Cross-Polarization Optical Coherence Tomography for Brain Tumor Imaging

Abstract

This paper considers valuable visual assessment criteria for distinguishing between tumorous and non-tumorous tissues, intraoperatively, using cross-polarization OCT (CP OCT)-OCT with a functional extension, that enables detection of the polarization properties of the tissues in addition to their conventional light scattering. Materials and Methods: The study was performed on 176 ex vivo human specimens obtained from 30 glioma patients. To measure the degree to which the typical parameters of CP OCT images can be matched to the actual histology, 100 images of tumors and white matter were selected for visual analysis to be undertaken by three "single-blinded" investigators. An evaluation of the inter-rater reliability between the investigators was performed. Application of the identified visual CP OCT criteria for intraoperative use was performed during brain tumor resection in 17 patients. Results: The CP OCT image parameters that can typically be used for visual assessment were separated: (1) signal intensity; (2) homogeneity of intensity; (3) attenuation rate; (4) uniformity of attenuation. The degree of match between the CP OCT images and the histology of the specimens was significant for the parameters "signal intensity" in both polarizations, and "homogeneity of intensity" as well as the "uniformity of attenuation" in co-polarization. A test based on the identified criteria showed a diagnostic accuracy of 87-88%. Intraoperative in vivo CP OCT images of white matter and tumors have similar signals to ex vivo ones, whereas the cortex in vivo is characterized by indicative vertical striations arising from the "shadows" of the blood vessels; these are not seen in ex vivo images or in the case of tumor invasion. Conclusion: Visual assessment of CP OCT images enables tumorous and non-tumorous tissues to be distinguished. The most powerful aspect of CP OCT images that can be used as a criterion for differentiation between tumorous tissue and white matter is the signal intensity. In distinguishing white matter from tumors the diagnostic accuracy using the identified visual CP OCT criteria was 87-88%. As the CP OCT data is easily associated with intraoperative neurophysiological and neuronavigation findings this can provide valuable complementary information for the neurosurgeon tumor resection.

Keywords: cross-polarization optical coherence tomography (CP OCT); glioblastoma; imaging assessment; intraoperative imaging; malignant brain tumors.

Figure 1

Design of the ex vivo and in vivo CP OCT study: (A) the ex vivo study was performed on material from operative biopsies: 30 patients with gliomas of different grades of malignancy; in total 274 ex vivo images were analyzed; (B) the in vivo study was performed on 17 patients with different grades of malignant brain tumors; in total 341 in vivo images were analyzed; (C) working area for CP OCT scanning with the experimental CP OCT device and on-mount optical probe; specimen with schematic marking of the scanning area along the central line (yellow dotted line); (D) CP OCT device approved for clinical use, with a handled OCT probe in a sterile cover; (F) in vivo, and (E) ex vivo CP OCT images in co- and cross-polarizations. The signal in cross-polarized image is orthogonally polarized backscattered light, which is mutually coherent with the incident one and can appear if the tissue has anisotropic structures such as myelinated fibers.

Figure 2

Illustration from the training set of CP OCT images: (A1–A6), (B1–B6), (C1–C6), (D1–D6) show examples of the assignment of certain characteristic to the CP OCT signal; (A1–A6)—intense signals are marked with blue rectangles, violet—a low-intensity signal; (B1,B2)—the regions of homogeneous signals are indicated by green rectangles; (B3–B6)—green areas and pale blue squares indicate areas of different intensities; (C1–C6)—blue arrows denote regions with high rates of signal attenuation, violet—with low signal attenuation rates; (D1,D2,D6)—the green rectangles indicate areas with uniform attenuation of the signal; (D3–D5)—rectangles of pale blue and green color indicate regions with different signal attenuation rates.

Figure 3

Examples of images from the training set for the second test: (A)—white matter, (B)—tumor. The responder identifies tissue type using main criteria: white matter is characterized by high intense signal in co- and cross-polarization unlike low intensity signal of tumorous tissue. In the doubtful case, the additional criteria can be used.

Figure 4

Ex vivo CP OCT images (A1–A5), (B1–B5) and corresponding histology (C1–C5) of white matter (A1–C1), diffuse astrocytoma Grade II (A2–C2), (A3–C3), diffuse astrocytoma Grade III (A4–C4) and glioblastoma (A5–C5); on the CP OCT image of diffuse astrocytoma Grade II can be seen microcysts typical for this type of tumor (A3,C3); (A1–A5)—CP OCT images in co-polarization and (B1–B5)—in cross-polarization. Histological images (C1–C5)-hematoxylin and eosin staining.

Figure 5

Intraoperative CP OCT images of cortex (A1,A2), white matter (B1,B2) and diffuse astrocytoma Grade II (C1,C2); the scanning areas of the corresponding tissue types in the surgical field are marked with green (A1,A2), blue (B1,B2) and violet (C1,C2) dotted lines (D,E), and interrelated with the neuronavigation data (F,G); (A1–C1)—CP OCT images in co-polarization and (A2–C2)—in cross-polarization. The white arrows show characteristic vertical striations arising from “shadows” of the blood vessels located just under the tissue surface.

Figure 6

Intraoperative CP OCT images in a patient with diffuse astrocytoma (Grade II): of the cortex (A1) with well-defined (A2,A3; A6—green dotted line) and invisible (A4,A5) tumor invasion, white matter (B1) close to the right corticospinal tract (B2—blue color dotted line; C3,C4—marked by red color); the scanning areas of corresponding tissue types in the surgical field (A6,B2) are marked with green (A6) and, blue (B2) dotted lines and interrelated with the neuronavigation data, where the tumor is marked with a green line and the corticospinal tract with a red line (C2,C3,C6). Preoperative MRI in T1 and T2 (C1) and the corticospinal tract reconstruction based on the DTI before (C4) and after (C5) operation. (A1–A5,B1)—CP OCT images in co-polarization (upper image) and cross-polarization (lower image).