Biomedical Information Processing and Visualization for Minimally Invasive Neurosurgery

Introduction

Neurosurgery is concerned with the diagnosis, treatment, and post-surgical rehabilitation of disorders that affect the nervous system. Medical information processing for image-guided neurosurgery has been used for both diagnosis and therapy. Surgeons can perform precision tumor resection to an accuracy of millimeters using a combination of position-tracked surgical instruments and diagnostic image guidance, such as magnetic resonance (MR) and computed tomography (CT) images. However, these traditional image setups do not allow the surgeon to directly view the surgical target in minimally invasive surgery.

Surgical navigations have been developed to provide surgeons with an augmented reality environment, which enables safe, easy, and accurate surgical diagnosis and therapy. Minimally invasive surgery requires accurate and precise image-guided surgical navigation based on pre-/intra-operative imaging. However, traditional image-guided navigation can have an error of as much as a few millimeters due to inaccurate registration of the pre-/intra-operative images (Gholipour et al., 2007; Liao et al., 2010), especially for the intra-operative issues, such as operation-related deformation and small-tumor detection.

Addressing the registration issue of intra-operative tumor and tissue deformation is a surgical navigation challenge. Although intra-operative images, such as ultrasound, can assist in the identification of target and critical areas in real-time, image resolution limits diagnosis and treatment accuracy during the operation. The use of open magnetic resonance imaging (MRI) allows for intra-operative identification of the remaining tumor tissue. However, it is difficult to resect complicated tumors with MRI guidance alone; imaging does not precisely delineate the tumor boundary or distinguish between tumor and healthy brain tissue. Finding a way to provide accurate intra-operative imaging for the remaining tumor issue or small tumors is a challenge. Magnetoencephalography (MEG) (Cohen, 1968) and functional MRI (Ogawa et al., 1990) are helpful in delineating the eloquent areas of the brain by means of functional brain mapping techniques (Muragaki et al., 2001; Fukaya et al., 2002). Although combining these techniques can provide better spatial and temporal resolution, an improvement in real-time information acquisition is required.

One technique that can address this is to illuminate brain tumor tissue with a fluorescent marker (Regula et al., 1995). For example, the use of 5-aminolevulinic acid (5-ALA)-induced protoporphyrin IX (PpIX) fluorescence for visualization of malignant glioma tissue enables more complete tumor removal (Ciburis et al., 2003). 5-ALA is the first compound in the porphyrin synthesis pathway, which leads to hemoglobin production in mammals. Gliomas can be detected with this method because this acid is a natural biochemical precursor of hemoglobin that stimulates synthesis and leads to fluorescent porphyrin accumulation in various epithelia and cancerous tissues. The fluorescence is used to identify residual malignant glioma intra-operatively, thus improving surgical treatment accuracy (Rossi et al., 1996; Stummer et al., 1998; Stummer et al., 2000). By utilizing fluorescence illumination in microscopic surgery, the surgeon can resect the tumor more accurately (Leblond et al., 2009). However, the combination of positioning accuracy in diagnosis and tumor ablation accuracy in treatment is required to resect tumor tissue safely and precisely.