First, I began by taking the coronal plane of my own abdomen/pelvis MRI scan (urogram protocol), and exported the images as .dcm (DICOM) files. The DICOM file format makes it easy to transfer and manipulate the images. 

I then imported the "raw" DICOM files into an open-source DICOM manipulation software known as "3D Slicer". After assessing both kidneys' appearances by making a MIP (maximum intensity projection) image, I elected to 3D print the right renal parenchyma due to its more homogenous enhancement. Additionally the left ureter was partially opacified, which would have introduced the possibility for error in segmentation.

The window level was changed so that the right kidney was more prominent than surrounding structures, after which automatic segmentation was attempted. The vasculature of the liver, however, resulted in messy data which would have delivered an unsatisfactory 3D print. To mitigate this, I manually placed an ROI around the perinephric space which was most visible on the axial slices. With the ROIs in place only the renal parenchyma enhanced, allowing me to more easily manually segment the kidney. 

Following fifteen minutes of manual segmentation, the right renal parenchyma was isolated. Each slice was inspected to ensure that there were no errors. Based on this segmentation, a .STL file (3D object file) was generated and exported. 

The 3D .STL file was imported into the 3D print pre-processing application known as "Voxelmaker". Within this software I inspected the 3D object, ensuring that it would print with a 15% infill, and providing supports where needed. I had to manually place supports in the renal pelvis. The program then divided the .STL file into thousands of layers that would be stacked inside the 3D printer in order to create a 3D object. 

Once the file was "sliced", it was then transferred to the 3D printer. Printing at 220 degrees Celsius, it took a total of six hours to print. Following the printing process, the model was actively cooled and removed from the print bed. The support material was removed without difficulty. 

The kidney is dimensionally accurate, and virtually identical in size to my actual right kidney. This was confirmed by manual measurement with a caliper versus the source MRI scan. 

This technique is easily reproducible, and I plan to employ it to 3D print 1:1 models of other organs, such as the brain. Of note, while it is possible to 3D print surface anatomy from both CT and MRI, it has been my experience that using T1W imaging from MRI has been most useful for soft tissue. CT on the other hand is more advantageous for osseus structures. 

Disclosure: This project was done to fulfill a "Hallmarks learning goal" for my undergraduate curriculum. I completed this project with no direct faculty support; I was however mentored by various professors and given access to faculty resources to make this project possible. I would like to thank the anatomy department at Thomas Jefferson University as well as the circulation staff at the Gutman Library for assisting with and influencing this project.