Pterygium and Ocular Surface Squamous Neoplasia: Optical Biopsy Using a Novel Autofluorescence Multispectral Imaging Technique

Abstract

In this study, differentiation of pterygium vs. ocular surface squamous neoplasia based on multispectral autofluorescence imaging technique was investigated. Fifty (N = 50) patients with histopathological diagnosis of pterygium (PTG) and/or ocular surface squamous neoplasia (OSSN) were recruited. Fixed unstained biopsy specimens were imaged by multispectral microscopy. Tissue autofluorescence images were obtained with a custom-built fluorescent microscope with 59 spectral channels, each with specific excitation and emission wavelength ranges, suitable for the most abundant tissue fluorophores such as elastin, flavins, porphyrin, and lipofuscin. Images were analyzed using a new classification framework called fused-classification, designed to minimize interpatient variability, as an established support vector machine learning method. Normal, PTG, and OSSN regions were automatically detected and delineated, with accuracy evaluated against expert assessment by a specialist in OSSN pathology. Signals from spectral channels yielding signals from elastin, flavins, porphyrin, and lipofuscin were significantly different between regions classified as normal, PTG, and OSSN (p < 0.01). Differential diagnosis of PTG/OSSN and normal tissue had accuracy, sensitivity, and specificity of 88 ± 6%, 84 ± 10% and 91 ± 6%, respectively. Our automated diagnostic method generated maps of the reasonably well circumscribed normal/PTG and OSSN interface. PTG and OSSN margins identified by our automated analysis were in close agreement with the margins found in the H&E sections. Such a map can be rapidly generated on a real time basis and potentially used for intraoperative assessment.

Keywords: autofluorescence; boundary detection; machine learning; ocular surface squamous neoplasia; pterygium.

Figure 1

(a) Right nasal pterygium with atypical changes at the superior margin. (b) Gelatinous lesion (arrows) contiguous with and arising in the super-limbal aspect of the pterygium—confirmed as OSSN by biopsy.

Figure 2

(a–e) Sample preparation and histological assessment. (a) Ocular surface biopsy collected from patients. (b) Histology sample processed following formalin fixation into paraffin embedded sections. (c) Two adjacent sections were cut using a microtome and then dewaxed. (d) Example cut tissue section, which was H&E stained and coverslipped for histology assessment and used as reference. (e) The unstained tissue section adjacent to that shown in (d). Such sections were placed on a slide, coverslipped, and used for multispectral imaging analysis. (e–j) Example tissue images in selected channels (channels number 3, 16, 22, 31, and 45, respectively). (k) H&E stained section of example tissue shown in (e–j).

Figure 3

Patient classification performance of the SVM classifier. (a) ROC curve obtained from PTG and OSSN classification. (b) ROC curve obtained from normalized normal and OSSN classification. (c) ROC curve obtained from normal and PTG classification. (d) ROC curve obtained from normal, PTG, and OSSN classification.

Figure 4

Spectral differences between normal, PTG, and OSSN. (a–c) H&E image for normal, PTG, and OSSN sections, respectively. (d–f/g–i) Channel 1/Channel 20 for normal, PTG, and OSSN sections, respectively. (j–l) PCA false color image for normal, PTG, and OSSN sections respectively. OSSN is green, while normal and PTG are violet.

Figure 5

Analysis of fluorophore signals in the tissue. (a) Intensity analysis for Channel 3 containing a contribution from elastin. (b) Intensity analysis for Channel 12 containing a contribution from lipopigment. (c) Intensity analysis for Channel 30 containing a contribution from flavins. (d) Intensity analysis for Channel 52 tentatively attributed to PPIX (*** represents p−value < 0.01). Corresponding sample channels are shown in Supplementary Figure S5.

Figure 6

False color map generated to locate the normal, PTG, and OSSN boundary on the testing section based on intra-patient classification framework compared to the associated histology images. The block data position on a single spectral channel (Chno = 10) image is colored in red/orange/green if they are predicted to be OSSN/PTG/normal. First/third column is the multispectral false-color map. Second/forth column is corresponding H&E section with red, orange, and green dash lines highlighting OSSN, PTG, and normal section.