An expert system for neurosurgical treatment planning

Abstract

In stereotactic neurosurgical practice, a very critical concern for a neurosurgeon is to excise a tumor with minimal damage to the patient after the operation. This puts extra demands on the pre-operative planning. Software development done correctly can assist the surgeon in this daily treatment process. Patients who can benefit directly are those with brain tumors requiring biopsy diagnosis and tumor excision. A solution is to provide the surgeon with three dimensional image visualization together with an expert system which calculates the consequences of the simulated operation. Knowledge acquired to finish this research includes neurosurgery, neuroanatomy, medical imaging, image analysis and expert systems. Different software tools have been written to facilitate the practical aspects of this system. The three-dimensional brain models can be reconstructed from two-dimensional medical images in the format of either photographic film or digital data from scanning machines such as Computed Tomography and Magnetic Resonance Imaging. Image processing techniques such as thresholding, filtering and contouring are performed for special-feature segmentation and the output coordinates are transferred to the developed expert system for path planning. To derive the safest probing path from the clinical data, a new method of path planning using the risk bearing regions and the path length has been developed and implemented. The inferencing rules and knowledge base are designed with inputs from experienced neurosurgeons in Hong Kong. The results displayed using three-dimensional projection method can help surgeons to assess the surgical risks. Four clinical cases with different types of tumors and image formats were used to validate the system.