Bone age detection method based on multi-modal adversarial training

Abstract

The invention provides a bone age detection method based on multi-modal adversarial training. The bone age detection method comprises the following steps: constructing a bone age prediction data set;constructing a bone age detection model based on multi-modal adversarial training for training; in the prediction stage, reserving only a discriminator of the bone age detection model, adding softmaxto the last layer, and loading and training the optimal discrimination model weight for result prediction. According to the bone age detection method based on multi-modal adversarial training, the method comprises the steps of: adopting the adversarial training mode to improve the accuracy of model prediction; carrying out multi-modal bone age detection by utilizing the medical images and the textmedical records; performing training by using the Chinese teenager X-ray data set to obtain a bone age recognition result conforming to Chinese teenagers; achieving bone age recognition of multi-modal data, wherein a recognition result is more accurate in combination with text information.

Description

technical field [0001] The present invention relates to the technical field of bone age detection, in particular to a bone age detection method based on multimodal confrontation training. Background technique [0002] As an important indicator of growth and development, bone age analysis plays an important role in the fields of medicine, sports and judicial identification. The degree of skeletal calcification in children is a determinant of bone age. Bone age can more accurately reflect the developmental level of each age stage in the process of human growth. Measuring a child's bone age is usually determined by a radiologist comparing x-rays of the child's hand to a standard state for their age. [0003] Adolescent bone age assessment plays an important role in the diagnosis of pediatric endocrine problems and growth disorders in children. It is often used to screen adolescents for endocrine disorder, growth and development delay, congenital adrenal hyperplasia and other ...

AI Technical Summary

Problems solved by technology

The traditional bone age detection method only uses convolutional neural network (CNN), and the network scalability is poor when processing images obtained by different detectors and different machines

Images