Image-Based Assay Performance Improvement

Abstract

The present invention is related to correct the errors in instruments, operation, and others using intelligent monitoring structures and machine learning, and others.

We claim: 
     
         1 . A system for assaying a sample using an optical system suspected of having an imperfection, comprising:
 a sample holder comprising a first plate, a second plate, and a plurality of monitoring marks, wherein
 (i) each of the first and second plates comprises a respective sample contact area contacting the sample that contains or is suspected of containing an analyte; 
 (ii) the plurality of monitoring marks comprise a plurality of spacers that are distributed in the sample contact areas, wherein the spacers regulate the spacing between the first and second plates, and wherein the plurality of monitoring marks have a flat surface that is parallel to the surface and the plates, have at least one predetermined parameter associated with a geometric property or an optical property of the spacers, and a lateral linear dimension of 1 μm or larger, and at least one lateral linear dimension of 300 μm or less; 
 (iii) the first and second plates sandwich at least a part of the sample into a thin layer having a substantially uniform thickness of 200 μm or less and 
 (iv) one of the first and second plates is flexible; the fourth power of the inter-spacer-distance (ISD) divided by the thickness (h) and the Young's modulus (E) of the flexible plate (ISD4/(hE)) is 5×10 6  μm 3 /GPa or less, and the thickness of the flexible plate times the Young's modulus of the flexible plate is in the range 60 to 750 GPa-μm; 
   an optical imaging system that is suspected of having an imperfection, wherein the optical imaging system is configured to take one or more images of the at least part of the sample, wherein the imperfection is associated with an image distortion and/or an illumination variation, and wherein the imperfection of the optical system causes an image imperfection in the image; and   a non-transitory computer-readable medium configured to store a correction algorithm that corrects the image imperfection using the at least one predetermined parameter and the image of the monitoring marks in the at least part of the sample.   
     
     
         2 . The system of  claim 1 , further comprising:
 a second non-transitory computer-readable medium storing a machine learning model, wherein the machine learning model is configured to transform an image taken by the optical imaging system into a transformed image, and the training and/or the image transformation of the machine learning model are performed after the images of the at least part of the sample has been corrected by the correction algorithm.   
     
     
         3 . A method of assaying a sample using an optical system suspected of having an imperfection, the method comprising:
 providing the system of  claim 1  or  2 , and   correcting, using the correction algorithm, the image imperfection using the at least one predetermined parameter and the one or more images of the monitoring marks in the at least part of the sample.   
     
     
         4 . The method of  claim 3 , further comprising:
 using a machine learning model to perform an image transformation of the image of the at least part of the sample; and   using the transformed image to measure the analyte,   wherein the training and/or the image transformation of the machine learning model are performed after the image of the at least part of the sample has been corrected by the correction algorithm.   
     
     
         5 . The system of  claim 1 , further comprises a computer program product for assaying a micro-feature in a sample, the program comprising computer program code applied and adapted for, in at least one image:
 (a) receiving an image of a sample and the monitoring mark(s), wherein the sample is loaded into device of any prior device claim, wherein the image is taken by an imager; and   (b) processing and analyzing the image to calculate the amount of the micro-feature, wherein the analyzing uses a detection model that is based on machine learning and the information provided by the image of the monitoring mark(s).