The overall goal of the following experiment is to use high throughput image analysis software to measure the size of three dimensional tumor S imaged with brightfield microscopy. This is achieved by first arranging all the images with 3D tumor steroid into the correct directory structure and file names for this steroid sizer program. As a second step steroid, sizer automatically computes the boundary of this steroid based on the active contour algorithm and measures the size of this steroid.
Next, a well-designed quality control sweep is made to ensure optimal results for every image. After analyzing the result in data, the growth of the 3D steroid under experimental treatment can be elucidated. The main advantage of this computer program is that it integrates a powerful automated image analysis algorithm and a quality control workflow to support high throughput image analysis.
The software tolerates uneven or noisy image background. It reduces labor remarkably and speeds up the analysis process. Implementing the software is beneficial for 3D tumor steroids to become a routine in vitro model for drug screens in both industry and academia.
Sphe sizer requires MATLAB with the signal processing and image processing toolboxes. An optional toolbox to use is parallel processing downloads OID sizer from the Rutgers server and unzip it to an installation directory, which will be needed in subsequent steps. Now prepare your images.
First, determine the image resolution of the imaging system and the absolute scale of the image in microns per pixel. Ultimately, the scale of each image in the experiment must be the same. Be sure to convert any proprietary image file formats to an accepted file format, such as TIFF or jpeg.
Now, save the files using specific file names and directory structure required by the software file names must be named by plate, row, and column file directories must be arranged as one experiment per directory with a single subdirectory for each time point in the experiment. Every image from every plate at a time point should share a subdirectory. Begin with opening MATLAB and in the command window orient to the installation directory of sphe sizer.
Then enter Sphe sizer one zero and hit return to launch this OID sizer program. Now click the browse button in the new window and select the directory prepared with all the images from the experiment. Then under the folder text field, select the include sub folders toggles for quality control.
Select the on the fly display option. The next step is to specify the resolution of the images, which must be done for the program to correctly convert pixels to millimeters. At this point, the user can enter the advanced menu to access more settings in this special color box.
Leave it at none for eight bit or 16 bit images. If the images are captured at 12 bit. Choose 12 bid.
After all the configurations are properly set up, start the computation by clicking compute as the computation is being carried out for the entire set of images. Individual segmentation results are displayed on the software on the fly when computation is complete, the results table displays the folder file volume, length, width, and a valid toggle for all the analyzed spheres. The valid toggle is for quality control.
When the user clicks on any cell in the results table, the original and the quality control images will display on the right side for review, examine all the images sequentially using the down arrow on the keyboard. When a suspicious spheroid boundary is encountered, refine it by clicking on manual initialize on the original image. Drag the ellipse tool to cover the steroid more accurately.
The result table will update the measurements accordingly. If the manual initialize fails to find the optimum boundary of the desired steroid, then use the hand draw button to display the original image and trace the boundary of the steroid. If the image does not contain a valid steroid, then simply uncheck the valid toggle for that image.
When quality control is completed, click on the format results box to save the results. The exported file names can be configured in the advanced to option window. The list output file is a tab delineated table that can contains all of the measurements.
The format output file is a tab delineated table and organizes the volume value into the original plate format For easy, visual and downstream data analysis. Sphe sizer can detect the boundary of the spheres on images with robustness even in various unfavorable image conditions. Occasionally, improper detection of this OID occurs then the manual initialized tool works by allowing the user to properly define the location and size of this steroid manually.
In extreme cases, the steroid may not be correctly segmented from distracting and noisy background automatically or with the manual initialized tool. In such cases, the program allows the user to hand draw the steroid boundary so as to generate accurate longitudinal and volumetric measurements. Efficiency of the software is validated by analyzing the same set of 288 images by manual analysis versus single core and multi-core computers.
Image analysis is over 18 times faster per image using steroid sizer than manual measurements to demonstrate repeatability of the software. A set of 24 steroid were analyzed three times by each method. Steroid sizer shown in green displayed smaller standard dation in measurements than manual analysis.
In an experiment made feasible with the help of steroid sizer tumor growth viewed as steroid was tested with HSP 90 inhibitor and cladribine treatments. The results showed that a combined treatment may have an anti-tumor effect in vivo. This study presents a fast, flexible, effective program for accurate determination of the size of 3D tumor steroids.
The ster size program is easy to use and it requires minimal user input. While attempting this procedure, it is important to remember that the steroids are imaged in the center of the field without touching the edge of the wall. All the images should be captured under the same objective and that all the files are correctly named and the arranged as indicated in the protocol.
