Thoracoscopic resection of pulmonary osteosarcoma metastases guided by artificial intelligence: A case series

Abstract

Background: Pulmonary metastasis is one of the main causes of death in patients with osteosarcoma. Timely and accurate identification of osteosarcoma micro-metastases and appropriate surgical interventions may improve the long-term survival rate of patients with osteosarcoma. Artificial-intelligence-assisted diagnosis based on three-dimensional convolutional neural networks is currently a cutting-edge technology. The basic components comprise convolution, activation, and pooling. The network weight is adjusted iteratively through the training data, and the backpropagation algorithm and nonlinear activation function are used to improve image accuracy. Cases presentation: Two adolescent patients with osteosarcoma were followed up regularly with thin-slice computed tomography (CT) after surgery to observe the presence of pulmonary metastatic nodules of osteosarcoma. The radiologist missed the diagnosis of small lung nodules at the first evaluation of the chest CT images of two patients but successfully identified them using artificial intelligence-assisted diagnostic technology. Pulmonary metastatic nodules of osteosarcoma were confirmed with thoracic surgical intervention. The two patients properly recovered after the operation and are still under follow-up. Conclusion: Medical artificial intelligence-assisted diagnostic technology can help identify small pulmonary metastatic nodules in patients with osteosarcoma, reduce the frequency of missed nodules, and guide their surgical resection.

Keywords

• Osteosarcoma
• Pulmonary nodules
• Artificial intelligence
• Deep convolutional neural network
• Case series