US 11,869,208 B2
Methods, apparatuses, and computer programs for processing pulmonary vein computed tomography images
Horng-Shing Lu, Hsinchu (TW); Chih-Min Liu, Taipei (TW); Shih-Lin Chang, Taipei (TW); Shih-Ann Chen, Taipei (TW); Yenn-Jiang Lin, Taipei (TW); Hung-Hsun Chen, Hsinchu (TW); and Wei-Shiang Chen, Hsinchu (TW)
Assigned to TAIPEI VETERANS GENERAL HOSPITAL, Taipei (TW); and National Yang Ming Chiao Tung University, Hsinchu (TW)
Filed by TAIPEI VETERANS GENERAL HOSPITAL, Taipei (TW); and National Yang Ming Chiao Tung University, Hsinchu (TW)
Filed on Mar. 15, 2021, as Appl. No. 17/201,167.
Claims priority of provisional application 62/990,254, filed on Mar. 16, 2020.
Prior Publication US 2021/0287365 A1, Sep. 16, 2021
Int. Cl. G06N 3/045 (2023.01); G06T 7/70 (2017.01); G06T 7/00 (2017.01); G06V 10/764 (2022.01); G06V 10/774 (2022.01)
CPC G06T 7/70 (2017.01) [G06N 3/045 (2023.01); G06T 7/0012 (2013.01); G06V 10/764 (2022.01); G06V 10/774 (2022.01); G06T 2207/10072 (2013.01); G06T 2207/20024 (2013.01); G06T 2207/20076 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30048 (2013.01)] 18 Claims
OG exemplary drawing
 
1. A method for processing pulmonary vein computed tomography (PVCT) images, comprising:
obtaining a plurality of input PVCT images from the upper border of a left atrium to the bottom of a heart;
determining, by a residual network model, whether each of the plurality of input PVCT images relates to a non-pulmonary vein (NPV) trigger origin, wherein a first convolution operation is performed based on the plurality of input PVCT images and a first convolution layer, and a second convolution operation is performed based on a second convolution layer, and
determining the plurality of input PVCT images relating to a NPV trigger origin when more than half of the plurality of input PVCT images are determined relating to a NPV trigger origin,
wherein:
a first filter of the first convolution layer and a second filter of the second convolution layer are determined based on a training set and an internal validation set,
the training set includes a first portion of the multiple patients and the corresponding training PVCT images,
the internal validation set includes a second portion of the multiple patients and the corresponding training PVCT images, and
the internal validation set determines whether the first filter of the first convolution layer and the second filter of the second convolution layer are convergent.