Weakly supervised deep learning automatic bone age evaluation method

Abstract

The invention relates to a weakly supervised deep learning automatic bone age evaluation method which comprises the following steps: marking bone age physiological anatomy key points for each X-ray film by taking a hand X-ray film as image data; constructing and training a neural network model, wherein the neural network model comprises a Faster R-CNN model and a U-net model, the Faster R-CNN model extracts a palm skeleton region and joint regions, and the joint regions comprise phalanx joints and elbow joints of fingers and comprise 16 joint regions in total; the U-net model extracts hand features of a palm skeleton area and key point features of a finger area; and summing the scoring result prediction values of all the key points, and taking a sum value as a bone age prediction value. The method has the advantages that the doctor experience is gathered in the data set to form the doctor group experience to label the data set, the X-ray film (image data) used for prediction each timecan be used as a new element to be supplemented into the data set, and the data set can be continuously expanded and accumulated.

Description

technical field [0001] The invention relates to the field of medical artificial intelligence, in particular to a method for automatically evaluating bone age in a deep learning manner. Background technique [0002] Bone age is a reliable indicator of physical maturity. It is a specific age sign and characteristic of children's bone age development, reflects the biological age of the body, and can accurately evaluate the maturity of the body. Especially in the clinical diagnosis and treatment of children with growth and development disorders, bone age assessment is very necessary. Bone age measurement can evaluate the bone development of boys and girls, and can early detect genetic and endocrine diseases with abnormal height, which plays an important role in treatment and monitoring. [0003] Bone age, unlike chronological age, tends to vary by gender and race. There are generally two widely used methods for bone age assessment using hand radiographs in current clinical pra...

AI Technical Summary

Problems solved by technology

Such a model is more bloated, runs slower, and cannot get scores for each key point

[0004] The bone age prediction methods mentioned above all rely on the doctor's personal experience to estimate the bone age. It is difficult for doctors to learn from and integrate with each other's experience, and it is difficult to quantify personal experience.

Images