gms | German Medical Science

Objective: When working with TMS and object naming tasks, we receive both negative and positive TMS spots, widely distributed and in relation to the patients’ pathologies. The TMS effect extends approximately 20-30mm into the brain tissue, potentially directly affecting white matter pathways. The tractography’s sensitivity to connect with TMS positive areas depends on the applied tracking algorithm. In this study, we used advanced tractography techniques with complimentary probabilistic tractography algorithms. Subsequently, we analyzed its potential for visualizing connections to TMS positive and negative areas in comparison to clinical standard deterministic tractography.

Methods: To improve the sensitivity and specificity of TMS positive cortico-subcortical connections, fiber tractography of a tumor related major language fascicles was performed with a deterministic tracking algorithm in Brainlab and both deterministic and probabilistic tractography algorithms in MRtrix3 software (n=10). Tractography was performed on B0 dMRIs, overlaid on 3T T1 MPRAGE images and visually inspected for connections to both TMS positive and negative spots.

Results: With state of the art probabilistic tractography algorithms like the ifod2, it was possible to connect a rTMS spot to the fiber bundle of interest in 70%, whereas fiber bundles, dissected with a deterministic algorithm like the sd_stream connected in 30% to the positive rTMS points. Unlike the deterministic algorithm, the probabilistic fiber bundle estimations show more precise fiber estimations in comparison to the white matter anatomy. Furthermore, fiber bundles dissected by probabilistic algorithms are more likely to feature cortical endings in addition to the tract’s obligatory pathway. Seeding from a positive rTMS spot delineated the bundle of interest with the ifod2 algorithm in 80% and with sd_stream in 30%.

Conclusion: When working with deterministic tracking algorithms, visualizing the cortical endings of the fiber bundle of interest is difficult. Therefore, the rTMS positive spots are not connecting to the fiber bundles. Probabilistic fiber tractography on the other hand reveals a clearer network-connectivity of positive TMS spots. It provides more detailed information about individual essential language areas and the overall language network. Challenges are to implement the more demanding and time consuming probabilistic tractography workflow into the clinical context and intraoperative neuronavigation.

Figure 1 [Fig. 1]