci. 2024;46(1):55\u0026ndash;68.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYildiz EP, EkiciB,Tatil B. Neonatal hypoxic ischemic encephalopathy:an update on disease pathogenesis and treatment. Expert Rev Neurother. 2017;17(5):449\u0026ndash;59.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRobertson CM, Perlman M. Follow-up of the term infant after hypoxic-ischemic encephalopathy. Paediatrics\u0026amp; Child Health. 2006;11(5):278\u0026ndash;82.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRizzo S, Botta F, Raimondi S, et al. Radiomics: the facts and the challenges of image analysis. EurRadiol Exp. 2018;2(1):36.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDouglas-Escobar M, Weiss MD. Biomarkers of hypoxic-ischemic encephalopathy in newborns. Front Neurol. 2012;3:144.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEinspieler C, Prechtl HF, Ferrari F, et al. The qualitative assessment of general movements in preterm, term and young infants\u0026mdash;review of the methodology. Early Hum Dev. 1997;50(1):47\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHaataja L, Mercuri E, Regev R, et al. Optimality score for the neurologic examination of the infant at 12 and 18 months of age. J Pediatr. 1999;135(2):153\u0026ndash;61.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLori S, Bertini G, Molesti E, et al. The prognostic role of evoked potentials in neonatal hypoxic-ischemic insult. J Maternal-Fetal Neonatal Med. 2011;24(sup1):69\u0026ndash;71.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eStewart AM, Chapman KE. Prognostication in Pediatrics. In: Husain AM, Sinha SR, editors. Continuous EEG Monitoring: Principles and Practice. Cham: Springer International Publishing; 2017. pp. 465\u0026ndash;81.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMartinez-Biarge M, Diez-Sebastian J, Kapellou O, et al. Predicting motor outcome and death in term hypoxic-ischemic encephalopathy. Neurology. 2011;76(24):2055\u0026ndash;61.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003evan Laerhoven H, de Haan TR, Offringa M, et al. Prognostic tests in term neonates with hypoxic-ischemic encephalopathy: a systematic review. Pediatrics. 2013;131(1):88\u0026ndash;98.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eUtsunomiya H. Diffusion MRI abnormalities in pediatric neurological disorders. Brain Develop. 2011;33(3):235\u0026ndash;42.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eParmar C, Leijenaar RT, Grossmann P, et al. Radiomic feature clusters and prognostic signatures specific for lung and head \u0026amp; neck cancer. Sci Rep. 2015;5:11044.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGillies RJ, Kinahan PE, Hricak H. Radiomics: images are more than pictures they are data. Radiology. 2016;278:563\u0026ndash;77.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAmerican Academy of Pediatrics, Committee on Fetus and Newborn. American College of Obstetricians and Gynecologists and Committee on Obstetric Practice. The Apgar score. Pediatrics. 2006;117:1444\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCasalaz DM, Marlow N, Speidel BD. Outcome of resuscitation following unexpected apparent stillbirth. Arch Dis Child Fetal Neonatal Ed. 1998;78:F112\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNatarajan G, Shankaran S, LaptookAR, et al. Apgar scores at 10 min and outcomes at 6\u0026ndash;7 years following hypoxic-ischaemic encephalopathy. Arch Dis Child Fetal Neonatal Ed. 2013;98(6):F473\u0026ndash;479.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePavel AM, O'Toole JM, Proietti J, et al. Machine learning for the early prediction of infants with electrographic seizures in neonatal hypoxic-ischemic encephalopathy. Epilepsia. 2023;64(2):456\u0026ndash;68.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShukla VV, Bann CM, Ramani M, et al. Predictive Ability of 10-Minute Apgar Scores for Mortality and Neurodevelopmental Disability. Pediatrics. 2022;149(4):e2021054992.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWu TW. TamraziB,HsuKH,et al. Cerebral lactate concentration inneonatal hypoxic-ischemic encephalopathy:in relation to time,characteristic of injury,and serum lactate concentration. Front Neurol. 2018;9:293.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang J, Li J, Yin X, et al. Cerebral hemodynamics of hypoxic-ischemic encephalopathy neonates at different ages detected by arterial spin labeling imaging. Clin HemorheolMicrocirc. 2022;81(4):271\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang J, Li J, Yin X, et al. The Value of Arterial Spin Labeling Imaging in the Classification and Prognostic Evaluation of Neonatal Hypoxic-ischemic Encephalopathy. Curr Neurovasc Res. 2021;18(3):307\u0026ndash;13.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eProisy M, Corouge I, LegouhyA, et al. Changes in brain perfusion in successive arterial spin labeling MRI scans in neonates with hypoxic-ischemic encephalopathy. Neuroimage Clin. 2019;24:101939.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYouwonShin. Brain MRI radiomics analysis may predict poor psychomotor outcome in preterm neonates. EurRadiol. 2021;31(8):6147\u0026ndash;55.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"hypoxic-ischemic encephalopathy, radiomics, Magnetic Resonance, predict prognosis","lastPublishedDoi":"10.21203/rs.3.rs-6097732/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6097732/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003ePurpose\u003c/h2\u003e\u003cp\u003eThe purpose of this study was to establish a combined model based on T1WI, T2WI, FLAIR images and clinical parameters to predict the prognosis of hypoxic-ischemic encephalopathy (HIE) in full-term newborns.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eBased on the results of cognitive scores and motor function scores at 12 months post-birth, the patients were classified into two groups: Group B for those with good prognosis (n\u0026thinsp;=\u0026thinsp;84) and Group W for those with poor prognosis (n\u0026thinsp;=\u0026thinsp;96). A total of 180 patients were retrospectively evaluated and assigned to either the training data set (n\u0026thinsp;=\u0026thinsp;126) or the testing data set (n\u0026thinsp;=\u0026thinsp;54). The clinical characteristics of both groups were compared first. Then, a clinical model, a radiomics model, and a combined model were developed. Finally, we evaluated the performance of the three constructed models using the receiver operating characteristic (ROC) curve and the area under the curve (AUC).\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eThe Apgar scores at 1 minute, 5 minutes, and 10 minutes were all higher in Group B compared to Group W, with P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 indicating a statistically significant difference. The clinical model showed that the Apgar score at 10 minutes was the most effective factor, with an AUC of 0.857 in the training data set and an AUC of 0.737 in the testing data set. For the radiomics model, 9 radiomics features were found to be significantly related to predicting the prognosis of HIE, with AUCs of 0.916 and 0.770 in the training and testing data sets, respectively. For the combined model, 7 radiomics features, Apgar scores at 5 minutes, and Apgar scores at 10 minutes were independent predictors for predicting the prognosis of HIE, with AUCs of 0.952 and 0.823 in the training and testing data sets, respectively. The combined model demonstrates better performance than both the clinical and radiomics models.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e\u003cp\u003eThe combined model, which incorporates MR-based radiomics signatures, and clinical factors, is effective in predicting the prognosis of HIE.\u003c/p\u003e","manuscriptTitle":"Combined model of radiomics and clinical features for predicting prognosis of term neonatal hypoxic-ischemic encephalopathy after one year: an exploratory study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-14 05:13:09","doi":"10.21203/rs.3.rs-6097732/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-09-30T06:13:50+00:00","index":"","fulltext":""},{"type":"reviewerAgreed","content":"248550165147111377857531483945585072446","date":"2025-09-30T01:58:55+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-30T01:43:12+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-29T14:42:54+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"195226278783417556108489899158468473681","date":"2025-09-25T00:24:20+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"81323317608851656510254181357524906522","date":"2025-09-24T09:25:01+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-24T02:59:20+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"94991654844477012150968594801850827026","date":"2025-09-21T09:10:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"177820132182110181505437905064495507258","date":"2025-07-16T23:13:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"107238322178444589501361475184320702501","date":"2025-07-16T11:02:49+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-06T05:03:36+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-04-05T02:36:15+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-04-05T02:35:09+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Imaging","date":"2025-02-24T14:09:04+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ad71f39e-6811-4777-8324-2aaf34660e2b","owner":[],"postedDate":"July 14th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-01-05T16:00:40+00:00","versionOfRecord":{"articleIdentity":"rs-6097732","link":"https://doi.org/10.1186/s12880-025-02020-5","journal":{"identity":"bmc-medical-imaging","isVorOnly":false,"title":"BMC Medical Imaging"},"publishedOn":"2025-12-29 15:57:47","publishedOnDateReadable":"December 29th, 2025"},"versionCreatedAt":"2025-07-14 05:13:09","video":"","vorDoi":"10.1186/s12880-025-02020-5","vorDoiUrl":"https://doi.org/10.1186/s12880-025-02020-5","workflowStages":[]},"version":"v1","identity":"rs-6097732","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6097732","identity":"rs-6097732","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}