Measurement of Imaging Indexes of Distal Radius based on deep Learning Key Point Detection.

The measurement of imaging indexes of distal radius is the basic work of diagnosis and recovery evaluation. Due to the problems of large morphological differences and insufficient clarity in radial images, the accurate measurement of indexes relies heavily on the clinical experience of doctors, with defects such as low accuracy and poor consistency. In order to solve the above problems, this paper proposes an intelligent technology to achieve accurate measurement of imaging indicators of the distal radius. Firstly, 12 key points needed for distal imaging indexes were determined: radius left vertex, radius right vertex, radius left neck point, radius right neck point, radius left lower point, radius right lower point, ulna left vertex, ulna right vertex, ulna left neck point, right neck point, ulna left lower point, ulna right lower point. Then, the deep convolutional neural network was used to estimate the geometric transformation features and regress the thermal map of the feature points, so as to achieve accurate positioning of the 12 feature points. Finally, each imaging index was calculated accurately by the preset computational method. The experimental results show that the proposed measurement method has advantages of high accuracy, good consistency and fast speed. The test accuracy is higher than the manual measurement results of inexperienced doctors, and the network model has good robustness and generalization ability. This method has practical value and can be applied in the field of medical imaging to help doctors improve the efficiency of diagnosis.