BACKGROUND: Preoperative three-dimensional (3D) planning has become the standard of care for complex orthopedic procedures. Quantitatively evaluating whether the preoperative plan is successfully performed remains a challenge due to the lack of standardized measurements, which makes comparison across studies difficult. Therefore, standardized measurements are required to close the feedback loop to improve upon our procedures and achieve safer or closer resections.

OBJECTIVE: This present study investigates the best suitable standardized approach for evaluating the procedural accuracu of the achieved plane compared to the preoperative plan, and to assess the achieved results in complex orthopedic procedures, with a main focus on bone tumor resections.

DESIGN: This single-centre retrospective cohort study included 10 patients who underwent complex orthopaedic procedures (7 oncological, 3 non-oncological), between 2021 and 2024, with a total of 28 cutting planes in patients and 17 in allografts. All patients had a 3D planned surgical approach with digital visualization and patient-specific instruments (PSIs). The achieved planes were determined using four methods (mesh, normal vector, manual and point cloud) and the procedural accuracy was assessed by multiple linear and angular outcome measures (distance between points, center of grafity (COG)-to-COG, COG-to-plane, angle between normal vectors and pitch and roll angles).
RESULTS: The mesh and point cloud methods were superior in terms of ease of use and objectivity, with the point cloud method being the most accessible due to the lack of segmentation requirements. The point cloud method performed comparably to the mesh method for linear and angular deviations, with a mean deviation within 0.8 mm (millimeters) and 0.2 degrees for roll and pitch angles in patient cases. The average flatness according to the corrected ISO-1101 standard was 2.5 mm, and a surgical margin difference of 1 mm was observed. Significant differences were observed between the COG-to-COG approach and other linear approaches for both methods, and between the corrected and non-corrected ISO-standards for the mesh method.

CONCLUSION: The point cloud method appeared to be a good alternative to the mesh method for identifying the achieved plane in bone (tumor) resections. Furthermore, the average outcomes observed in this study provide a useful baseline for assessing procedural accuracy in complex orthopaedic procedures using 3D planning and computer assisted surgery (CAS). Further research with larger sample sizes is needed to validate these findings.

KEYWORDS: bone resection, computer-assisted surgery, linear deviation, angular deviation, cutting plane, procedural accuracy

Disease of bone or trauma can result in significant bony deformation. Functional impairment as a result of such deformity is an indication for surgical correction. Indicated surgery is often closed wedge osteotomy and internal fixation. Manual design of a preoperative plan using computer-aided design software is the current golden standard for complex correction surgery planning. However, these manually designed plannings are time consuming and subjective. The goal of this study was to develop an algorithm capable of generating good quality preoperative plans for correction surgery, in less time than an expert in preoperative planning.
A software solution integrated in Blender (Blender Foundation, The Netherlands), has been developed to generate semi-automatic preoperative plans. The software optimizes one to three closed wedge osteotomies, by minimizing the dissimilarity from a deformity shape after osteotomy, to a target shape. The deformity shape is the centreline of the deformed femur. The target shape in this study is the centreline of an intramedullary nail.
A cost function was created that makes use of the Hausdorff distance and the root mean squared error to quantify the dissimilarity between the deformity shape after osteotomy and the target shape. Clinical constraints to the model are femoral length and collum anteversion angle. Using a multi-objective genetic algorithm the cost function is minimized within the bounds of the set clinical constraints.
To validate the method of semi-automatic generation, 20 bone models of deformed femora have been created from retrospectively collected CT data. The proposed solution was used to generate preoperative plans, which were scored by two independent assessors. After the creation of an initial preoperative plan and a maximum of two revisions of the plan, 18 of 20 semi-automatically generated preoperative plans were eligible for surgery. Four cases of femoral deformity were randomly selected from the dataset and counterparts to the existing automatic preoperative plans were created manually. 30% to 575% more time was needed to design a preoperative plan manually compared to the semi-automatic generation. In all preoperative plans in the automatic group, the collum anteversion angle of the postoperative configuration was within the normal range of 8-15 degrees. In the manual group, one of four postoperative configurations had a collum anteversion angle within this normal.
The proposed method for semi-automatic preoperative plan generation is a novel, versatile approach with the ability to optimize multiple osteotomies to mimic a given target shape. The automatic planning tool is a promising aide to both technicians and clinicians for a fast preoperative planning withing the boundaries of clinical anatomical normal ranges.