Thrust Rational, Organization and Strategy: In Thrust 2, we are developing systems that can perform the large and rapidly growing family of minimally invasive surgical interventions that involve the planned placement of some linear surgical instrument. Examples include needle placement procedures (aspirations, injections, local ablation therapies, and brachytherapy), external beam radiation therapy, and orthopedic implants and resections. The majority of these interventions today are performed percutaneously (i.e., across the skin), but a rapidly growing variety of these procedures are also deployed through alternative access routes from within body cavities (rectum, sinus, throat) and the vascular and respiratory systems.

These procedures can be represented in a model that is analogous to industrial manufacturing systems: if the right information is available, they can be planned ahead of time and executed in a reasonably predictable manner. (This contrasts with the strategy for Thrust 1, which involves interventions that are highly interactive, difficult to plan with precision, and where the engineered system acts with a constantly changing level of autonomy.) We, therefore, have classified them as Surgical CAD/CAM systems, having three key concepts: 1) Surgical CAD in which medical images, anatomical atlases, and other information are combined preoperatively to model an individual patient. The computer then assists the surgeon in planning and optimizing an appropriate intervention; 2) Surgical CAM in which real time images and other sensor data are used to "register" the preoperative plan to the actual patient, and the model and the plan are updated throughout the procedure. The physician performs the actual surgical procedure with the assistance of the computer, using appropriate technology for the particular intervention; and 3) Surgical TQM (total quality management) which reflects the important role that the computer can play in reducing surgical errors and in promoting more consistent and improved execution of procedures. Successful procedures are also included in procedural statistical atlases and fed back into the system for pre- and intra-operative planning.

The main strategic focus of the thrust has been to implement Surgical CAD/CAM paradigm in the context of needle placement & external beam therapy procedures that share numerous commonalities, irrespective of the imaging modalities, target organs and procedural details required of a specific intervention. In particular, we are developing prototype applications and components to prove the viability of the two fundamental engineering concepts of our approach to Surgical CAD/CAM: (1) One stop shopping: We are developing prototype systems that can integrate and optimize the entire spectrum of a minimally invasive procedure, from pre-operative planning through execution, assessment, and follow-up. Thus, the patient need not return to a medical facility several times, as is presently the case, to complete the various stages of the procedure. We are designing systems that can perform the entire range of activity with the speed and convenience comparable to current outpatient diagnostic procedures. (2) Plug and play: We are developing a modular family of multi-purpose and factorable systems that are invariant, to a reasonable extent, to the actual imaging modality, target organ, and procedural details. This will enable us to apply the systems to a broad range of clinical conditions by interchanging components relatively quickly and easily, with predictable and certifiable performance.

Initially, we have been focusing on systems for solid organ therapy (prostate, liver, and spine/bone), but will be extending these results to other organ systems. The goal is to achieve financing for all of our Surgical CAD/CAM clinical testbeds from sources other than NSF core funding. Doing so will enable us to concentrate core NSF funding on basic technology development affecting multiple synergistic clinical applications. We have been benefiting from NIH grants and leveraging industry partnerships with a growing number of SBIR and STTR grants. We are placing special emphasis on partnership with companies that own advanced local therapy technology that can benefit from image guidance, new delivery devices, or system integration. We currently have four STTR/SBIR grants of this type. We are also pursuing partnerships with medical imaging companies, to integrate their advanced imaging technologies with therapy planning and delivery. Three such projects, conducted in partnership with Siemens, are: an image overlay system incorporating a CT scanner; a strain imaging system using a high-end ultrasound scanner; and a bone imaging system using a C-Arm.

Thrust 2 has five tasks. Task 2.1 concentrates on system integration and the deployment of various system embodiments in clinical testbeds and human trials. As we are delivering more systems and applications to clinical trials, this task has grown to be the largest of Thrust-2. Task 2.2 is concerned with image analysis and data fusion. This covers robust and automated identification of structures in a variety of medical imaging modalities, as well as the fusion of various medical imaging modalities and sensory inputs together. Task 2.3 focuses on the study of deformity of biological structures, in the context of intra-operative adaptation of surgical plans as the targeted structures are deforming during the delivery of the procedure. Task 2.4 is concerned with the physical interface with the patient: interventional devices, surgical robots, intra-operative imagers, sensors, and appropriate combination of these implemented in needles catheters, and other minimally invasive surgical tools. Task 2.5 concentrates on three-dimensional tomographic reconstruction and registration of two- and three-dimensional image data.

Key milestones for Thrust 2 are identified in Figure B-3 and a full set of milestones are listed, by task, in Volume II. The key faculty for each task are identified in Table 2.

Relationship to the broad strategy and other thrusts: Thrust 2 aims to model larger segments of minimally invasive surgery processes, while Thrust 1 aims to enhance micro, and some macro, elements of interventions. Accordingly, the two thrusts have begun to converge in multiple ways. One such course is using a mixed model for surgical execution. An example is the “Robot-assisted fine needle placement in rodents” project, in which the needle placement robot is deployed with both image-based pre-planned control and steady-hand like cooperative force control. This enables us to pre-plan and execute a “gross access” to a particular surgical site with Surgical CAD/CAM techniques, and then a finer and more interactive local manipulation can take place with the use of Surgical Assistants techniques. This approach is being prototyped in the percutaneous liver surgery testbed, where the plan is to place a percutaneous ultrasound probe needle with Surgical CAD/CAM guidance into the liver, and then to apply image mosaic and other interactive imaging techniques for local guidance and monitoring.