Virtual Environments for Training in Surgery

Contact person: Frank Tendick (frank@milo.berkeley.edu)

Overview

Training in surgery is traditionally based on an apprenticeship model. Students learn by watching and participating in cases, taking greater roles with each procedure. Although this can be an effective method of teaching, it has great drawbacks in an era when techniques and technology change rapidly, yet economy in health care is of increasing importance.

Training in the operating room can increase risk to the patient and slows the operation, resulting in greater costs. It also has drawbacks in teaching effectiveness. A stressful environment can reduce learning, and students are not free to experiment with different techniques to see which one might be best for them. Because every mentor teaches his or her own technique, it is impossible to develop standards for training or assessment.

Other methods of training have limitations. Books are not interactive and cannot portray anatomy in three dimensions. Cadavers and live animals are expensive and usually cannot demonstrate pathologies. Animal anatomy is not the same as human anatomy. In vitro training models made of synthetic materials can be useful, but it would be difficult to maintain a library of models with all important pathologies and anatomical variations, especially if the models are of little use after being ``dissected.''

Computer-based training has many potential advantages. It is interactive, yet an instructor's presence is not necessary, so students may practice in their free moments. Any pathology or anatomical variation could be created. Simulated positions and forces can be recorded to compare with established performance metrics for assessment and credentialing. Students could also try different techniques and look at anatomy from perspectives that would be impossible during surgery.


Projects

1. Navigating in a Virtual Surgical Model

In minimally invasive surgery of the abdomen, surgeons watch a video image from a rigid lens endoscope inserted through an incision in the skin. The objective lens of some scopes is fixed at 30 or 45 degrees from the scope axis, so that the user can peer over or around objects as with a periscope. Often an assistant holds the scope, and surgeons are unfamiliar with this capability or how to use it to best advantage.

This project will include developing a virtual environment made of objects of various geometries representing organs. The user will hold a special 4 degree-of-freedom "joystick" with a fulcrum as if orienting the scope through an incision. He or she must navigate through the environment to look for hidden targets. A second "joystick" representing a grasping instrument might be used to open trapdoor-like folds of tissue to reveal some of the targets.

At the end of the semester, we will use this simulation in a course at UCSF where practicing surgeons come to learn minimally invasive techniques. If we can show that training in a VE transfers to improved performance in vivo, this work will be quite exciting and will make a significant publication.

2. Understanding 3-D Surgical Anatomy

A student or practicing surgeon must develop mental 3-D models of typical and patient-specific surgical anatomy from 2-D textbooks or from narrow 2-D images obscured by fat, scar tissue, and connective tissue. Even an experienced surgeon may have an inaccurate mental model he or she has used in 100's of cases, only to find that it fails disastrously in an unusual situation (s)he has not encountered before. We would like to be able to train for such situations, just as pilots train for events that happen rarely but must be quickly handled when they do occur. Three projects relate to understanding anatomy in 3 dimensions:

a. Interactive Viewing of Complex Anatomy. Navigating through a 3-D model of a complex organ or anatomical region, such as the eye. Perhaps this could include interacting with the model, e.g., taking it apart, activating muscles, or looking through it as the optics changed.

b. Misunderstanding the anatomy of ducts between the liver and gall bladder can lead a surgeon to cut the wrong duct when removing the diseased gall bladder. This can potentially lead to a liver transplant or death. A relatively simple model can be developed which must be "dissected" by removing small chunks of fat. The user would be asked to identify the 3-D configuration of the ducts after removing as little fat as possible.

c. I have an MRI data set of a head, both in slices and with volume interpolation. A project could be designed to segment brain parts, reconstruct them in three dimensions, and/or render surfaces from the segmentation with interactive viewing.

3. Modeling Deformable Tissue

It is obviously desirable to have soft tissue capable of deforming in a natural way in a simulation. Unfortunately, finite element methods for accurate modeling are orders of magnitude too slow for real time. Several computationally efficient methods exist to create deformable objects. It is possible that they behave realistically enough for training some aspects of surgery. This project would include implementing and comparing models for efficiency, visual realism, and haptic realism in modeling interaction forces.

4. Suturing

Knot tying is a difficult skill in microsurgery or minimally invasive surgery. I think it would be possible to model tying a knot by treating the suture as a spline between endpoint constraints. Changing topology as the suture was knotted or in contact with the non-grasping instrument would be handled as a sequence of states. The instruments would be modeled as infinitely thin rods to avoid problems with collision detection.


WWW Maven: Dan Garcia (ddgarcia@cs.berkeley.edu) (finger me) Send me feedback

(CS 294)