A Novel Framework for Forensic Bone Age Estimation Using CNN Architecture and Multimodal Fusion

This study introduces an approach to forensic bone age estimation by employing a multimodal fusion strategy within a Convolutional Neural Network (CNN) architecture. The challenge of low-quality images in the dataset is addressed through the incorporation of the Contrast Limited Adaptive Histogram Equalization (CLAHE) preprocessing technique, effectively enhancing image quality and bolstering model accuracy. Furthermore, efficient net transfer learning techniques are implemented for additional refinement. The validation of the proposed method, assessed using Mean Squared Error (MSE) and the coefficient of determination $R^{2}$ score, demonstrates compelling results with an MSE value of 18.21 and an $R^{2}$ score of 0.59. These outcomes surpass those of existing methods, highlighting the superior accuracy and robustness of our framework. This marks a significant advancement in forensic bone age estimation.