gms | German Medical Science

Objective: The combination of thermography and an intravenous cold saline injection as a contrast agent might be a promising approach for intraoperative cortical perfusion imaging. The aim of this study is the evaluation of the captured cold signal among different cortical regions to improve the algorithmic analysis of continuous thermographic imaging.

Methods: Thermographic imaging was performed using an infrared camera with a resolution of 30 mK, 125 µm per pixel and an acquisition rate of 60 Hz. The exposed cerebral cortex was recorded during application of a cold bolus (4 °C, 0.9% NaCl, 50 ml). First, a preprocessing procedure was performed for extracting desired signal components. For perfusion parameter determination, an asymmetric Gaussian function was approximated to the preprocessed thermal profiles. To verify the usage of this algorithm we used imaging sequences from 11 patients, which had provided informed consent preoperatively. 15 bolus injections during intracranial surgery (5 AVM, 3 Infarction, 4 Epilepsy surgery, 1 Aneurysm clipping, 2 GBM) were analyzed regarding cortical perfusion parameters of arterial vessels and parenchymal regions.

Results: The cold bolus was detected in 43.8 ± 17% of all cortical pixels among all patients. The cold signal arrived 19 ± 6.1 s after injection in arterial vessels and 21.4 ± 7.3 s in the parenchyma. The signal peak in arteries of 120 ± 30 mK was reached earlier (after 5.6 ± 1.8 s) than the corresponding signal peak of 100 ± 50 mK of parenchymal regions (6.1 ± 2.7 s).The delays were due to the cold bolus mixing and distribution pattern. A warming up during brain tissue passage leads to a decrease of signal intensity in the parenchyma. The signal duration in the parenchyma with 11.3 ± 3.8 s was clearly shorter and less variant compared to arterial vessels with 15 ± 7.7 s. This observation implies to use different model functions and algorithms for the analysis of different cortical regions. In arterial vessels, the signal rise time was a steep flank while the increase to baseline is flattened. In contrast, the parenchymal signal behavior was characterized by a flatter signal decrease followed by a sharper rise after the peak temperature.

Conclusion: Implementation of the varying temperature signals provided from intraoperative thermal imaging of various cortical perfusion patterns, we were able to improve the sensitivity and reliability of intraoperative thermographic perfusion imaging substantially.