Bone Age Prediction Using Edge Detection and ResNet50

This project enters into the field of medical imaging, particularly giving an address to integral task of predicting children’s bone age by exploring and examining their hand X-ray images. By utilizing the dataset of RSNA 2017 challenge, this research focuses on discovering the powerful models using the new and modern techniques like ResNet50 transfer learning and hyperparameter tuning. The RSNA 2017 dataset comprises a combination of scanned and digital version of X-ray images along with an extensive CSV file that has essential information like age (the variable to be predicted) and gender, which act as foundation for this study. On the far side of the typical procedures and approaches, this project amalgamates advanced modes and methods for preprocessing and image enhancement. Putting into practice of algorithms like edge detection, the hand X-ray images go through detailed cleaning to pull out the exact hand part and insights which is fundamental for the clear-cut age prediction. Additionally, the image augmentation techniques implementation reinforces the model’s flexibility by uncovering it to manifold variations in the dataset, which accumulates its potentiality to popularize results in more authentic estimations. This study helps in the field of healthcare, espousing creative procedures to propel extremities of bone age prediction from X-ray images. The collaboration of RSNA Radiology Informatics Committee with renowned proportions in the province, instrumented and supervise this provocation, discovering a standard point for evolution in pediatric radiology. By employing ResNet50 transfer learning with meticulous hyperparameter tuning, this project not only aids to advance the model that predicts bone age but also takes into consideration of insertion of gender information as an external, which has the major exert influence on estimation task. The envisioned result of this research grasps remarkable assurance for the upcoming generation of radiology and medical image vetting. Precise age estimation models not only bestow to improved pediatric care of patients but also smooth the path of enhanced diagnostic precision, undoubtedly amending practices of radiology. Eventually, this study endeavors to make consequential benefaction to healthcare, which benefits both suppliers of healthcare and patients by strengthening sharper and more regularized methods for holding X-ray images and therefore enhancing technology of medical imaging.