摘要目的 实现Tanner-Whitehouse(简称TW2)骨龄算法网络化,提高放射科医师阅读骨龄X线片的效率.方法 采用TW2骨龄算法,用Java脚本执行语言编写骨龄算法并放在服务器中,通过HTTP服务,客户端可以访问集成有算法的可视化Web页面,在可视化图形界面上对照20块参考骨来选择X线骨龄片中相应骨的不同成熟阶段(每块参考骨顺应TW2法,划分为8个成熟阶段),同时允许医师再输入其他具体数据,如出生日期、性别等,然后通过计算机计算出骨龄.对50份2～18岁儿童手腕部骨龄片,共1000块骨进行网络TW2骨龄箅法和人工骨龄评分法所用时间和一致性比较进行U检验.结果 骨龄算法网络化使用简单,具有友好的界面,准确性和可重复性也优于人工判读,1000块骨通过前者进行骨成熟度分级一致的骨数为838块,占总数的83.8%,相差1级(+1和-1)的有149块,占总数的14.9%,相差2级(+2和-2)的有13块,占总数的1.3%.通过后者进行骨成熟度分级1致的骨数为657块,占总数的65.7%,相差1级(+1和-1)的有272块,占总数的27.2%,相差2级(+2和-2)的有71块,占总数的7.1%.两者一致率的差异具有统计学意义(U=9.31595,P<0.01).两者骨龄评估平均所花费的时间分别为3～5 min和15～20 min,前者明显提高了放射科医师阅读X线骨龄片的效率.结论 网络TW2骨龄算法可以使骨龄算法快速简便地实现共享.

abstractsObjective Improve the efficiency of radiology doctors on reading bone age films by sharing Tanner-Whitehouse(TW2)bone algorithm on the network. Methods The bone age algorithm web was Programmed with a Java script implementation of Tanner-Whitehouse Method and putted on a Web server based on HTTP service. The program allows to select a stage (from a set of 8 stages) for every bone (from a set of 20 bones), and also allows doctors to input some specific data such as birthday, sex. Based on TW2 reported values, selected and input data, the program computes the bone age. We assessed the bone ages on 50 left hand and wrist X-ray films of Chinese children aged 2-18 (1000 bones totally) with computer-aided method and manual method. The grading agreement of bone development and the time spent for bone age assessment were compared (U test) between the two methods. Results Computer-aided method is easy to use, better than manual method in accuracy of bone development grading, and it also has a friendly interface. For the 1000 bones assessed by TW2, the rate of the same maturity classification was 83.8%(838/1000), the rate of one-level maturity difference (+ 1 and - 1) was 14. 9% (149/1000), the rate of two-level maturity difference (+2 and -2) was 1.3% (13/1000). For the 1000 bones assessed by manual method, the rate of the same maturity classification was 65.7% (657/1000), the rate of one-level maturity difference (+ 1 and -1) was 27. 2% (272/1000), the rate of two-level maturity difference (+ 2 and -2) was 7.1% (71/1000). TW2 bone algorithm was significantly better than manual method (U =9.31595 ,P <0.01). The average time of assessing bone age by the two methods was 3-5 min and 15-20 min, and the TW2 method saved time for radiologists doctor. Conclusion Sharing TW2 bone age algorithm through the network can be quick and easy.