Tissue microarray
互联网
Creating an interactive database for image viewing and data input.
The major problem of using tissue microarray is how to simultaneously examine each core at the microscope, know its identity and input the score in a database next to the correct identifier.
Solutions proposed in the literature [Refs 1, 2 ] request knowledge of computer programs such as Access and the help of a computer wizard to link your image array to the database.
A major effort toward unifying and making available tools for the analysis of TMA has been recently made by a Stanford-based consortium. Browse their website at the Stanford TMA software website and companion to "Software Tools for High-Throughput Analysis and Archiving of Immunohistochemistry Staining Data Obtained with Tissue Microarrays" (2002) and get their AJP paper .
Most of the time you may need to handle simpler arrays with less complex databases. Still, scoring between 50 and 300 images arrayed is a daunting task.
Here we show how to set up a Microsoft Excel Database which allows exactly this. Examples and a template are also provided.
Let's assume you have a 3 x 9 tissue array made from nine surgical specimens (3 cores per specimen), and an Excel spreadsheet containing the identifier for each core.
Creating the relational database in Excel.
Your database with the cases, which was your template to construct the tissue block looks like this:
On the left, column A with the cases (numbers are fictitious), in numerical order.
Center top: the 3x9 database used to construct the TMA
Below, the row and column identifier for each core, A1 through C9.
Each case has been cored with a 1mm Beecher Instruments arrayer in triplicate. This is how an H&E stain looks like.
Each case is then colored, to facilitate identification and occupies three Excel cells.
The position of each core is recorded next to his Pathology identifier. This to control for accuracy of linking when making the links. When cases are not aligned in the TMA, this is crucial.
Inserting Excel links.
Excel can insert functions in a database cell. The "Hyperlink" function is described as:
You can customize this path to make your database reach the images on your HardDisk, in a folder, of your CDrom etc.
On your computer HardDisk, the files and folders are placed as in the example below. Is easy to understand the path which drives Excel to the folder "Container1", which contains two folders, "StainX" and "StainY", this latter containing 27 .jpg images. The path is Container1/StainY/a1.jpg
The first hyperlink for the first image, a1.jpg, is inserted, aligned with the case which represents.
The first part within parenthesis "Container/FolderX/a1.jpg" is the address by which Excel locates and opens the image named a1.jpg .
The second part is something you like to appear in the cells where the link originate. Here is named "placeHolder" but can be an asterisk (see below) or else.
Then the hyperlink for the second image, a2.jpg, is added
and for the third
Finally, all the links are put in place, this time with an asterisk ~undefined) instead of the "place holder".
When your mouse goes over the links, the arrow turns into a hand pointer 
and the link path is highlighted. In the example below, cell E7 is highlighted and correctly points to image b5.jpg Check for accuracy of the link by comparing with the position of the cores in column B.
One or more columns next to the now named linked columns are used to input the score of the stain.
To scale-up the input of hyperlinks (e.g. >100 cores), see the "advanced technique " file.
Duplicating and renaming linked columns.
The column is highlighted and copyed
then pasted in the free adjacent columns
The new column is re-named.
To re-address Excel to a new folder and a new stain via the links, each link is modified with the Find-Replace function.
The new pasted columns are highlighted and the name of the folder to be changed, replaced. Be careful to change only that name and not to disrupt the link path.
Now the renamed links show the new path. Once you check the correctness of links in the original columns, you don't need to re-check in the copies.
Color-coding of the columns makes it easier to follow longitudinally a single case.
Now your database is finished and you can add ad libitum columns with new stains. Note how easy is to browse the database per stains (columns) or per case (rows).
The best way to become familiar with the sheet is to dowload the demo file and analyzing it (see below ).
Image taking.
To link the images to the database, images of each core of the whole tissue array are taken sequentially according to the database order and named A1.jpg through C9.jpg, as in our example, irrespectively of the stain, magnification or other identifiers.
This makes image taking error-free, rapid (150 H&E at 20x can be taken in 1h 30min), and efficient.
You only have to care to focus and select the most representative field. The analysis will be done at the computer screen.
Images are saved as .JPG compressed files. A 3.5Mb image is shrunk down to approx 1MB, still has enough detail to be viewed at 200% (equivalent to a 40x).
Then the images are allocated in a folder named with the name of the stain or any name of choice, but indicative of what is the content. You can put a text file with the all the information (stain name, notes, magnification) in each folder.
The correct identification of your core relies on the correct placement of the image links in your relational database.
Data input.
By clicking the hyperlink cell, Excel opens up the image(s) in Adobe Photoshop (see note for Mac Users). Open sequentially all the images you need.
After you evaluate that, return to the Excel database and input your score and notes in the cell next to the case.
The database can be sorted and rearranged with Excel, to suit viewing, scoring and analyzing.
Note for Macintosh users.
Links to .jpg files in Microsoft Excel 2001 open with PictureViewer, instead of Adobe Photoshop, resulting in much less versatility of the viewing ability.
You can trick Excel and force it to use Adobe Photoshop by manually changing the file name from XYZ.jpg to XYZ.psd . You can do it automatically with an AppleScript file (download it here . Is a file compressed with StuffitExpanderTM ).
Importing and testing the template database.
here or a Demo Folder for Mac (Approx 28 Mb, compressed with Stuffit ExpanderTM ) or Demo Folder for Windows containing the demo Excel file and 27 images.
Please note that a moderately skilled user of Excel, supported by the Help menu tips, is able to construct such a tool. I will not be available to help you.
However, you can e-mail me your comments.
