Osirix Imaging - Articles, Tutorials & Resources for Medical Illustrators

The region growing tool allows you to isolate structures out of a study to visualize it in 3D. A contrast enhanced structure is ideal to use with the region growing tool because of its density difference. Also, the lungs can easily be separated from other tissues with this tool.

Originally, I had worked on a tutorial on segmenting the colon with the region growing tool and the “Invert Data” plugin using multiple ROIs (region of interest) in a contrast enhanced CT scan of the colon. Unfortunately, because of the large size of the study and a lack of RAM on my MacBook Pro, I experienced numerous crashes and was unable to render the ROI in 3D.

Because of the processor and memory limitations on my computer, and without the 64-bit version of OsiriX, I decided to use a smaller study and a simpler ROI to segment the lungs. Chapter 4 in the first edition of the OsiriX the Pocket Guide was used as reference.

1. Download a study from the DICOM sample image sets . For this tutorial, I used TRAGICOMIX, a CT study of the thorax.

image via http://pubimage.hcuge.ch:8080/

2. Copy the study to your desktop. Right click and open the folder to unzip it.

3. To import the study into OsiriX, click the unzipped folder from your desktop and drag it to the Local DICOM database window. Or click the Import button and choose your folder from the desktop. You can also select File > Import > Import Files and choose your folder.

In the OsiriX Database dialog box, click on Copy Files.

4. Once the files are copied in to the Local Database, select a series from the study and click on the 2D viewer button to open it. I used the series A Aorta w-o 2.0 B10f 85%Smooth (images 1-261), 2mm CT slices of the thorax.

5. After the series opens in the 2D viewer window, select ROI > Grow Region (2D/ 3D segmentation).

6. In the Segmentation Parameters dialog box, select

3D Growing Region (entire series)

Threshold (interval) for the Algorithm, and

1000HU for the interval. (this number is similar to the tolerance level when using the magic wand tool in Photoshop; it lets you select more or less pixels in the image)

Change the ROI name if you have multiple ROIs in the same study or if you want to save and import the ROI at a later time.

7. Click once on an area of the lungs and click compute in the Segmentation Parameters dialog box.

The lungs in this study required only one ROI for the series, but other studies may require multiple ROIs. If so, animate the series and click on another area of the image and click compute and repeat this process until the ROI selection is complete. If you click on the wrong part of the image, simply click on another area before you click compute. Or if you already clicked compute and decide that you don’t need the ROI, you can deselect or delete ROIs in the series under the ROI menu.

8. To add the pulmonary vessels in the ROI, select ROI > Brush ROIs > Closing .

9. Set the structuring element radius to 10.

10. Animate the series to view the region growing segmentation applied to the entire stack of images.

11. Select ROI > Set pixel values to…

This will open up a dialog box so that you can set the pixels of everything outside the ROI to the value of air, -1024.

12. When I render this ROI as is, the 3D volume render doesn’t show anything. So under the ROI menu, I changed all ROIs of everything inside the ROI to -500. If you can suggest a better pixel value for the inside of the ROI, please let me know in the comments below.

13. Under the 2D/3D button, click on 3D Volume Rendering.

14. I adjusted the 16-bit CLUT editor of the 3D Volume Render. Also, a 3D surface rendering of the lungs can be used to export an .obj as a guide to model it in a 3D program like 3DS Max.

Esther Pulley is a recent graduate of the Biomedical Visualization program at the University of Illinois at Chicago.