US 12,067,699 B2
Production method of learned model, brightness adjustment method, and image processing apparatus
Wataru Takahashi, Kyoto (JP); and Yusuke Tagawa, Kyoto (JP)
Assigned to SHIMADZU CORPORATION, Kyoto (JP)
Appl. No. 17/282,229
Filed by Shimadzu Corporation, Kyoto (JP)
PCT Filed Oct. 3, 2018, PCT No. PCT/JP2018/037075
§ 371(c)(1), (2) Date May 19, 2021,
PCT Pub. No. WO2020/070834, PCT Pub. Date Apr. 9, 2020.
Prior Publication US 2021/0350513 A1, Nov. 11, 2021
Int. Cl. G06T 5/00 (2024.01); A61B 6/00 (2006.01); G06N 20/00 (2019.01); G06T 5/50 (2006.01); G06T 5/92 (2024.01)
CPC G06T 5/92 (2024.01) [A61B 6/5258 (2013.01); G06N 20/00 (2019.01); G06T 5/50 (2013.01)] 12 Claims
OG exemplary drawing
 
1. A method of producing a trained model by machine learning using training data including an input training image and ground truth data, the trained model configured for receiving an input radiographic image reflecting a subject to output a brightness adjustment parameter of the radiographic image, the method comprising:
inputting the training image to a training model, the training model being under production, to output a brightness adjustment parameter;
acquiring a value of a loss function based on the output brightness adjustment parameter and the ground truth; and
adjusting the value of the loss function to be decreased, to optimize the training model,
wherein the adjusting the value of the loss function includes adjusting the loss function so that the value thereof is higher in a first situation compared to a second situation, causing a brightness-adjusted image, generated based upon the output brightness adjustment parameter, to have a higher contrast in a predetermined area therein,
wherein the first situation being such that the contrast of a predetermined area becomes higher compared to the training image, and
wherein the second situation being such that the contrast of a predetermined area becomes lower compared to the training image.