Optimized resection planning and novel tool development for bone tumour surgery

Correct surgical treatment of a bone tumour requires that all the diseased tissue be removed from the patient, in one piece, along with a cuff of healthy tissue surrounding the diseased mass to ensure no cancerous cells remains in the surgical site. To achieve this, a surgeon will make cuts in the bone around the tumour. The position and alignment of these cuts are specified in a pre-operative planning stage, where medical images can be used by a clinician to decide where each cutting plane should pass through the bone, its distance from the tumour, and whether it is possible to preserve any nearby critical healthy tissue. This resection plan can also be used for the development of a reconstructive solution for the affected region, which returns function and structure to the compromised anatomy, and improves the post-operative quality of life to the patient. Preparing these resection plans can be a technically demanding and time-consuming process, with only limited assistance available to clinicians through software and virtual tools. Positive treatment outcomes are heavily dependent on good quality resection plan, which itself relies on a clinician's experience with similar cases, the availability of medical imaging software capable of planning a bone tumour resection, as well as the user's familiarity with that software. Measurements can be performed on a resected mass post-operatively to check whether the disease has been completely removed from the surgical site, and the consistency between a resection plan to how it was performed. However there are no quantifiable measures of a resection plan itself, and the quality of a resection plan is largely a subjective assessment made by the surgeon, based primarily on whether the approach will result in removal of all diseased tissue. The absence of a standardized measure of quality for a resection plan, the aggressive nature of malignant bone tumours, and the importance of timely treatment limits a clinician's opportunity to consider multiple resection plans for a single patient. Additionally, proposing elaborate resection plans to preserve critical anatomy is a risky proposition, as novel surgical approaches may result in some unforseen complications, meaning resection plans similar to others with a history of success are deemed less likely to fail, and are therefore the typical treatment approach. Recent technological advances of additively manufactured patient-specific implants and robot-assisted orthopaedic surgery are on the cusp of being integrated into the bone tumour surgical treatment workflow, and may provide both intraoperative and post-operative improvements. However, in order to maximize the capabilities of these technologies, the planning stage must be standardized such that the selected surgical approach can be quantified, and the plan specifying the intraoperative execution and reconstructive solution is one which provides an optimal post-operative treatment outcome. Furthermore, proposing any novel surgical approach would require the development of specialized tooling to facilitate the optimal treatment. This thesis reports the research performed in quantifying bone tumours and their resection plans; a method for automatically generating planar and elaborate resection plans optimized for minimum healthy bone loss; and novel tool designs to assist in performing the proposed optimized resections. A selection of 20 anonymized osteosarcoma and chondrosarcoma within the femur and tibia are used to develop consistent and comparable measurements of bone tumours, whereby a tumour's size is measured volumetrically, and its infiltration and shape are measured as a percentage of that volume. In collaboration with an experienced surgeon, a resection plan is manually prepared for each tumour case, using software tailored for bone tumour resections. A measurement is proposed for the manually-prepared resection plans, independent of the tumour size, serving as a measure of the resection plan efficiency. Using the bone waste value from the manually-prepared resection plans as a baseline for each case, two general methods of optimizing resection plans are developed, the first for planar cuts, the second for more elaborate, non-planar geometries. A stochastic optimization approach (particle swarm) configures the position and alignment of a resection plan defined by the tumour geometry data, by minimizing healthy bone collaterally removed with the tumour (bone waste). For planar resections, a validation check on the alignment of the cutting planes ensures all resection plans can be extracted from the bone. Both methods are found to improve upon the bone waste of the manually-prepared resection plans. More generally, it is found that the more a resection plan conforms to the curvature of the tumour, the lower the bone waste, and that using more than five cutting planes results in diminishing returns to bone waste. To assist with the treatment workflow, a system of selecting the optimal surgical approach is described where each resection plan is scored by its intraoperative complexity and the benefit provided to the patient. Factors of each resection plan such as the number of cutting planes, the accessibility of each cut, and the ease of implanting the reconstructive solution contributed to a two value score, allowing for an objective comparison of multiple resection plans for a single tumour. Finally, two novel tool designs are investigated to facilitate curved and planar cuts underneath a tumour. First, a tool is presented with a 2 degree-of-freedom kinematic structure in which the cutting tip of a robotic linkage is bound to the surface of a parametrically-defined torus. Second, an investigation into cutting behaviour of wire saws is conducted, and a prototype tool for performing a reciprocating sawing action underneath a volume of material is proposed. The future direction of this work will focus on optimizing resection plan generation, development of novel resection geometries, application to tumours in different anatomy, integration with surgical robotics and additive manufacturing, and clinical evaluation of the optimized resection plans.