Fluorescence and Multiphoton Imaging for Tissue Characterization of a Model of Postmenopausal Ovarian Cancer

Abstract

Background and objectives: To determine the efficacy of targeted fluorescent biomarkers and multiphoton imaging to characterize early changes in ovarian tissue with the onset of cancer.

Study design/materials and methods: A transgenic TgMISIIR-TAg mouse was used as an animal model for ovarian cancer. Mice were injected with fluorescent dyes to bind to the folate receptor α, matrix metalloproteinases, and integrins. Half of the mice were treated with 4-vinylcyclohexene diepoxide (VCD) to simulate menopause. Widefield fluorescence imaging (WFI) and multiphoton imaging of the ovaries and oviducts were conducted at 4 and 8 weeks of age. The fluorescence signal magnitude was quantified, and texture features were derived from multiphoton imaging. Linear discriminant analysis was then used to classify mouse groups.

Results: Imaging features from both fluorescence imaging and multiphoton imaging show significant changes (P < 0.01) with age, VCD treatment, and genotype. The classification model is able to classify different groups to accuracies of 75.53%, 69.53%, and 86.76%, for age, VCD treatment, and genotype, respectively. Building a classification model using features from multiple modalities shows marked improvement over individual modalities.

Conclusions: This study demonstrates that using WFI with targeted biomarkers, and multiphoton imaging with endogenous contrast shows promise for detecting early changes in ovarian tissue with the onset of cancer. The results indicate that multimodal imaging can provide higher sensitivity for classifying tissue types than using single modalities alone. Lasers Surg. Med. © 2020 Wiley Periodicals, Inc.

Keywords: fluorescence imaging; multiphoton imaging; ovarian cancer.

Fig. 1.

Example images taken using MMPSense dye for wild type (a) and TAg+ (b) female mice at eight weeks dosed with sesame oil.

Fig. 2.

Signal intensity for widefield fluorescence imaging using MMPSense (a), FolateRSense (b), and Integrisense (c). (4 - four weeks age, 8 - eight weeks age, W - wild type, T - TAg+, S - treated with vehicle sesame oil, V - treated with VCD). Statistical significance is only denoted for comparisons between genotype groups. Differences were considered statistically significant for p < 0.05 (denoted *), p < 0.01 (denoted **), p < 0.001 (denoted ***).

Fig. 3.

Composite MPM images (green - SHG, red - 2PEF) for different depths for an eight-week wild type mouse. The ovaries deep (a), and superficial (b). The oviducts deep (c), and superficial (d).

Fig. 4.

Sample SHG (top row) and 2PEF (bottom row) images for a wild type and TAg+ female mouse at eight weeks, both dosed with sesame oil.

Fig. 5.

One example texture feature chosen for each: (a) SHG ovaries, (b) SHG oviducts, (c) 2PEF ovaries, (d) 2PEF oviduct. (4 - four weeks age, 8 - eight weeks age, W - wild type, T - TAg+, S - treated with vehicle sesame oil, V - treated with VCD). Statistical significance is only denoted for comparisons between genotype groups. Differences were considered statistically significant for p < 0.05 (denoted *), p < 0.01 (denoted **), p < 0.001 (denoted ***).

Fig. 6.

Example of the classification model between four-week wild type and four-week TAg+ female mice, both dosed with sesame oil (a), and the corresponding ROC curve for the decision boundary (b) to differentiate mice into different genotypes.