Back to results
US 2025/0054135 A1Published application
Machine learning based evaluation of lateral flow tests
AssigneeLiftric GmbH
InventorsJakob HUBER, Ben JOHN, Thorsten KNOELLER
Life of this patent
Estimating the term…
Loading patent…
Loading patent…
AssigneeLiftric GmbH
InventorsJakob HUBER, Ben JOHN, Thorsten KNOELLER
Estimating the term…
The present invention concerns computer-implemented methods for use in lateral flow test evaluation. One embodiment of such a method comprises obtaining, preferably by a mobile electronic device (110), a digital image (202) that depicts at least one test cassette (100), wherein the test cassette (100) comprises at least one viewport (102) and wherein the viewport (102) comprises at least one test indicator (104). The method may comprise performing, preferably by the mobile electronic device (110), an image segmentation step (204) to recognize at least one test indicator (104) depicted in the digital image (202), and performing an evaluation step (206) for producing at least one evaluation result (208) based, at least in part, on the recognized at least one test indicator (104). The image segmentation step (204) may comprise generating at least one object marker (306, 406, 506), in particular at least one mask and/or bounding box, based on a downscaled version of the obtained digital image (202) and applying the at least one object marker (306, 406, 506) to the obtained digital image (202) or to a part thereof.
A computer-implemented method for lateral flow test evaluation, wherein the method comprises the following steps:
The method of any one of the preceding claims, wherein generating the at least one object marker (306, 406, 506) comprises processing the respective downscaled image (302, 402, 502) with at least one segmentation machine-learning model (304, 404, 504), preferably with a separate segmentation machine-learning model (304, 404, 504) for each object marker (306, 406, 506).
The method of any one of the preceding claims, wherein the evaluation step (206) comprises:
The method of any one of the preceding claims, wherein the downscaled digital image (302) is downscaled by a factor selected from the range of 5 to 15, more preferably from the range of 8 to 12, most preferably by a factor of 10 relative to the obtained digital image (202), or by a factor selected from the range of 2 to 8, more preferably from the range of 4 to 6, most preferably by a factor of 5 relative to the obtained digital image (202);
The method of any one of the preceding claims, wherein the downscaled digital image (302) adheres to an RGB color model, the downscaled test cassette image (402) adheres to an RGB color model and/or the downscaled viewport image (502) adheres to a Lab color model.
The method of any one of the preceding claims, which comprises at least one of the following further steps:
The method of any one of the preceding claims, wherein the first segmentation machine-learning model (304), the second segmentation machine-learning model (404), the third segmentation machine-learning model (504), the prediction machine-learning model (604, 704), the test cassette validation machine-learning model (412) and/or the viewport validation machine-learning model (512) comprises an artificial neural network, in particular a convolutional neural network, CNN.
The method of any one of the preceding claims, wherein the first segmentation machine-learning model (304), the second segmentation machine-learning model (404) and the third segmentation machine-learning model (504) is based on a U-Net or a Yolo model and the prediction machine-learning model (604, 704), the test cassette validation machine-learning model (412) and/or the viewport validation machine-learning model (512) is based on a MobileNetv2.
The method of any one of the preceding claims, being performed, at least in part, by a mobile electronic device (110), preferably a handheld device, in particular a handheld consumer device such as a smartphone or tablet computer.
Dependencies are inferred from the claim text.