Identification of cardiac structure via MRI for 3D printing
Principal Investigator: Dr Niall Haslam
Approved Research ID: 43384
Approval date: November 21st 2018
The aim of this project is to build a machine learning algorithm to identify anatomy in medical images for the purposes of 3D printing scale replicas of anatomy. To create 3D printed models the raw images must be annotated and the anatomy identified. These manual labels will be combined with the raw images and used to teach algorithms to identify the relevant anatomy in the image and create a 3D model of the anatomy from the 2D images. Medical 3D printing offers a host of benefits to service providers and service users across the healthcare continuum. Typically imaging from CT, MRI is viewed on 2D screens in black and white. In some cases 3D visualisations are rendered on the screen to provide the illusion of 3D volumes on a 2D screen. This is known to have problems with viewing angle, depth, transparency and lighting anomalies. 3D printed models offer advantages over screen based visualisations by allowing complex anatomical relationships can be better appreciated at scale, physically. Offering 3D printing as a part of a comprehensive care pathway for complex injuries facilitates procedural efficiency, improved treatment outcomes, and reduces downstream re-intervention costs, offering high potential value. Research shows time savings ranging from 27% to 40% on surgery where a 3D print has been used to prepare. This time saving not only has a knock-on impact on costs (less theatre time, only necessary equipment is sterilised, reduction in loan kits) but by impacts directly on scheduling efficiency, increasing the number of patients being treated. In addition, having access to the 3D printed 1:1 replica of the patient's anatomy supports physical simulations of surgery, as well as pre-bending and pre-fitting of plates prior to surgery. Research also shows that 'touch' re-calibrates the visual perception so that it is better able to infer depth from the retinal projection. Procedures with longer operating times, uncertainty, and risk of complications are those which will most greatly justify the financial and resource cost in creating 3D printed patient models, and will result in the greatest impact. In addition to a better prepared surgical team and increased hospital efficiency, the patient not only can benefit from likely improved outcomes, but patient understanding and satisfaction is increased by seeing and interacting with models of their anatomy. Clinicians report that increased patient understanding aids in informed consent discussions and facilitates improved patient cooperation in the procedures.