US 12,243,207 B2
Image taking method, image analysis method, method for training an image analysis neural network, and image analysis neural network
Gábor Bayer, Budapest (HU)
Assigned to 77 Elektronika Müszeripari Kft., Budapest (HU)
Appl. No. 17/609,547
Filed by 77 Elektronika Müszeripari Kft., Budapest (HU)
PCT Filed Mar. 31, 2020, PCT No. PCT/HU2020/050011
§ 371(c)(1), (2) Date Nov. 8, 2021,
PCT Pub. No. WO2020/225580, PCT Pub. Date Nov. 12, 2020.
Claims priority of application No. P1900144 (HU), filed on May 8, 2019; and application No. P2000067 (HU), filed on Feb. 24, 2020.
Prior Publication US 2022/0230295 A1, Jul. 21, 2022
Int. Cl. G06T 7/00 (2017.01); G01N 15/04 (2006.01); G02B 21/36 (2006.01); G06N 3/08 (2023.01); H04N 23/67 (2023.01)
CPC G06T 7/0002 (2013.01) [G01N 15/042 (2013.01); G02B 21/365 (2013.01); G06N 3/08 (2013.01); H04N 23/67 (2023.01); G06T 2207/10016 (2013.01); G06T 2207/10028 (2013.01); G06T 2207/10056 (2013.01); G06T 2207/10148 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30168 (2013.01)] 14 Claims
OG exemplary drawing
 
1. An image analysis method for analysing images, said images being taken by an image taking method, said image taking method comprising: taking said images with a microscope and selecting a subset of the taken images for image analysis, in the image taking method taking the images of a sediment of a liquid filled in an analyzing space located between a transparent upper window portion and a transparent lower window portion of a container adapted for analyzing the liquid, wherein the sediment is centrifuged on an inner flat surface of the transparent lower window portion, wherein the image analysis method comprises:
taking, with the microscope and in a first spatial region of the analyzing space, a first depth image sequence at first focal plane positions shifted with an equal step size in a depth direction perpendicular to the flat surface,
selecting, from the first depth image sequence and by applying a focusing algorithm, an image with a best contrast for the image analysis,
taking, with the microscope and in a second spatial region of the analyzing space, a second depth image sequence at second focal plane positions shifted with an equal step size in the depth direction, wherein the step size of the second depth image sequence is the same or is smaller than that of the first depth image sequence, wherein the second depth image sequence has fewer images than the first depth image sequence, and wherein an intermediate focal plane position, of intermediate focal plane positions from the second focal plane positions of the second depth image sequence, is the same as a focal plane position with a best contrast of the first depth image sequence,
selecting, from the second depth image sequence and by applying the focusing algorithm, an image with the best contrast for the image analysis,
also selecting, for the image analysis, at least one additional image from each depth image sequence, wherein a respective focal plane position of the at least one additional image is located at a same relative distance from a best contrast focal plane position in each depth image sequence, and
providing each selected image to an image analysis module having a neural network trained for the image analysis to cause the neural network to perform the image analysis.