Aligning Small Datasets Using Domain Adversarial Learning: Applications in Automated in Vivo Oral Cancer Diagnosis.

Deep learning approaches for medical image analysis are limited by small data set size due to factors such as patient privacy and difficulties in obtaining expert labelling for each image. In medical imaging system development pipelines, phases for system development and classification algorithms often overlap with data collection, creating small disjoint data sets collected at numerous locations with differing protocols. In this setting, merging data from different data collection centers increases the amount of training data. However, a direct combination of datasets will likely fail due to domain shifts between imaging centers. In contrast to previous approaches that focus on a single data set, we add a domain adaptation module to a neural network and train using multiple data sets. Our approach encourages domain invariance between two multispectral autofluorescence imaging (maFLIM) data sets of in vivo oral lesions collected with an imaging system currently in development. The two data sets have differences in the sub-populations imaged and in the calibration procedures used during data collection. We mitigate these differences using a gradient reversal layer and domain classifier. Our final model trained with two data sets substantially increases performance, including a significant increase in specificity. We also achieve a significant increase in average performance over the best baseline model train with two domains (p = 0.0341). Our approach lays the foundation for faster development of computer-aided diagnostic systems and presents a feasible approach for creating a robust classifier that aligns images from multiple data centers in the presence of domain shifts.

Clinical label-free endoscopic imaging of biochemical and metabolic autofluorescence biomarkers of benign, precancerous, and cancerous oral lesions.

Early detection is critical for improving the survival rate and quality of life of oral cancer patients; unfortunately, dysplastic and early-stage cancerous oral lesions are often difficult to distinguish from oral benign lesions during standard clinical oral examination. Therefore, there is a critical need for novel clinical technologies that would enable reliable oral cancer screening. The autofluorescence properties of the oral epithelial tissue provide quantitative information about morphological, biochemical, and metabolic tissue and cellular alterations accompanying carcinogenesis. This study aimed to identify novel biochemical and metabolic autofluorescence biomarkers of oral dysplasia and cancer that could be clinically imaged using novel multispectral autofluorescence lifetime imaging (maFLIM) endoscopy technologies. In vivo maFLIM clinical endoscopic images of benign, precancerous, and cancerous lesions from 67 patients were acquired using a novel maFLIM endoscope. Widefield maFLIM feature maps were generated, and statistical analyses were applied to identify maFLIM features providing contrast between dysplastic/cancerous vs. benign oral lesions. A total of 14 spectral and time-resolved maFLIM features were found to provide contrast between dysplastic/cancerous vs. benign oral lesions, representing novel biochemical and metabolic autofluorescence biomarkers of oral epithelial dysplasia and cancer. To the best of our knowledge, this is the first demonstration of clinical widefield maFLIM endoscopic imaging of novel biochemical and metabolic autofluorescence biomarkers of oral dysplasia and cancer, supporting the potential of maFLIM endoscopy for early detection of oral cancer.

End-to-End Neural Network for Feature Extraction and Cancer Diagnosis of In Vivo Fluorescence Lifetime Images of Oral Lesions.

In contrast to previous studies that focused on classical machine learning algorithms and hand-crafted features, we present an end-to-end neural network classification method able to accommodate lesion heterogeneity for improved oral cancer diagnosis using multispectral autofluorescence lifetime imaging (maFLIM) endoscopy. Our method uses an autoencoder framework jointly trained with a classifier designed to handle overfitting problems with reduced databases, which is often the case in healthcare applications. The autoencoder guides the feature extraction process through the reconstruction loss and enables the potential use of unsupervised data for domain adaptation and improved generalization. The classifier ensures the features extracted are task-specific, providing discriminative information for the classification task. The data-driven feature extraction method automatically generates task-specific features directly from fluorescence decays, eliminating the need for iterative signal reconstruction. We validate our proposed neural network method against support vector machine (SVM) baselines, with our method showing a 6.5%-8.3% increase in sensitivity. Our results show that neural networks that implement data-driven feature extraction provide superior results and enable the capacity needed to target specific issues, such as inter-patient variability and the heterogeneity of oral lesions.Clinical relevance- We improve standard classification algorithms for in vivo diagnosis of oral cancer lesions from maFLIm for clinical use in cancer screening, reducing unnecessary biopsies and facilitating early detection of oral cancer.

Machine-Learning Assisted Discrimination of Precancerous and Cancerous from Healthy Oral Tissue Based on Multispectral Autofluorescence Lifetime Imaging Endoscopy.

Multispectral autofluorescence lifetime imaging (maFLIM) can be used to clinically image a plurality of metabolic and biochemical autofluorescence biomarkers of oral epithelial dysplasia and cancer. This study tested the hypothesis that maFLIM-derived autofluorescence biomarkers can be used in machine-learning (ML) models to discriminate dysplastic and cancerous from healthy oral tissue. Clinical widefield maFLIM endoscopy imaging of cancerous and dysplastic oral lesions was performed at two clinical centers. Endoscopic maFLIM images from 34 patients acquired at one of the clinical centers were used to optimize ML models for automated discrimination of dysplastic and cancerous from healthy oral tissue. A computer-aided detection system was developed and applied to a set of endoscopic maFLIM images from 23 patients acquired at the other clinical center, and its performance was quantified in terms of the area under the receiver operating characteristic curve (ROC-AUC). Discrimination of dysplastic and cancerous from healthy oral tissue was achieved with an ROC-AUC of 0.81. This study demonstrates the capabilities of widefield maFLIM endoscopy to clinically image autofluorescence biomarkers that can be used in ML models to discriminate dysplastic and cancerous from healthy oral tissue. Widefield maFLIM endoscopy thus holds potential for automated in situ detection of oral dysplasia and cancer.

Handheld tunable focus confocal microscope utilizing a double-clad fiber coupler for in vivo imaging of oral epithelium.

A reflectance confocal endomicroscope with double-clad fiber coupler and electrically tunable focus lens is applied to imaging of the oral mucosa. The instrument is designed to be lightweight and robust for clinical use. The tunable lens allows axial scanning through > 250 ?? ? m in the epithelium when the probe tip is placed in contact with tissue. Images are acquired at 6.6 frames per second with a field of view diameter up to 850 ?? ? m . In vivo imaging of a wide range of normal sites in the oral cavity demonstrates the accessibility of the handheld probe. In vivo imaging of clinical lesions diagnosed as inflammation and dysplasia illustrates the ability of reflectance confocal endomicroscopy to image cellular changes associated with pathology.

Objective Detection of Oral Carcinoma with Multispectral Fluorescence Lifetime Imaging In Vivo.

Successful early detection and demarcation of oral carcinoma can greatly impact the associated morbidity and mortality rates. Current methods for detection of oral cancer include comprehensive visual examination of the oral cavity, typically followed by tissue biopsy. A noninvasive means to guide the clinician in making a more objective and informed decision toward tissue biopsy can potentially improve the diagnostic yield of this process. To this end, we investigate the potential of fluorescence lifetime imaging (FLIM) for objective detection of oral carcinoma in the hamster cheek pouch model of oral carcinogenesis in vivo. We report that systematically selected FLIM features can differentiate between low-risk (normal, benign and low-grade dysplasia) and high-risk (high-grade dysplasia and cancer) oral lesions with sensitivity and specificity of 87.26% and 93.96%, respectively. We also show the ability of FLIM to generate "disease" maps of the tissue which can be used to evaluate relative risk of neoplasia. The results demonstrate the potential of multispectral FLIM with objective image analysis as a noninvasive tool to guide comprehensive oral examination.

A novel multimodal optical imaging system for early detection of oral cancer.

OBJECTIVES: Several imaging techniques have been advocated as clinical adjuncts to improve identification of suspicious oral lesions. However, these have not yet shown superior sensitivity or specificity over conventional oral examination techniques. We developed a multimodal, multi-scale optical imaging system that combines macroscopic biochemical imaging of fluorescence lifetime imaging with subcellular morphologic imaging of reflectance confocal microscopy for early detection of oral cancer. We tested our system on excised human oral tissues. STUDY DESIGN: In total, 4 tissue specimens were imaged. These specimens were diagnosed as either clinically normal, oral lichen planus, gingival hyperplasia, or superficially invasive squamous cell carcinoma. The optical and fluorescence lifetime properties of each specimen were recorded. RESULTS: Both quantitative and qualitative differences among normal, benign, and squamous cell carcinoma lesions can be resolved with fluorescence lifetime imaging reflectance confocal microscopy. The results demonstrate that an integrated approach based on these two methods can potentially enable rapid screening and evaluation of large areas of oral epithelial tissue. CONCLUSIONS: Early results from ongoing studies of imaging human oral cavity illustrate the synergistic combination of the 2 modalities. An adjunct device based on such optical characterization of oral mucosa can potentially be used to detect oral carcinogenesis in early stages.

Reflectance confocal endomicroscope with optical axial scanning for in vivo imaging of the oral mucosa.

This paper presents the design and evaluation of a reflectance confocal laser endomicroscope using a miniature objective lens within a rigid probe in conjunction with an electrically tunable lens for axial scanning. The miniature lens was characterized alone as well as in the endoscope across a 200 µm axial scan range using the tunable lens. The ability of the confocal endoscope to probe the human oral cavity is demonstrated by imaging of the oral mucosa in vivo. The results indicate that reflectance confocal endomicroscopy has the potential to be used in a clinical setting and guide diagnostic evaluation of biological tissue.

Estimation of the number of fluorescent end-members for quantitative analysis of multispectral FLIM data.

Multispectral fluorescence lifetime imaging (m-FLIM) can potentially allow identifying the endogenous fluorophores present in biological tissue. Quantitative description of such data requires estimating the number of components in the sample, their characteristic fluorescent decays, and their relative contributions or abundances. Unfortunately, this inverse problem usually requires prior knowledge about the data, which is seldom available in biomedical applications. This work presents a new methodology to estimate the number of potential endogenous fluorophores present in biological tissue samples from time-domain m-FLIM data. Furthermore, a completely blind linear unmixing algorithm is proposed. The method was validated using both synthetic and experimental m-FLIM data. The experimental m-FLIM data include in-vivo measurements from healthy and cancerous hamster cheek-pouch epithelial tissue, and ex-vivo measurements from human coronary atherosclerotic plaques. The analysis of m-FLIM data from in-vivo hamster oral mucosa identified healthy from precancerous lesions, based on the relative concentration of their characteristic fluorophores. The algorithm also provided a better description of atherosclerotic plaques in term of their endogenous fluorophores. These results demonstrate the potential of this methodology to provide quantitative description of tissue biochemical composition.

Handheld multispectral fluorescence lifetime imaging system for in vivo applications.

There is an increasing interest in the application of fluorescence lifetime imaging (FLIM) for medical diagnosis. Central to the clinical translation of FLIM technology is the development of compact and high-speed clinically compatible systems. We present a handheld probe design consisting of a small maneuverable box fitted with a rigid endoscope, capable of continuous lifetime imaging at multiple emission bands simultaneously. The system was characterized using standard fluorescent dyes. The performance was then further demonstrated by imaging a hamster cheek pouch in vivo, and oral mucosa tissue both ex vivo and in vivo, all using safe and permissible exposure levels. Such a design can greatly facilitate the evaluation of FLIM for oral cancer imaging in vivo.

Optical axial scanning in confocal microscopy using an electrically tunable lens.

This paper presents the use and characterization of an electrically focus tunable lens to perform axial scanning in a confocal microscope. Lateral and axial resolution are characterized over a >250 µm axial scan range. Confocal microscopy using optical axial scanning is demonstrated in epithelial tissue and compared to traditional stage scanning. By enabling rapid axial scanning, minimizing motion artifacts, and reducing mechanical complexity, this technique has potential to enhance in vivo three-dimensional imaging in confocal endomicroscopy.

Reflectance confocal microscopy of oral epithelial tissue using an electrically tunable lens.

We present the use of a commercially available electrically tunable lens to achieve axial scanning in a reflectance confocal microscope. Over a 255 µm axial scan range, the lateral and axial resolutions varied from 1-2 µm and 4-14 µm, respectively, dependent on the variable focal length of the tunable lens. Confocal imaging was performed on normal human biopsies from the oral cavity ex vivo. Sub-cellular morphologic features were seen throughout the depth of the epithelium while axially scanning using the focus tunable lens.

Fluorescence lifetime imaging and reflectance confocal microscopy for multiscale imaging of oral precancer.

Optical imaging techniques using a variety of contrast mechanisms are under evaluation for early detection of epithelial precancer; however, tradeoffs in field of view (FOV) and resolution may limit their application. Therefore, we present a multiscale multimodal optical imaging system combining macroscopic biochemical imaging of fluorescence lifetime imaging (FLIM) with subcellular morphologic imaging of reflectance confocal microscopy (RCM). The FLIM module images a 16×16 mm² tissue area with 62.5 µm lateral and 320 ps temporal resolution to guide cellular imaging of suspicious regions. Subsequently, coregistered RCM images are acquired at 7 Hz with 400 µm diameter FOV, <1 µm lateral and 3.5 µm axial resolution. FLIM-RCM imaging was performed on a tissue phantom, normal porcine buccal mucosa, and a hamster cheek pouch model of oral carcinogenesis. While FLIM is sensitive to biochemical and macroscopic architectural changes in tissue, RCM provides images of cell nuclear morphology, all key indicators of precancer progression.