Abstract Winner – Robotic Technology & Virtual Assistants

Background: Heart catheterizations rely almost exclusively on fluoroscopic imaging for guidance which exposes the operator, staff and patient to harmful radiation. Human error is also an important source of morbidity and mortality in cardiac catheterization. The ultimate goal of this project is a self-directed robot (CATH-BOT) capable of performing remotely supervised cardiac catheterizations without ionizing radiation.

CATH-BOT will create a patient specific 3-D anatomical model by accessing all recent pre-procedural imaging within the hospital electronic record system (MRI, CT, ultrasound, etc) as well as uploaded images from outside sources. The patient-specific 3-D anatomical model is co-registered (X,Y,Z planes) with the patient via real-time imaging immediately prior to the procedure. Temporal syncing is performed via respiratory/ECG-gating. Catheter location and course are continuously tracked via an electromagnetic tracking system (EMTS) which is also co-registered to the real-time patient anatomy. Thus, CATH-BOT is aware of real-time catheter position in relation to patient anatomy, without the need for fluoroscopy.

Catheter movement is accomplished using a robotically steerable catheter (Sensei Robotic System). CATH-BOT will have a vast database of pre-programmed catheter manipulation algorithms (CMAs). CMA selection is achieved through integrating catheter position, desired catheter location, and real-time patient anatomical map. Machine learning will be employed to optimize CMA selection. Safeguards will include redundant force sensors, mechanical restraints and predefined acceptable catheter zones. Future integration of computational creativity and deep learning via artificial neural network will raise the level of catheter manipulation to that (or above) the human operator, all without radiation. Traditional fluoroscopy can be used to supervise CATH-BOT as necessary, however the ultimate goal is procedural independence.

The first step in CATH-BOT creation is the development of the EMTS. We tested our patented prototype EMTS in a phantom model with human/EMTS guidance versus human/fluoroscopic guidance as a gold standard.

Methods: An image loading engine was created to load and rescale DICOM images to an isotropic volume. An Aurora™ magnetic tracking system (NDI, Waterloo, ON) was used with a 0.5M cubic field in conjunction with a 6-DOF sensor-embedded catheter. Validation was performed utilizing a Procrustes affine rigid point based registration algorithm with an iterative closest point surface based registration. A simplified cardiac phantom was constructed to mimic the cardiac atria with a septal defect. Forty catheterizations were performed by a human operator using fluoroscopy, then repeated using EMTS. Times were compared by student’s T-test.
Results: All mock catheterizations were successfully completed. There was no significant difference between fluoroscopy and EMTS in time across the mock septal defect (Fluoroscopy = 7 sec, EMTS = 6.5 sec) while total procedure time was significantly lower using EMTS (Fluoroscopy = 25.7 sec, EMTS = 21.8 sec).
Conclusions: The EMTS prototype performed well and represents the first step of the CATH-BOT project. The next phase of development will integrate the robotically steerable catheter with direct catheter position feedback and development of the CMA database.


Author: Hamilton Baker

Coauthor(s): David Kwartowicz, PhD

Status: Work In Progress