145 - Developing an Algorithm for Complex Zygoma Surgical Reconstruction Following Trauma

Poster Board Number: 145
There are separate poster presentation times for odd and even posters.
Odd poster #s – first hour
Even poster #s – second hour

Abstract Body : Intro/Objective: Technological advances can aid in the assessment, planning, and management of complex craniomaxillofacial injuries. At University of Colorado Hospital, the current approaches to planning zygoma fracture repair is limited to viewing virtual 3D models constructed from computed tomography (CT) scans. These only rotate in the x-plane and individual bones cannot be isolated from the full head CT. Viewing models without the ability to manipulate bone fragments to confirm proper repair sequences causes physicians to rely heavily on visuospatial abilities, which is individually variable. Errors in visuospatial planning can lead to errors in the proper sequence of fragment manipulation. These errors cause surgeons to re-manipulate fragments during the surgery, causing longer exposed patient surgical time, more anesthesia, and even wasted hardware. The objective of this study is to develop and evaluate an algorithm for the proper sequence of fragment manipulation of the zygomatic bone in complex fractures following trauma.

Methods: DICOM imaging from previous University of Colorado Hospital patients with craniofacial fractures is used as the source data. The fractured zygoma is segmented from the full head CT using crop volume and a threshold technique in 3D Slicer. The model is labeled, and the sequence of manipulation is determined for that specific model given the developed constraints. Our approach starts with developing novel zygoma categorization based on regions of the bone affected during the traumatic event. First, fractures along suture lines are labeled, as these are a great source of stability. Secondly, the bone is divided into labeled quadrants based on the crossing of the maxillary buttresses through the body of the zygoma. Using these subdivisions and labels of the zygoma, a sequencing hierarchy can be established based on stability and exceptions. The first exception is based on surgical incisions sites. Next is the geometry of the bone. Deeper repairs need to be manipulated before superficial fragments. Fragment edges also need to be aligned with each other. Furthermore, any fragments with irregular, centralized projections need to be manipulated prior to neighboring, flat edged fragments, or else there is a risk of the irregular fragment getting left out.

Results: Our results show that the order of reconstruction should be based on (1) stability of the bone based on intact suture lines, (2) fragment accessibility based on incisions, (3) fragments closest to the buttresses, and (4) irregular projections of fragments. The later rule overrides the previous rule when present.

Conclusion: The successful completion of our process will derive the constraints and patterns for digital reconstruction of facial fractures prior to surgery. These results will lay the foundations for generating the surgical reconstruction sequencing using an automated system.

Significance: Developing these constraints is a step closer to the automation of craniofacial surgical planning through machine learning. With the successful development of an algorithm for zygoma reconstruction, the exposed patient time will reduce during surgery. Sequence error incidences will also decrease using our methods. A prospective clinical trial can compare these methods to the sequences a trained plastic surgeon would use in the same patient. These methods allow for a plastic surgeon to confirm their visuospatial-derived surgical plan prior to exposing the fractures.