Mediastinal object identification method, system and device

• Main Authors: GAO ZEBIN, NIE FANGXING, DU QIANG, GUO YUCHEN
• Format: Patent
• Language: Chinese; English
• Published: 20.07.2021
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses a mediastinal object identification method, system and device. The method comprises the following steps of S1, preprocessing a CT image of a mediastinal tumor, S2, taking a plurality of continuous slices from the preprocessed CT image, and processing each slice into a matrix with a plurality of window widths and window levels, and S3, inputting a plurality of slices processed into a matrix with a plurality of window widths and window levels into 2.5 DUNet of a two-stage self-attention mechanism, performing grouping convolution on the plurality of slices, and inputting the slices into a two-stage self-attention mechanism module for fusion to obtain an identification result. According to the method, deep learning is used, an attention mechanism is combined, and the advantages of 2D and 3D are utilized at the same time, so that the model precision is high, and the reasoning speed is high; in addition, due to the fact that the database is large in scale, the generalization performance can