Quantitative Insights Into Pediatric Neurodevelopment: Optimizing Magnetic Resonance Fingerprinting for Enhanced Imaging and Characterization of Developmental Disorders

Pediatric neurodevelopmental disorders are difficult and pose challenges in both diagnosis as well as treatment. Therefore, it becomes necessary to use advanced imaging methodologies to ensure precise evaluation. Magnetic Resonance Fingerprinting (MRF) is one of the upcoming image modality that can help to diagnose pediatric neurodevelopment disorders. In this paper, we have proposed a novel method of LSTM and Ridge Regression. This method is implemented into the processed tabular data extracted from MRF images, where the first step is to convert MRF images into tabular layout. Each Individual row encapsulates a distinct MRF sample characterized by multiple features. Subsequently, are used to acquire representations from the tabular MRF data which are then channeled into LSTM network to obtain a cutting-edge level of performance is attained in the classification of neurodevelopmental disorders based on MRF data. The results of our experiments demonstrate that our proposed model reached an accuracy rate of 83.1 percent in the case of developmental disorders, surpassing CNN (78.2 percent), RNN (75.6 percent), and MLP (80.3 percent). This increased level of accuracy has the capacity to assist healthcare practitioners in the prompt identification and treatment of patients, thereby improving overall clinical outcomes. In essence, this paper explores the promise of LSTM and Ridge Regression models in enriching the analysis of MRF data.