Automatic generation of vascular risk map for trajectory planning in stereotactic interventions

It is an important issue to reach the target point in the brain safely. It is aimed not to enter critical brain structures during the traction (entry-target). In classical planning, brain structures (AC, PC, MC), entry and target points are marked manually by the surgeon. This time-consuming and high-risk process needs to be supported by computer-aided software. When the studies conducted in this field are examined, it is seen that there is no targeting system where brain structures (brain, tumor, vascular network) are detected and the entry risk map is produced autonomously. In addition, a unique traction risk calculation model is proposed. This model converts the two-level vascular network to gray scale according to the distance to the boundary and regulates the effect of the boundary and the center point in the vascular structure on the risk. This update in the vascular network provided a more accurate risk map. HD-BET and SwinUnetr models were used respectively in the detection of brain and tumor structures. Manual traction data of two different neurosurgeons were used in the verification activities of the study. It is seen that the proposed automatic traction detection approach can determine tractions with approximately 50% lower risk values compared to surgeons. In the next study, risk maps will be produced by targeting STN structures instead of tumors and the aim will be to increase the safety of epilepsy operations.