Dual Stream Enhanced Lung Cancer Growth Evolution Predictive Network Under Time Series

[Purposes] In this paper, a dual stream enhanced lung cancer growth evolution predictive network under time series (DSGNet) is proposed. [Methods] With full use of the advantages of CNN and Transformer, PSGNet can be used to extract the static features of tumors through CNN-based branches, and strengthen the extracted feature representations in a multi-scale manner. The Transformer-based branch helps obtain sequential dependencies between tumor sequence images, with its core component being the multi-head self-attention layer proposed in this paper. This branch maps lesion sequence images into a feature map sequence and then inputs the sequence into a deep network with multi-head self-attention, from which the complete inter-tumor growth relationship is extracted. [Findings] The proposed algorithm is evaluated on the lung cancer NLST dataset and a dataset from cooperative hospitals. The experimental results show that DSGNet achieves a precision of 92.45% and a Dice coefficient of 82.78% in predicting tumor growth. Compared with other tumor prediction algorithms, DSGNet proposed in this paper has been improved to a certain extent in all aspects and has been proven to be applicable to clinical research in many ways.