European Radiology: Deep learning makes it possible to quickly classify bone tumors on plain radiographs!

Bone tumors are the third leading cause of death in cancer patients under the age of 20 .

Bone tumors are the third leading cause of death in cancer patients under the age of 20 .

Data-driven machine learning and deep learning (DL) techniques have been extensively studied and have been used in many clinical fields at this stage

Recently, a study published in the journal European Radiology evaluated the ability of DL in differentiating benign, malignant and intermediate bone tumors , and compared DL with the diagnostic level of radiologists for rapid and accurate clinical evaluation of bone tumors.

This retrospective study collected data on pathologically confirmed bone tumors between 2012 and 2019

A total of 982 radiographs from 643 patients (median age, 21 years; interquartile range, 12-38 years; 244 women) were included in the final evaluation

Figure Receiver operating characteristic curve and area of ​​curve (ROC-AUC) for radiologists, fusion models and radiologists on the test set .

Figure Receiver operating characteristic curve and area of ​​curve (ROC-AUC) for radiologists, fusion models and radiologists on the test set .

In this study , a model incorporating radiological and clinical information was developed for the identification of benign , intermediate, and malignant tumors .

Original source :

Original source :

Renyi Liu , Derun Pan , Yuan Xu , et al.

Renyi Liu Derun Pan Yuan Xu ,et al.
.
10.
1007/s00330-021-08195-zLeave

a message here