Jinling Wang, PhD1, Katherine S Lin, MD2, Keith A Watson, MD2, Caroline G Cao1. 1Wright State University, 2Miami Valley Hospital
INTRODUCTION: Robot-assisted hysterectomy is becoming more common due to its advantages over traditional laparoscopic surgery, such as an ergonomic working posture for the surgeon, a 3D view of the surgical site, and seven degrees-of-freedom in motion. One of the essential and critical steps in robot-assisted surgery is port placement. Well-positioned ports can avoid collisions between instruments, and provide adequate reach and visualization of the surgical site. Several factors must be considered for proper port placement, including the type of surgical intervention, location of the target anatomy, access to the target organ, size of the patient, size of the robotic arms and surgical tools, the physical coupling between the patient and patient cart, and prior surgical history. However, the guidelines provided by the manufacturers are only for a generic patient and the most common procedures. Patient-specific information is needed to assist surgeons in port placement. To develop a tool for patient-specific, optimal port placement, it is necessary to build a 3D model of the patient’s abdomen for before/after insufflation. A database of patient abdomen dimensions will help to develop an algorithm to predict the change of shape after insufflation thus aiding pre-operative planning. Towards this goal, this study devised a method to build a patient-specific abdominal model.
METHODS AND PROCEDURES: The procedures of this method are as follows: (1) 25 markers were drawn on the patient’s abdomen around umbilicus in a 5×5 matrix with the markers 2 cm apart; (2) the Artec 3D scanner was used to image the surface of the abdomen; (3) the coordinate of each marker was obtained from the scanned image; and (4) a cubic interpolation method was applied to generate a 3D model.
RESULTS: Data were collected on two patients both before and after insufflation (4 sets of scanned data). Four models were built using the proposed method. To test the accuracy of the 3D models, 32 points were randomly selected in each model. The coordinates of these 32 points from the 3D scanner were compared with the ones calculated by the generated model. The mean errors and standard deviations for the four models were 1.06±1.31mm, 1.00±1.69mm, 0.57±0.54mm, and 0.30±0.33mm. During data processing, step (3) is subjective since a human operator must select one point from multiple candidate points in the scanned data for each marker to obtain the coordinate. To verify the reliability of this method, two investigators processed the 4 sets of data independently while one of them did it twice. The correlation results show that both inter- and intra-rater reliability have a correlation coefficient of greater than 0.99.
CONCLUSION: Preliminary results show that the models are highly accurate. This proposed method can be applied to build a database for the 3D abdominal model and help with the development of the port placement tool in the future.