Devices, systems and methods for automated quantitative scoring of digitized tissue images
Methods and systems for generating a heat map that reduces bias in selecting FOVs are disclosed. Some disclosed methods include annotating a primary stained image, registering the annotation to a secondary serial specific stained image, using an image analysis algorithm to compute a scoring criteria specific to the tissue and assay type for tiled regions in the image, using a sliding window in the annotated tumor region to compute values for each pixel in a heat map which correlate to the specific scoring criteria, displaying the heat map at low resolution, ranking and selecting hot spots, selecting FOVs from the hot spot regions which results in displaying the slide-level score for the FOVs. The systems comprise, among other things, software configured to perform the referenced method.
1. A method of generating a heat map representation of a specified scoring criteria, the method comprising:
selecting a digitized staining slide of a tumor image depicting at least part of a tumor;
computing a scoring criteria specific to the tumor and staining type at scanned resolution; and
generating a low-resolution heat map by correlating pixel values in the low-resolution heat map to the scoring criteria, wherein correlating the pixel values in the low-resolution heat map to the scoring criteria comprises:
computing an average value for the scoring criteria in a windowed region of predetermined size in an annotated tumor region which average value correlates to a pixel value; and
sliding the windowed region across the annotated tumor region to iteratively compute pixel values and generate the low-resolution heat map.
2. The method of claim 1 , further comprising:
annotating the tumor image on the digitized staining slide; and after annotating the tumor image, computing the scoring criteria.
3. The method of claim 2 , wherein the digitized staining slide is a digitized primary staining slide and the method further comprises registering annotations from the digitized primary staining slide to a digitized specific staining slide of the tumor image.
4. The method of claim 3 , wherein annotating the tumor image comprises displaying the digitized primary staining slide on a display so that a user can manually demarcate a whole tumor region using an input device.
5. The method of claim 3 , wherein annotating the tumor image in the digitized primary staining slide comprises processing the digitized primary staining slide with a tumor identification algorithm so that the tumor image is automatically demarcated.
6. The method of claim 3 , wherein, registering annotations from the digitized primary staining slide includes applying an inter-marker image registration algorithm to map tumor region annotations from the digitized primary staining slide to the digitized specific staining slide.
7. The method of claim 1 , further comprising saving the computed scoring criteria to a database, and selecting Field-of-Views from locally dominant regions in the low-resolution heat map which returns the computed scoring criteria for the selected Field-of-Views to obtain a slide-level score.
8. The method of claim 1 , wherein computing scoring criteria comprises gridding the annotated tumor region into tiles of a predetermined size and computing the scoring criteria for each tile at scanned resolution.
9. The method of claim 1 , further comprising using an algorithm to rank, select and highlight hot spot regions having a high density of specific staining.
10. The method of claim 9 , wherein ranking comprises selecting local maxima within a predetermined neighborhood of values in the low-resolution heat map and sorting them in order.
11. The method of claim 1 , wherein computing scoring criteria comprises computing multiple scoring criteria and generating a heat map comprises generating a heat map for each one of the multiple scoring criteria.
12. A system comprising:
one or more data processors; and
a non-transitory computer readable storage medium containing instructions which, when executed on the one or more data processors, cause the one or more data processors to perform actions including:
selecting a digitized staining slide of a tumor image depicting at least part of a tumor;
computing a scoring criteria specific to the tumor and staining type at scanned resolution; and
generating a low-resolution heat map by correlating pixel values in the low-resolution heat map to the scoring criteria, wherein correlating the pixel values in the low-resolution heat map to the scoring criteria comprises:
computing an average value for the scoring criteria in a windowed region of predetermined size in an annotated tumor region which average value correlates to a pixel value; and
sliding the windowed region across the annotated tumor region to iteratively compute pixel values and generate the low-resolution heat map.
13. The system of claim 12 , wherein the actions further include annotating the tumor image on the digitized staining slide; and after annotating the tumor image, computing the scoring criteria.
14. The system of claim 13 , wherein the digitized staining slide is a digitized primary staining slide and the method further comprises registering annotations from the digitized primary staining slide to a digitized specific staining slide of the tumor image.
15. The system of claim 14 , wherein annotating the tumor image comprises displaying the digitized primary staining slide on a display so that a user can manually demarcate a whole tumor region using an input device.
16. 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 actions including:
selecting a digitized staining slide of a tumor image depicting at least part of a tumor;
computing a scoring criteria specific to the tumor and staining type at scanned resolution; and
generating a low-resolution heat map by correlating pixel values in the low-resolution heat map to the scoring criteria, wherein correlating the pixel values in the low-resolution heat map to the scoring criteria comprises:
computing an average value for the scoring criteria in a windowed region of predetermined size in an annotated tumor region which average value correlates to a pixel value; and
sliding the windowed region across the annotated tumor region to iteratively compute pixel values and generate the low-resolution heat map.
17. The computer-program product of claim 16 , wherein the actions further include:
saving the computed scoring criteria to a database, and selecting Field-of-Views from locally dominant regions in the low-resolution heat map which returns the computed scoring criteria for the selected Field-of-Views to obtain a slide-level score.
18. The computer-program product of claim 16 , wherein computing scoring criteria comprises gridding the annotated tumor region into tiles of a predetermined size and computing the scoring criteria for each tile at scanned resolution.