Automated Fetal Brain Growth Assessment Using AI For Early Diagnosis of Neurological Disorders

Resumen

Fetal brain development is a sensitive indicator of neurological well-being, and early detection of malformation critically impacts perinatal outcome. Conventional fetal neuroimaging modalities like ultrasound and MRI are inhomogeneous, interpreter-dependent, and time-consuming. A new Automated Fetal Brain Growth Assessment (AFBGA) system using Artificial Intelligence (AI) and deep machine learning algorithms is introduced to facilitate rapid and homogeneous fetal brain growth assessment. Spatiotemporal Neural Growth Network is thought to be a new architecture of whole-fetal brain growth pattern modeling and whole-fetal brain segmentation. The model receives serial ultrasound and MRI scans through morphological and structural brain development characteristics of fetal brain development. To extract features accurately, a Neuro-Optimized Feature Extraction Algorithm (NOFEA) consists of an adaptive selection mechanism, which automatically selects informative features from multi-scale neuroimaging data. It eliminates redundant information and enhances interpretability, which helps in enhancing brain development measurement accuracy. Another Growth Trajectory Prediction Network (GTP-Net) is created that integrates Long Short-Term Memory (LSTM) and a Neuro-Adaptive Variational Model (NAVM) to predict fetal brain growth variation profiles with high temporal resolution. Blockchain-secured federated learning environment maintains data privacy and supports collaborative training among multi-institutional fetal image data sets to achieve more generalizability across heterogeneous populations. Experimental validation on large-scale fetal neuroimaging datasets confirms that the proposed AFBGA framework provides greater accuracy, sensitivity, and specificity compared to conventional evaluation protocols. The system can perform real-time risk stratification of microcephaly, ventriculomegaly, and neural tube defects, allowing timely intervention. This research contributes to AI-driven fetal medicine by providing an automated, scalable, and privacy-preserving solution for early neurological disorder diagnosis. Future work will focus on integrating multi-modal imaging fusion and deploying AI-assisted fetal brain assessment tools in clinical practice.