A 3D interactive simulation of cardiac ultrasound.
A series of parametric and patient-specific airway models was 3D printed and used as a variable-difficulty bronchoscopy training system.
A parametric airway model was obtained from an online medical model repository. Patient airways were segmented from anonymized patient CT DICOM images. The airway models were each hollowed with a 1 mm offset from the external surface. The parametric airway was separated into seven distinct regions: trachea, bifurcation, left & right bronchi and primary bronchi to upper left, lower and middle right lobes. Anatomical regions were printed with different colours using a fused deposition modelling 3D printer. Patient-specific airway models were 3D printed as unibody objects. Markers to distinguish different anatomical features were fitted and printed for one patient-specific model.
Parametric models of thoracic vertebrae 3 - 10 was downloaded from an online medical model repository. Vertebrae were positioned in free 3D modelling software, Blender™ and rudimentary intervertebral disks were designed between each vertebrae to ensure a congruent model for 3D printing. Free, anonymized scoliotic-patient CT DICOM images were obtained from osirix-viewer.com. Thoracic vertebrae were semi-automatically segmented using open-source medical segmentation tool ITK-SNAP. Models were imported into free 3D printing slicing software Cura™ and exported as a .gcode toolpath file describing speed, temperature and geometric details necessary for 3D printing.
The Aortic Root is a complex apparatus that is best described as a composite 3D structure. We developed a technique to render and 3D print the aortic root from 3D TEE datasets.
Anonymized patient CT DICOM data sets were downloaded from a free, online DICOM repository www.osirix-viewer.com/datasets. For this experiment, two DICOM sets were chosen: one with adequate resolution of the patient’s spine and the other focused on the patient’s upper airway. DICOM data was viewed in free medical segmentation software ITK-SNAP; a semi-automated, region-growth selection module was used to generate a voxel model of the patient’s thoracic spine and upper airway. These voxel models were exported as a stereolithographic (.stl) file type and opened in free 3D modelling suite Meshmixer (Autodesk, Inc., San Rafael, CA, US). Within Meshmixer, models were repaired and surface deformities were smoothed. Modified spine/airway models were prepared for 3D printing within the FLOSS program, Slic3r, as a 3D printer readable .gcode file. Models were then uploaded to commercial 3D printer Lulzbot Taz 5 (Aleph Objects Inc.) for fabrication.
We develop a low-cost ultrasound phantom based on cardiac Ct using 3D printing and ballistics gel. We then tested the quality and accuracy of TEE images obtained from ballistics gel model. Images from a CT dataset were segmented to isolate patient myocardium to create virtual 3D model of the heart and subsequently 3D printed with nylon as a mock-up. Commercially available synthetic ballistics gel was heated to 130 degrees Centigrade until liquefaction when the 3D printed model was submerged and the solution was allowed to cool to room temperature. Once solidified, the 3D print was carefully removed leaving the internal cast of the heart intact.
Read APIL publications and research in progress papers.
Learn 3D segmentation and printing techniques.