Computer implemented method and system for automatically detecting target objects in 3D image

Abstract

The invention relates to a computer implemented method and system for automatically detecting target objects in a 3D image. The method includes receiving the 3D image captured by an imaging device, and further includes detecting, by a processor, a plurality of bounds including a target object by utilizing a 3D learning network. The learning network is trained to generate a set of feature mappingshaving different scales based on a 3D image. The method further includes determining, by the processor, a set of parameters recognizing each detected bound by utilizing the 3D learning network, and positioning, by the processor, the target object based on the set of parameters. The method can rapidly, accurately and automatically detect the target object from the 3D image by utilizing the 3D learning network.

Description

[0001] Cross References to Related Applications [0002] This application claims priority to US Provisional Application No. 62 / 542,890, filed August 9, 2017, the entire contents of which are incorporated herein by reference. technical field [0003] The present disclosure generally relates to image processing and analysis. More specifically, the present disclosure relates to methods and systems for automatic localization and detection of target objects from 3D images. Background technique [0004] The accuracy of diagnosis and the effect of treatment depend on the quality of medical image analysis, especially the detection of target objects (such as organs, tissues, target sites, etc.). Compared with conventional two-dimensional imaging, volume (3D) imaging, such as volume CT, can capture more valuable medical information, thereby facilitating more accurate diagnosis. However, objects of interest are usually detected by experienced medical personnel, such as radiologists, ...

AI Technical Summary

Problems solved by technology

[0006] Although some basic machine learning methods are introduced for detection, these methods usually define features artificially and thus have low detection accuracy

Moreover, such machine learning is usually limited to 2D image learning, but cannot directly detect target objects in 3D images due to the lack of 3D spatial information and the considerable computing resources required for 3D learning.

Images