US 12,249,044 B2
Ocular fundus image processing method, ocular fundus image processing device, ocular fundus image processing program, and recording medium having said program recorded thereon
Kibo Ote, Hamamatsu (JP); Fumio Hashimoto, Hamamatsu (JP); Hidenao Yamada, Hamamatsu (JP); and Akira Obana, Hamamatsu (JP)
Assigned to HAMAMATSU PHOTONICS K.K., Hamamatsu (JP); and Social Welfare Corporation Seirei Social Welfare Community, Hamamatsu (JP)
Appl. No. 17/800,062
Filed by HAMAMATSU PHOTONICS K.K., Hamamatsu (JP); and Social Welfare Corporation Seirei Social Welfare Community, Hamamatsu (JP)
PCT Filed Jan. 5, 2021, PCT No. PCT/JP2021/000077
§ 371(c)(1), (2) Date Aug. 16, 2022,
PCT Pub. No. WO2021/171788, PCT Pub. Date Sep. 2, 2021.
Claims priority of application No. 2020-034202 (JP), filed on Feb. 28, 2020.
Prior Publication US 2023/0078077 A1, Mar. 16, 2023
Int. Cl. G06T 5/50 (2006.01); A61B 3/12 (2006.01); G06T 5/00 (2006.01)
CPC G06T 5/00 (2013.01) [A61B 3/12 (2013.01); G06T 5/50 (2013.01); G06T 2207/10024 (2013.01); G06T 2207/10064 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20224 (2013.01); G06T 2207/30041 (2013.01)] 16 Claims
OG exemplary drawing
 
1. An ocular fundus image processing method comprising:
acquiring a first image which is a fluorescence image of an ocular fundus of a subject generated by irradiating the ocular fundus with excitation light of a first wavelength;
acquiring a second image which is a fluorescence image of the ocular fundus of the subject generated by irradiating the ocular fundus with excitation light of a second wavelength different from the first wavelength;
generating a plurality of trained deep learning models for predicting a correction factor for calculating a quantity of macular pigment of the subject from input images including at least the first image and the second image through training using a plurality of different initial values;
predicting a plurality of correction factors by inputting the input images including at least the first image and the second image to the plurality of trained deep learning models;
calculating a statistical value of the plurality of correction factors and deriving the statistical value as the correction factor of the subject; and
calculating a quantity of macular pigment of the subject on the basis of at least one of the first image and the second image and the correction factor of the subject.