Abstract
Purpose :
Deep anterior lamellar keratoplasty (DALK) challenges surgeons to insert needles precisely into the corneal stroma despite limited depth visualization through the operating microscope. We demonstrate improved performance in ex vivo DALK needle insertions using a cooperatively-controlled robot assistant that stabilizes needle insertions while using live optical coherence tomography (OCT) with real-time cornea segmentation to monitor needle depth.
Methods :
For cooperative surgeon-robot needle control, we used an IRB 120 robot arm (ABB Robotics; Zurich, Switzerland) with a Gamma force torque sensor (ATI Industrial Automation; Apex, NC). The robot moved only in response to surgeon force input through the needle handle (Fig. 1). For corneal OCT imaging, we used a telecentric scanner and a 100 kHz swept-source laser centered at 1060 nm (Axsun Technologies; Billerica, MA). A monitor displayed live cornea segmentation and a synthetic B-scan along the tracked needle shaft to the surgeon. For efficacy analysis, we completed a user study in which operators performed needle insertions into human cadaveric corneas. Operators performed needle insertions either manually, using a microscope and a syringe-mounted needle, or robot-assisted (cooperatively), using OCT and the robot-stabilized needle. Operators alternated between manual and robot-assisted insertions over two corneas at eight insertions per cornea. We provided coaching on DALK technique and robot assistant use during the initial eight insertions. For the non-training insertions only, we analyzed for differences in mean final needle depth for perforation-free trials using ANOVA and perforation rate using Fisher’s exact test.
Results :
Three operators performed 24 total needle insertions and demonstrated a mean insertion depth of 39% manually and 71% cooperatively, excluding perforations (Fig. 2). The perforation rate was 2/12 manually and 3/12 cooperatively. The difference in 32% of corneal depth was statistically significant (F(1,13)=6.7, p=0.022). There was no statistically significant difference in perforation rate (p=1.0).
Conclusions :
Operators using our cooperatively-controlled DALK assistant achieved an 84% improvement in perforation-free needle depth, a significant factor in pneumodissection success.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.