This is a prospective and cross-sectional observational study with pre-, intra-, and postoperative clinical and imaging data collection. This study followed the guidelines of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for writing observational studies (29) and was approved by the local ethics committee (CAAE 32871020.4.0000.5258/2019). Informed consent was obtained from all patients. Patients with a histologically confirmed diagnosis of primary (gliomas and meningiomas) and secondary brain tumors (brain metastases) aged over 18 years with surgical indication were included. Patients with deep-positioned tumors (skull base, insular or basal ganglia, and thalamus), patients with an inconclusive histological diagnosis, and patients with recurrent brain tumors or previously irradiated brain tumors were excluded.

Patient age, gender, the occurrence of tumor recurrence/relapse, intraoperative blood pressure, surgical position, type of anesthesia, as well as patient and operating room temperatures, and ambient humidity information, were collected. All patients had their temperature continuously monitored by an esophageal thermometer attached to the anesthesia station. Neuronavigation, iUS, and fluorescence-guided surgery (sodium fluorescein) were not routinely used. Surgeries were performed under general or local anesthesia for intraoperative brain mapping. The decision regarding stopping surgical resection was made by the surgical team, regardless of the findings of the thermographic maps.

Thermographic imaging was performed on a FLIR SC620 camera (FLIR systems, USA), which provides high-spatial resolution (640 × 480 pixels) with a thermal sensitivity of 0.02 °C. To avoid movement artifacts, the camera was placed on a tripod positioned 30 cm–50 cm from the surgical site. Surgical lights were redirected from the surgical field for infrared measurements to avoid tissue heating.

Background static and sequential digital images were obtained (60 images per 1-min trial, 3 trials for patient) during irrigation-induced hypothermia of the exposed dura mater or cortex after the instillation of saline solution at room temperature (cold challenge test). In this case, imaging acquisition started approximately 10 s before and ended 1 min after the end of its application. A sudden temperature drop was identified after saline instillation, thereby generating thermal recovery curves of the exposed tissues (24). Image acquisition was performed before durotomy, after durotomy, and at the end of the tumor resection (as indicated by the surgical team) and did not exceed 10 min in any case. Infrared imaging was analyzed postoperatively with the aid of the FLIR Tools® software (FLIR systems, USA) in previously established locations (Figure 1), namely: •

Healthy cortex (control; at craniotomy margins, away from the tumor)—anterior (A-Spot), posterior (P-Spot), medial (M-Spot), and lateral (L-Spot);

Preoperative T2-weighted FLAIR (fluid-attenuated inversion recovery), T1-weighted gadolinium-enhanced, and diffusion-weighted sequences were performed on a Siemens® 1.5T scanner (Erlangen, Germany). Tumor location, depth (distance between the cortical surface and most superficial tumor location; cortical or subcortical location), and tumor dimensions were determined. Postoperatively, patients underwent imaging within 72 h (computed tomography and/or MRI). Postoperative images were analyzed using Radiant® software [Medixant. RadiAnt DICOM Viewer (Software). Version 2020.1. 9 March 2020. URL: https://www.radiantviewer.com] by two different investigators, a neurosurgeon (DRM) and an experienced neuroradiologist (FR), who was blinded for intraoperative findings and thermographic evaluations. Both investigators determined the extent of resection, the location of potential residual tumor (cavity walls and/or tumor bed), as well as the assessment of acute ischemia.

Descriptive analyses were performed using summaries of numerical variables (mean and percentile) and graphs (thermal recovery curve). After dural opening and full exposure of the tumor, intraoperative inspection was taken as the gold standard for diagnostic accuracy of IIT. Then, a 2 × 2 contingency table was created to calculate IIT sensitivity and specificity for tumor location and the significance of the proportion was assessed by the χ² test. To analyze the variation of the thermographic maps at different surgical steps, the Friedman test was used. The significance level was set at p < 0.05. If Friedman's analysis demonstrated statistical significance, the Wilcoxon test was performed and adjusted for multiple comparisons with a Bonferroni correction, which is the level of confidence (p = 0.05) divided by the number of comparisons, in this study three. This correction results in a level of confidence of p = 0.017 to be considered statistically significant. The paired Wilcoxon test was used to compare the thermographic maps before and after the cold challenge test, while the Mann–Whitney test was used to compare the thermographic maps of the tumor and the healthy cortex. All analyses were performed using the R 4.0.0 software (http://www.r-project.org/).

Fifteen patients comprised the study group, 8 (53.3%) were female with a mean age of 59.4 years (range, 45–73 years). The mean esophageal temperature during surgical procedures was 35.4 °C (range, 33.7 °C–36.7 °C), while the mean operating room temperature was 22.5 °C (range, 20.5 °C–24.6 °C). Fourteen patients were operated on under general anesthesia and one patient underwent awake surgery for brain mapping. Tumors generally affect cortical frontal and parietal lobes on the left side. Subcortical tumors averaged 1.2 cm in depth (range, 0.25 cm–2 cm). Table 1 illustrates the demographics.

Intraoperative thermographic images of the exposed dura mater revealed a mixed vascular pattern of meningeal vessels, veins, and cortical arteries (Figure 2). The temperature difference between vessels, healthy cortex, and intracranial tumors or adjacent cortex was evident. Tumor location was adequately observed in 93.3% of patients through the exposed dura mater (sensitivity 93.3%, specificity 100%, p < 0.00001). In three patients, iUS corroborated the thermographic findings regarding lesion localization.

Static thermographic maps of the healthy cortex and tumors immediately after craniotomy showed great variation [mean of 31.9 °C (range, 28.8–34 °C); and mean of 29.7 °C (range, 26.4–33.5 °C), respectively]. Tumors were consistently hypothermic in comparison to the healthy cortex with a mean temperature difference of 2.2 °C (p = 0.014) across the exposed dura mater. The coolest site in the tumoral region was the C-Spot (29.4 °C; range, 25–34.1 °C). The cold challenge test significantly decreased the mean temperature both in the healthy cortex and tumors [mean of 30.8 °C (range, 27.2–34.5 °C; p = 0.031); mean of 27.8 °C (range, 25–32.6 °C; p = 0.006); respectively]. Such exacerbation of absolute temperature differences increased from 2.2 °C to 3 °C, thereby facilitating the identification of tumoral areas. The subcortical location did not interfere with the recognition of tumors up to 2 cm in depth.

Cortical and tumor area temperatures recovered fast during dural opening reaching baseline values similar to the exposed dura mater [mean of 31.9 °C (p = 0.865), and 30.3 °C (p = 0.912), respectively]. On the other hand, visualization of cortical vessels was facilitated after dura mater retraction due to increased brightness. Intraparenchymal tumors or the cortex overlying the tumors, in the case of subcortical tumors, were on average 1.6 °C cooler than normal parenchyma (p = 0.048). Knowledge of tumor location and vascular relations allowed tumor resection or corticectomy to be precisely planned in all cases.

Tumors of the glial lineage and brain metastases were primarily hypothermic (mean of 1.8 °C, and 3.5 °C, respectively), while meningiomas were hyperthermic (1.2 °C). The analysis of the C-Spot temperatures significantly elucidated histological subtypes (p = 0.055), as well as the cold challenge test (p = 0.015) (Figure 3). Interestingly, two patients with meningiomas developed the hottest thermographic maps after dural opening, due to intense bipolar coagulation for hemostasis of dural feeders (2.9 °C and 12.9 °C).

Once the resection was deemed complete by the surgical team, the thermographic profiles of the tumor bed were measured. In two glioblastomas, the estimation of the extent of resection was done by fluorescence-guided tissue staining (fluorescein sodium). Static thermographic maps of the tumor bed revealed an average increase of 2.8 °C in gliomas (range, 31.2–34.3 °C), and of 4 °C in brain metastases (range, 30.3–34 °C). Glioma tumor resection beds had the highest absolute temperatures when compared to other histological subtypes. In general, the cold challenge test reduced the mean temperature of the tumor bed to 30.5 °C (range, 27.4–33.8 °C), but did not reach statistical significance.

Ten patients underwent postoperative MR imaging to estimate the degree of tumor resection and to identify acute ischemia. From these, 50 thermographic measurements were done from the tumor bed. Gross total resection was achieved in 60% of the patients (n = 6). In four patients (three glioblastomas and 1 metastasis), the resection was deemed near-total or subtotal. Nine areas (18%) of residual tumors on imaging had their thermographic maps retrospectively evaluated. Residual glioblastomas were isothermic (31.9 °C), while residual metastasis was hypothermic in comparison to the surrounding healthy cortex (28.6 °C and 30 °C, respectively). Besides static thermographic imaging, we also evaluated temperature recovery curves after the cold challenge test, which revealed a characteristic delay in the recovery of the thermographic profile at residual tumor spots, when compared to the healthy cortex (Figure 4). In addition, the final absolute temperature was reduced.

Seventeen spots (34%) suggested acute cerebral ischemia, based on postoperative diffusion-weighted sequences. Such areas were also retrospectively evaluated revealing a consistent hypothermic map [mean of 1.8 °C (z = −4.7827; p < 0.0001)] (Figure 5). However, none of the patients presented clinical deterioration.

The present study included 15 patients with intracranial cortical and subcortical tumors of different histological subtypes. Static and dynamic thermographic maps (cold challenge test) were acquired intraoperatively with a high-resolution camera at pre-established time points. Intraoperative thermographic imaging of the exposed dura mater revealed a mixed vascular pattern of meningeal vessels, veins, and cortical arteries. Intra-axial tumors were reliably identified by demonstrating hypothermic areas, which were exacerbated by the cold challenge test.

Dural opening facilitated both the identification of tumors and their neurovascular relations. The analysis of the C-Spot temperatures significantly elucidated histological subtypes. At the end of the surgical resection, there was a consistent rise in the temperature of gliomas and metastases tumor beds. Residual tumors on imaging were retrospectively evaluated with ITT, presenting a typically delayed temperature recovery curve after the cold challenge test. Acute ischemia was consistently hypothermic.

The present study has several limitations. First, the limited number of patients precluded the analysis of different thermographic maps in diverse histological subtypes. Second, it is widely recognized that the perilesional cortex exhibits changes in regional cerebral blood flow (24). Thus, it would be possible that the evaluation of thermographic profiles of the contralateral healthy cortex would reveal greater temperature discrepancies. For obvious reasons, we used the temperature of the healthy cortex in the areas exposed by craniotomy as a control. Third, the thermographic camera did not generally have an orthogonal orientation with the surgical bed. Then, some hidden spots could have been missed in the analysis of the extent of resection. Finally, despite postoperative imaging revealing residual tumor tissue, these were not confirmed with tissue biopsy. We are currently running a study to correlate dynamic thermographic maps with residual tumors. Future studies should assess the feasibility of IIT camera integration into the surgical microscope to favor its potential translation into clinical practice and widespread use of this technology in all the issues that we presented.