<scp>SCLe</scp> ‐Net: Siamese Convolutional Le‐Network for Hand Bone Age Estimation Using X‐Ray Image

ABSTRACT Hand bone age estimation is used to assess skeletal maturity and predict growth patterns by analyzing the bones in the hand. The prediction of bone age manually is subjective, and it varies among clinicians because of the variation in the analysis of an image. Various models have been presented, but they did not have the ability to handle large databases and did not offer consistent results. Hence, a new method termed Siamese Convolutional Le‐Net (SCLe‐Net) has been devised to estimate the hand bone age. First, the x‐ray image is considered as input and preprocessed by adaptive bilateral filter (ABF), which is exploited for eradicating the noise. Next, ROI extraction is performed to mine the required region from the image. Following this, the image is augmented by applying color and position augmentations. After this, the Local Gradient XoR Pattern (LGxP) with moment invariant feature is mined. Finally, hand bone age is estimated using SCLe‐Net, which integrates LeNet and Siamese Convolutional Neural Network (SCNN). Further, the experimental outcomes illustrate that the SCLe‐Net computed a maximum negative predictive value (NPV), accuracy, and positive predictive value (PPV) of 92%, 95.9%, and 93.9%.