User Reference Manual

Transcription

1 User Reference Manual IN SITU HYBRIDIZATION QUANTIFICATION (for research purposes only)

2 Contents Overview... 3 System Requirements... 3 Compatible File Formats... 3 Input Parameters... 4 Output Parameters... 7 Histograms... 7 Quantifying Single Probe RNA ISH Slides Quantifying Single Probe SISH & CISH Slides Quantifying Dual Probe SISH & CISH Slides Contact Information... 13

3 Overview The Indica Labs ISH quantification tool is intended for research use only. It can be used to measure SISH, CISH, Dual CISH, or RNA ISH signals in digital microscopic images. The algorithm generates a markup image that is used by the pathologist to assess performance of the algorithm and for algorithm tuning purposes. It also generates quantitative data describing the analyzed regions of tissue. These data include total number of cells, total number of signals, average number of signals per cell, and a histogram which displays the frequency of cells that have 1, 2, 3,4 signals. All output data can be easily exported to an excel file, CSV (comma separated value) file. System Requirements The Indica Labs ISH quantification tool is designed to integrate seamlessly into the Aperio release 11 platform. This software can be installed to run server-side with the Aperio Spectrum software as well as client side within the Aperio ImageScope software. Both ImageScope and Spectrum should be at version 11.0 or later. Compatible File Formats Since this tool can be fully integrated into the Aperio platform it can be used to analyze all image formats that are compatible with Aperio ImageScope and Spectrum. At the time of writing this document, this includes the following image formats: JPG TIFF SVS (Aperio) NDPI (Hamamatsu) MRXS (Zeiss) JP2 CWS (Olympus)

4 Input Parameters This image analysis tool allows the user to fine tune the algorithm to meet the user s specific needs or to improve performance on a specific stain or tissue type. The following section describes each of the input parameters are in detail: Nuclear Stain specifies the optical densities of the RGB color components of the nuclear stain. Default is Haematoxylin. This parameter should consist of three comma separated decimal values each ranging from 0.0 to 1.0. The simplest way to generate these values is to download the free ImageScope EyeDropper Plugin and click on a representative nucleus. The EyeDropper plugin will calculate the RGB OD values for you and copy them to the clipboard. Note, this value is not used if Positive Signal Stain is set to 1.0, 1.0, 1.0 (see below). Positive Signal Stain 1 RNA ISH and CISH assays often identify the signals by staining them with a particular color. This parameter allows you to specify that color by providing comma separated optical densities of the RGB color components of the positive signal stain. Default is DAB. For non-black signals, the simplest way to generate the values for this parameter is to download the free ImageScope EyeDropper Plugin and click on a representative signal to calculate the RGB OD values. Then use the Markup Image parameter to verify the stain separation is working properly. Note, if your positive signals are black (as in SISH), then this value should be set to 1.0,1.0,1.0. Positive Signal Stain 2 For dual probe ISH assays, this value should define the stain for probe 2. Note, it is important that probe 1 and probe 2 are not swapped because the algorithm will calculated the ratios based on the number of probe 1 signals divided by the number of probe 2 signals. Set this value to: o 0.0,0.0,0.0 for single probe ISH. o 1.0,1.0,1.0 for dual probe ISH where probe 2 is black. o X,Y,Z for dual probe ISH where probe 2 is colored. Nuclear Contrast Threshold In order to detect nuclei, the algorithm relies on sufficient contrast between the darker nucleus and the lighter background staining. This value dictates how much contrast is required for nuclear detection. The value ranges from zero to one. Decreasing the value results in fewer nuclei detected, increasing the value results in more. Note, sometimes increasing this value has the unexpected effect of removing nuclei. This occurs because

5 multiple nuclei are joined together and then disqualified because their large size exceeds the Maximum Nuclear Size parameter. Minimum Nuclear Optical Density This parameter allows the user exclude faintly stained objects that are misinterpreted as nuclei. Since nuclei are typically stained darkest with haematoxylin you can set this threshold such that darker objects are true nuclei and lighter objects are not nuclei. The value of this parameter can be a decimal value ranging from 0 (totally white) to 1 (totally black). Values closer to zero will result in more nuclei detected and values closer to 1 will result in only the darkest nuclei being detected. Minimum Nuclear Size This parameter allows the algorithm to exclude objects that are too small in area to be nuclei. The specified value in microns squared should be set to the smallest nucleus size that you d like to detect. Objects smaller than this will not be counted and objects larger than this might be counted. Maximum Nuclear Size This parameter allows the algorithm to exclude objects that are too large in area to be nuclei. The specified value in microns squared should be set to the largest nucleus size that you d like to detect. Objects larger than this will not be counted and objects smaller than this might be counted. Nuclear Segmentation Aggressiveness Oftentimes multiple nuclei can be touching or overlapping resulting in what looks can be mistaken for a single connected object. This tool uses a segmentation algorithm to separate the connected nuclei and this parameter controls the level of segmentation that will occur. It ranges from 0.0 to 1.0. Increasing this value will result in more aggressive segmentation and ultimately a higher number of smaller vacuoles. Cell Radius From Nuclei This parameter allows the user to specify a cytoplasmic region around each nucleus (colored yellow in the markup image). For signal counting purposes, these areas are considered part of the cell. To only count signals within the nucleus, set this value to zero. Signal 1 Contrast Threshold This parameter controls how sensitive the algorithm should be when detecting signals (spots) for probe 1. Decreasing this value will cause will cause more probe 1 signals to be detected, but may result in

6 some false positives. Increasing this value will cause less signals to be detected but may result in false negatives. Signal 2 Contrast Threshold This parameter controls how sensitive the algorithm should be when detecting signals (spots) for probe 2. Decreasing this value will cause will cause more probe 2 signals to be detected, but may result in some false positives. Increasing this value will cause less signals to be detected but may result in false negatives. For single probe applications this parameter is not used. Signal 1 Size Quite often the CISH or SISH signals are tightly clustered and cannot be separated by eye or by computer algorithm. In these situations our software uses patent pending techniques to estimate the total number of signals within a cluster of signals. This value represents the estimated spot size in microns squared. Decreasing this value will cause signal 1 clusters to be counted as more signals and will result in a higher total signal 1 count and increasing this value will cause clusters to be calculated as less signals resulting in a lower total signal 1 count. In order to tune this parameter it is recommended that you create several annotation regions each one around individual cells. Then based on the results you can determine which signals are being counted as individual or multiple and can adjust the spot size as needed. Signal 2 Size Quite often the CISH or SISH signals are tightly clustered and cannot be separated by eye or by computer algorithm. In these situations our software uses patent pending techniques to estimate the total number of signals within a cluster of signals. This value represents the estimated spot size in microns squared for signal 2. Decreasing this value will cause signal 2 clusters to be counted as more signals and will result in a higher total signal 2 count and increasing this value will cause clusters to be calculated as less signals resulting in a lower total signal 2 count. In order to tune this parameter it is recommended that you create several annotation regions each one around individual cells. Then based on the results you can determine which signals are being counted as individual or multiple and can adjust the spot size as needed. Output Image This parameter is useful for verifying the effectiveness of the stain separation. After modifying Nuclear Stain, Positive Signal Stain 1, or Positive Signal Stain 2 be sure to test each of the corresponding output images to be sure they represent the separated stain.

7 Output Parameters This image analysis tool reports a set of results for each distinct analysis region and for the entire digital slide if analyzed. These results include the following output parameters: Total Cell Count total number of nuclei that were detected. Total Signal 1 Count - the total number of copies that were counted for probe 1. Total Signal 2 Count - the total number of copies that were counted for probe 2. Signal1/Signal2 total signal 1 count divided by total signal 2 count. Total Signal 1 Area this value represents the total area (in microns squared) of all counted copies of signal 1. Total Signal 2 Area this value represents the total area (in microns squared) of all counted copies of signal 2. Avg Signal 1 Per Cell this value represents the average number of signal 1 copies in a cell or the Total Signal 1 Count divided by the Total Cell Count. Avg Signal 2 Per Cell this value represents the average number of signal 2 copies in a cell or the Total Signal 2 Count divided by the Total Cell Count. Avg Signal 1 Area Per Cell this value (in microns squared) is the Total Signal 1 Area divided by the Total Cell Count. Avg Signal 1 Area Per Cell this value (in microns squared) is the Total Signal 2 Area divided by the Total Cell Count. Avg Signal 1 Optical Density this value describes how dark (optically dense) the signal 1 stain is within the signal 1 copies. It ranges from 0.0 to 1.0 where larger values are darker and smaller values are lighter. Avg Signal 2 Optical Density this value describes how dark (optically dense) the signal 2 stain is within the signal 2 copies. It ranges from 0.0 to 1.0 where larger values are darker and smaller values are lighter. Histograms This image analysis tool also generates a set of histograms which describe the frequency of cells having certain characteristics. These histograms are:

8 Probe 1 Signals Per Cell Histogram Displays the distribution of cells containing 0, 1, 2, 3, 4 copies of Probe1 Probe 2 Signals Per Cell Histogram Displays the distribution of cells containing 0, 1, 2, 3, 4 copies of Probe 2

9 Signal 1 / Signal 2 Ratio Histogram Displays the distribution of cells containing various ratios of signal 1 / signal 2.

10 Quantifying Single Probe RNA ISH Slides The algorithm can also be tuned to measure and count RNA ISH signals as shown below The blue areas represent the nuclei of cells found by the algorithm, the yellow areas represent the area that s been designated as cytoplasm surrounding the nucleus, and the red areas represent the RNA signals. In order to analyze RNA ISH slides, please note the following suggestions: Specify the nuclear stain and the positive signal stain 1. To simplify this, process, download the Free ImageScope EyeDropper Tool. For single probe applications be sure to unset the positive signal stain 2 (i.e. set it to 0.0,0.0,0.0) If the cell signals fall outside of the nucleus, but you still want to count them on a per cell basis, then adjust the Cell Radius from Nuclei parameter. This will in increase or decrease the size of the yellow areas in the image shown above.

11 Quantifying Single Probe SISH & CISH Slides In the example below, you can see the results of running the algorithm in a small area of a digitized SISH slide. The blue areas represent the nuclei of cells found by the algorithm, and the yellow areas represent the individual signals and red areas represent the clustered signals. Individual signals are counted as 1 copy and clustered signals are counted as two or more according to their size. In order to analyze single probe SISH and CISH slides, please note the following suggestions: Specify the nuclear stain and the positive signal stain 1. To simplify this, process, download the Free ImageScope EyeDropper Tool. Note, since the probe color is typically colorless (black) set the positive signal stain 1 optical densities to 1.0,1.0,1.0. Unset the positive signal stain 2 (i.e. set it to 0.0,0.0,0.0) Since the signals are expected to lie within the cell nuclei you can optionally set Cell Radius from Nuclei to zero so that only signals within the nuclei are counted and the signals from tangentially cut cells are not counted.

12 Quantifying Dual Probe SISH & CISH Slides In the example below, you can see the results of running the algorithm in a small area of a Ventana INFORM Her2 Dual ISH slide. The blue areas represent the nuclei of cells found by the algorithm, and the orange areas represent the individual probe 1 (Her2) signals and red areas represent the clustered Her2 signals. Chr17 is specified as probe 2 and it s individual and clustered signals are colored cyan and green respectively. In order to analyze single probe SISH and CISH slides, please note the following suggestions: Specify the nuclear stain, positive signal stain 1, and positive signal stain 2. To simplify this, process, download the Free ImageScope EyeDropper Tool. Typically for dual CISH, probe 1 will be black, in which case positive signal stain 1 should be set to 1.0,1.0,1.0. Since the signals are expected to lie within the cell nuclei you can optionally set Cell Radius from Nuclei to zero so that only signals within the nuclei are counted and the signals from tangentially cut cells are not counted.

13 Contact Information For technical support and other information, please contact Indica Labs, Inc. Visit: Dial: +1 (505)