Quality metrics for automatic evaluation of dual ISH images
The present disclosure is directed to a computer system designed to (i) receive a series of images as input; (ii) compute a number of metrics derived from focus features and color separation features within the images; and (iii) evaluate the metrics to return (a) an identification of the most suitable z-layer in a z-stack, given a series of z-layer images in a z-stack; and/or (b) an identification of those image tiles that are more suitable for cellular based scoring by a medical professional, given a series of image tiles from an area of interest of a whole slide scan.
1. A computer system for determining the most suitable z-layer in a z-stack of digital images, the z-stack of digital images depicting different layers of a single sample, comprising one or more processors and one or more memories, the one or more memories storing non-transitory computer-readable instructions for execution by the one or more processors to cause the one or more processors to: (a) compute (i) one or more focus metrics for each z-layer within the z-stack of digital images, and (ii) one or more color separation metrics for each z-layer within the z-stack of digital images; and (b) evaluate the computed one or more focus metrics and/or the computed one or more color separation metrics to determine a most suitable z-layer within the z-stack, wherein the computed one or more focus metrics comprise a focus quality score for each z-layer, wherein the focus quality score is derived from one or more focus features in an image tile derived from the z-stack of digital images, and wherein the one or more focus features are selected from the group consisting of: (i) a mean/median of the 0 th Difference of Gaussian (“DoG”) layer for all dot pixels in the image tile; (ii) a mean/median of the 0 th DoG layer for all black dot pixels in the image tile; (iii) a mean/median of the max DoG values of the 0 th layer for all dot blobs in the image tile; and (iv) a mean of the top DoG values for the 0 th layer for all dot blobs in the image tile.
2. The system of claim 1 , wherein the computed one or more color separation metrics comprise a color separation quality score for each z-layer.
3. The system of claim 2 , wherein the color separation quality score is derived from one or more color separation features.
4. The system of claim 3 , wherein the one or more color separation features are selected from the group consisting of: (i) A max , (ii) A sigmax ; (iii) a maximum in an unmixed black channel; (iv) a gradient value for at least one channel selected from the group consisting of a luminance channel, an unmixed red channel, an A channel, an A sig channel, and a green channel; (v) DoG values in at least one of a channel selected from the group consisting of a green channel, an A channel, and an optical density domain absorbance channel; and (vi) color domain features based on modeling color information in terms of ellipses.
5. The system of claim 1 , wherein the computed one or more focus metrics or the computed one or more color separation metrics for each z-layer are computed within an empirically determined color space optimized for signals of an in situ hybridization assay applied to the sample, wherein the sample is a tissue sample.
6. The system of claim 1 , further comprising identifying the z-layer having either a best focus and/or a best color separation.
7. A non-transitory computer-readable medium storing computer-executable instructions, which, when executed by one or more processors, cause the one or more processors to: (i) compute one or more focus metrics and one or more color separation metrics for each z-layer within a z-stack of images, each z-layer within the z-stack of images corresponding to an image of a biological sample, and (ii) evaluate the focus metrics or color separation metrics to determine a most suitable z-layer within the z-stack wherein the computed one or more focus metrics comprise a focus quality score for each z-layer, wherein the focus quality score is derived from one or more focus features in an image tile derived from the z-stack of digital images, and wherein the one or more focus features are selected from the group consisting of: (i) a mean/median of the 0 th Difference of Gaussian (“DoG”) layer for all dot pixels in the image tile; (ii) a mean/median of the 0 th DoG layer for all black dot pixels in the image tile; (iii) a mean/median of the max DoG values of the 0 th layer for all dot blobs in the image tile; and (iv) a mean of the top DoG values for the 0 th layer for all dot blobs in the image tile.
8. The non-transitory computer-readable medium of claim 7 , wherein the computed one or more color separation metrics comprise a color separation quality score for each z-layer.
9. The non-transitory computer-readable medium of claim 8 , wherein the color separation quality score is derived from one or more color separation features.
10. The non-transitory computer-readable medium of claim 9 , wherein the one or more color separation features are selected from the group consisting of: (i) A max , (ii) A sigmax ; (iii) a maximum in an unmixed black channel; (iv) a gradient value for at least one channel selected from the group consisting of a luminance channel, an unmixed red channel, an A channel, an A sig channel, and a green channel; (v) DoG values in at least one of a channel selected from the group consisting of a green channel, an A channel, and an optical density domain absorbance channel; and (vi) color domain features based on modeling color information in terms of ellipses.
11. The non-transitory computer-readable medium of claim 7 , wherein the instructions further comprise identifying the z-layer having either a best focus and/or a best color separation.
12. The computer system of claim 1 , wherein the one or more focus features are derived from a difference of gaussian layer.
13. The non-transitory computer-readable medium of claim 7 , wherein the one or more focus features are derived from a difference of gaussian layer.
14. A non-transitory computer-readable medium storing computer-executable instructions, which, when executed by one or more processors, cause the one or more processors to: (i) compute one or more focus metrics or one or more color separation metrics for each z-layer within a z-stack of images, each z-layer within the z-stack of images corresponding to an image of a biological sample, and (ii) evaluate the focus metrics or color separation metrics to determine a most suitable z-layer within the z-stack; wherein the computed one or more focus metrics comprise a focus quality score for each z-layer; wherein the focus quality score for each z-layer is derived from one or more focus features in an image tile derived from the z-stack of digital images; and wherein the one or more focus features are selected from the group consisting of: (i) a mean/median of the 0 th DoG layer for all dot pixels in the image tile; (ii) a mean/median of the 0 th DoG layer for all black dot pixels in the image tile; (iii) a mean/median of the max DoG values of the 0 th layer for all dot blobs in the image tile; and (iv) a mean of the top DoG values for the 0 th layer for all dot blobs in the image tile.