Weakly supervised identification of microscopic human breast cancer-related optical signatures from normal-appearing breast tissue.

With the latest advancements in optical bioimaging, rich structural and functional information has been generated from biological samples, which calls for capable computational tools to identify patterns and uncover relationships between optical characteristics and various biomedical conditions. Constrained by the existing knowledge of the novel signals obtained by those bioimaging techniques, precise and accurate ground truth annotations can be difficult to obtain. Here we present a weakly supervised deep learning framework for optical signature discovery based on inexact and incomplete supervision. The framework consists of a multiple instance learning-based classifier for the identification of regions of interest in coarsely labeled images and model interpretation techniques for optical signature discovery. We applied this framework to investigate human breast cancer-related optical signatures based on virtual histopathology enabled by simultaneous label-free autofluorescence multiharmonic microscopy (SLAM), with the goal of exploring unconventional cancer-related optical signatures from normal-appearing breast tissues. The framework has achieved an average area under the curve (AUC) of 0.975 on the cancer diagnosis task. In addition to well-known cancer biomarkers, non-obvious cancer-related patterns were revealed by the framework, including NAD(P)H-rich extracellular vesicles observed in normal-appearing breast cancer tissue, which facilitate new insights into the tumor microenvironment and field cancerization. This framework can be further extended to diverse imaging modalities and optical signature discovery tasks.

Real-time three-dimensional histology-like imaging by label-free nonlinear optical microscopy.

BACKGROUND: The current gold-standard formalin-fixed and paraffin-embedded (FFPE) histology typically requires several days for tissue fixing, embedding, sectioning, and staining to provide depth-resolved tissue feature visualization. During these time- and labor- intense processes, the in vivo tissue dynamics and three-dimensional structures undergo inevitable loss and distortion. METHODS: A simultaneous label-free autofluorescence multiharmonic (SLAM) microscope is used to conduct ex vivo and in vivo imaging of fresh human and rat tissues. Four nonlinear optical imaging modalities are integrated into this SLAM microscope, including second harmonic generation (SHG), two-photon fluorescence (2PF), third harmonic generation (THG), and three-photon fluorescence (3PF). By imaging fresh human and rat tissues without any tissue processing or staining, various biological tissue features are effectively visualized by one or multiple imaging modalities of the SLAM microscope. In particular, some of the most essential features in hematoxylin and eosin (H&E)-stained histology, such as collagen fibers and nuclei, are also present in the SLAM microscopy images with good contrast. Because nuclei are evident from negative contrast, the nuclei are segmented from the SLAM images using deep learning. Finally, a color-transforming algorithm is developed to convert the grey-scale images acquired by the SLAM microscope to the virtually H&E-stained histology-like images. The converted histology-like images are later compared with the FFPE histology at the same tissue site. In addition, the nuclear-to-cytoplasmic ratios (N/C ratios) of the cells in the SLAM image are quantified, which has diagnostic relevance for cancer. RESULTS: Various histological correlations are identified with high similarities for the color-converted histology-like SLAM microscopy images. By applying the color transforming algorithm on real-time SLAM image sequences and 3D SLAM image stacks, we report, for the first time and to the best our knowledge, real-time 3D histology-like imaging. Furthermore, the quantified N/C ratio of the cells in the SLAM image are overlaid on the converted histology-like image as a new image contrast. CONCLUSIONS: We demonstrated real-time 3D histology-like imaging and its future potential using SLAM microscopy aided by color remapping and deep-learning-based feature segmentation.

Complementary use of polarization-sensitive and standard OCT metrics for enhanced intraoperative differentiation of breast cancer.

We report the development and implementation of an intraoperative polarization-sensitive optical coherence tomography (PS-OCT) system for enhancing breast cancer detection. A total of 3440 PS-OCT images were intraoperatively acquired from 9 human breast specimens diagnosed by H&E histology as healthy fibro-adipose tissue (n = 2), healthy stroma (n = 2), or invasive ductal carcinoma (IDC, n = 5). A standard OCT-based metric (coefficient of variation (CV)) and PS-OCT-based metrics sensitive to biological tissue from birefringence (i.e., retardation and degree of polarization uniformity (DOPU)) were derived from 398 statistically different and independent images selected by correlation coefficient analysis. We found the standard OCT-based metric and PS-OCT-based metrics were complementary for the differentiation of healthy fibro-adipose tissue, healthy stroma, and IDC. While the CV of fibro-adipose tissue was significantly higher (p<0.001) than those of either stroma or IDC, the CV difference between stroma and IDC was minimal. On the other hand, stroma was associated with significantly higher (p<0.001) retardation and significantly lower (p<0.001) DOPU as compared to IDC. By leveraging the complementary information acquired by the intraoperative PS-OCT system, healthy fibro-adipose tissue, healthy stroma, and IDC can be differentiated with an accuracy of 89.4%, demonstrating the potential of PS-OCT as an adjunct modality for enhanced intraoperative differentiation of human breast cancer.

Intraoperative optical coherence tomography of the human thyroid: Feasibility for surgical assessment.

Thyroid nodules assessed with ultrasound and fine-needle aspiration biopsy are diagnosed as "suspicious" or "indeterminate" in 15%-20% of the cases. Typically, total thyroidectomy is performed in such cases; however, only 25%-50% are found to be cancerous upon final histopathologic analysis. Here we demonstrate optical coherence tomography (OCT) imaging of the human thyroid as a potential intraoperative imaging tool for providing tissue assessment in real time during surgical procedures. Fresh excised tissue specimens from 28 patients undergoing thyroid surgery were imaged in the laboratory using a benchtop OCT system. Three-dimensional OCT images showed different microstructural features in normal, benign, and malignant thyroid tissues. A similar portable OCT system was then designed and constructed for use in the operating room, and intraoperative imaging of excised thyroid tissue from 6 patients was performed during the surgical procedure. The results demonstrate the potential of OCT to provide real-time imaging guidance during thyroid surgeries.

Intraoperative optical coherence tomography for assessing human lymph nodes for metastatic cancer.

BACKGROUND: Evaluation of lymph node (LN) status is an important factor for detecting metastasis and thereby staging breast cancer. Currently utilized clinical techniques involve the surgical disruption and resection of lymphatic structure, whether nodes or axillary contents, for histological examination. While reasonably effective at detection of macrometastasis, the majority of the resected lymph nodes are histologically negative. Improvements need to be made to better detect micrometastasis, minimize or eliminate lymphatic disruption complications, and provide immediate and accurate intraoperative feedback for in vivo cancer staging to better guide surgery. METHODS: We evaluated the use of optical coherence tomography (OCT), a high-resolution, real-time, label-free imaging modality for the intraoperative assessment of human LNs for metastatic disease in patients with breast cancer. We assessed the sensitivity and specificity of double-blinded trained readers who analyzed intraoperative OCT LN images for presence of metastatic disease, using co-registered post-operative histopathology as the gold standard. RESULTS: Our results suggest that intraoperative OCT examination of LNs is an appropriate real-time, label-free, non-destructive alternative to frozen-section analysis, potentially offering faster interpretation and results to empower superior intraoperative decision-making. CONCLUSIONS: Intraoperative OCT has strong potential to supplement current post-operative histopathology with real-time in situ assessment of LNs to preserve both non-cancerous nodes and their lymphatic vessels, and thus reduce the associated risks and complications from surgical disruption of lymphoid structures following biopsy.

Real-time Imaging of the Resection Bed Using a Handheld Probe to Reduce Incidence of Microscopic Positive Margins in Cancer Surgery.

Wide local excision (WLE) is a common surgical intervention for solid tumors such as those in melanoma, breast, pancreatic, and gastrointestinal cancer. However, adequate margin assessment during WLE remains a significant challenge, resulting in surgical reinterventions to achieve adequate local control. Currently, no label-free imaging method is available for surgeons to examine the resection bed in vivo for microscopic residual cancer. Optical coherence tomography (OCT) enables real-time high-resolution imaging of tissue microstructure. Previous studies have demonstrated that OCT analysis of excised tissue specimens can distinguish between normal and cancerous tissues by identifying the heterogeneous and disorganized microscopic tissue structures indicative of malignancy. In this translational study involving 35 patients, a handheld surgical OCT imaging probe was developed for in vivo use to assess margins both in the resection bed and on excised specimens for the microscopic presence of cancer. The image results from OCT showed structural differences between normal and cancerous tissue within the resection bed following WLE of the human breast. The ex vivo images were compared with standard postoperative histopathology to yield sensitivity of 91.7% [95% confidence interval (CI), 62.5%-100%] and specificity of 92.1% (95% CI, 78.4%-98%). This study demonstrates in vivo OCT imaging of the resection bed during WLE with the potential for real-time microscopic image-guided surgery.