Systems and methods for analyzing electronic images for quality control
Systems and methods are disclosed for receiving a digital image corresponding to a target specimen associated with a pathology category, determining a quality control (QC) machine learning model to predict a quality designation based on one or more artifacts, providing the digital image as an input to the QC machine learning model, receiving the quality designation for the digital image as an output from the machine learning model, and outputting the quality designation of the digital image. A quality assurance (QA) machine learning model may predict a disease designation based on one or more biomarkers. The digital image may be provided to the QA model which may output a disease designation. An external designation may be compared to the disease designation and a comparison result may be output.
1 . A computer-implemented method for quality control of at least one digital image of a sample mounted on a slide, comprising:
accessing the at least one digital image that depicts the sample mounted on the slide, wherein the at least one digital image was captured using at least one imaging device of a slide imaging apparatus;
determining, using a first quality control machine learning model, a quality score of the at least one digital image and a quality score classification by determining one or more artifacts of sub-regions of at least one region of interest of the at least one digital image by:
applying, using the first quality control machine learning model, at least one digital image filter, scanning technique, or pixel comparison technique to the at least one region of interest to detect the one or more artifacts, and
comparing, using the first quality control machine learning model, a number and/or extent of the one or more artifacts within the at least one region of interest to a threshold value of artifacts, the one or more artifacts comprising a blur, missing tissue, line, folded tissue, bubbles, cracks in glass, scratches, dust, and/or pen marking of the at least one digital image, wherein the quality score classification of the at least one region of interest is determined depending on the comparison; and
generating, using a second quality control machine learning model, at least one quality designation based on the quality score classification of the quality score and the one or more artifacts, wherein:
the at least one quality designation is either an approved designation or a rejected designation, and
the second quality control machine learning model having been trained to predict the at least one quality designation based on a plurality of digital images, a plurality of quality scores, a plurality of quality score classifications, an approval scale, and a rejection scale.
2 . The method of claim 1 , further comprising:
outputting a visual output of the at least one quality designation.
3 . The method of claim 2 , wherein the visual output is a visual indicator, and the visual indicator highlights a given artifact of the at least one digital image.
4 . The method of claim 3 , wherein the visual output is an overlay that highlights a location of the artifacts on the at least one digital image.
5 . The method of claim 3 , wherein the visual output is an overlay that highlights a degree of a presence of the artifacts.
6 . The method of claim 1 , wherein the artifacts include scanning results and/or errors of the at least one digital image.
7 . The method of claim 1 , wherein the at least one region of interest is the at least one digital image or a section of the at least one digital image.
8 . The method of claim 1 , wherein one or both of the first quality control machine learning model or the second quality control machine learning model is based on a convolutional neural network.
9 . A system comprising:
a memory storing instructions; and
a processor executing the instructions to perform operations comprising:
accessing at least one digital image that depicts a sample mounted on a slide, wherein the at least one digital image was captured using at least one imaging device of a slide imaging apparatus;
inputting the at least one digital image into a quality control machine learning model;
automatically determining, using the quality control machine learning model, a quality score of the at least one digital image and a quality score classification by determining one or more artifacts of sub-regions of at least one region of interest of the at least one digital image by:
applying, using the quality control machine learning model, at least one digital image filter, scanning technique, or pixel comparison technique to the at least one region of interest to detect the one or more artifacts, and
comparing, using the quality control machine learning model, a number and/or extent of the one or more artifacts within the at least one region of interest to a threshold value of artifacts, the one or more artifacts comprises a blur, missing tissue, line, folded tissue, bubbles, cracks in glass, scratches, dust, and/or pen marking of the at least one digital image, wherein the quality score classification of the at least one region of interest is determined depending on the comparison; and
automatically generating, using the quality control machine learning model, at least one quality designation based on the quality score classification of the quality score and the one or more artifacts, wherein:
the at least one quality designation is either an approved designation or a rejected designation, and
the quality control machine learning model has been trained to predict the at least one quality designation based on a plurality of digital images, a plurality of quality scores, a plurality of quality score classifications, an approval scale, and a rejection scale.
10 . The system of claim 9 , further comprising the at least one imaging device.
11 . The system of claim 9 , wherein the slide imaging apparatus includes least one controlling and evaluation device.
12 . The system of claim 9 , further comprising outputting a visual output of the at least one quality designation.
13 . The system of claim 12 , wherein the visual output is a visual indicator, and the visual indicator highlights a given artifact of the at least one digital image.
14 . The system of claim 13 , wherein the visual output is an overlay that highlights a location of the artifacts on the at least one digital image.
15 . The system of claim 12 , wherein the visual output is an overlay that highlights a degree of a presence of the artifacts.
16 . The system of claim 9 , wherein the artifacts include scanning results or errors of the at least one digital image.
17 . A computer-program product tangibly embodied in a non-transitory machine-readable storage medium, including instructions configured to cause one or more data processors to perform a set of operations comprising:
accessing at least one digital image that depicts a sample mounted on a slide, wherein the at least one digital image was captured using at least one imaging device of a slide imaging apparatus;
inputting the at least one digital image into a first quality control machine learning model;
automatically determining, using the first quality control machine learning model, a quality score of the at least one digital image and a quality score classification by determining one or more artifacts of sub-regions of at least one region of interest of the at least one digital image by:
applying, using the first quality control machine learning model, at least one digital image filter, scanning technique, or pixel comparison technique to the at least one region of interest to detect the one or more artifacts, and
comparing, using the first quality control machine learning model, a number and/or extent of the one or more artifacts within the at least one region of interest to a threshold value of artifacts, the one or more artifacts comprises a blur, missing tissue, line, folded tissue, bubbles, cracks in glass, scratches, dust, and/or pen marking of the at least one digital image, wherein the quality score classification of the at least one region of interest is determined depending on the comparison; and
automatically generating, using a second quality control machine learning model, at least one quality designation based on the quality score classification of the quality score and the one or more artifacts, wherein:
the at least one quality designation is either an approved designation or a rejected designation,
the at least one quality designation is indicative of artifacts in the at least one digital image, and
the second quality control machine learning model has been trained to predict the at least one quality designation based on a plurality of digital images, a plurality of quality scores, a plurality of quality score classifications, an approval scale, and a rejection scale that is indicative of an image quality.
18 . The method of claim 1 , further comprising:
determining a report based on the at least one quality designation, the report including types of artifacts found, number of each type of artifact, clinical impact of artifacts, time to rectify artifact, and/or potential patterns or correlations in artifacts.
19 . The system of claim 9 , further comprising:
determining a report based on the at least one quality designation, the report including types of artifacts found, number of each type of artifact, clinical impact of artifacts, time to rectify artifact, and/or potential patterns or correlations in artifacts.
20 . The computer-program product of claim 17 , further comprising:
determining a report based on the at least one quality designation, the report including types of artifacts found, number of each type of artifact, clinical impact of artifacts, time to rectify artifact, and/or potential patterns or correlations in artifacts.