Autofluorescence Multispectral Imaging Research Articles

Single cell, Label free Characterisation of Human Mesenchymal Stromal cell Stemness and Future Growth Potential by Autofluorescence Multispectral Imaging.

To use autofluorescence multispectral imaging (AFMI) to develop a non-invasive assay for the in-depth characterisation of human bone marrow derived mesenchymal stromal cells (hBM-MSCs). hBM-MSCs were imaged by AFMI on gridded dishes, stained for endpoints of interest (STRO-1 positivity, alkaline phosphatase, beta galactosidase, DNA content) then relocated and results correlated. Intensity, texture and morphological features were used to characterise the colour distribution of regions of interest, and canonical discriminant analysis was used to separate groups. Additionally, hBM-MSC lines were cultured to arrest, with AFMI images taken after each passage to investigate whether an assay could be developed for growth potential. STRO-1 positivity could be predicted with a receiver operator characteristic area under the curve (AUC) of 0.67. For spontaneous differentiation this was 0.66, for entry to the cell-cycle it was 0.77 and for senescence it was 0.77. Growth potential (population doublings remaining) was estimated with an RMSPE = 2.296. The Mean Absolute Error of the final prediction model indicated that growth potential could be predicted with an error of ± 1.86 doublings remaining. This non-invasive methodology enabled the in-depth characterisation of hBM-MSCs from a single assay. This approach is advantageous for clinical applications as well as research and stands out for the characterisation of both present status as well as future behaviour. The use of data from five MSC lines with heterogenous AFMI profiles supports potential generalisability.

Computer-assisted discrimination of cancerous and pre-cancerous from benign oral lesions based on multispectral autofluorescence imaging endoscopy

Computer-assisted discrimination of cancerous and pre-cancerous from benign oral lesions based on multispectral autofluorescence imaging endoscopy

#328 Non-invasive assessment of urinary exfoliated proximal tubule cells using multispectral autofluorescence imaging features for early detection of CKD

Abstract Background and Aims Early detection of CKD is important to allow for timely intervention which may reduce adverse clinical outcomes and cost of renal replacement therapies. Histopathological diagnosis of CKD remains the gold standard, but routine kidney biopsy is costly and associated with risks such as pain and bleeding. There is a need to develop accurate non-invasive methods for early diagnosis and prognostication of CKD. Kidney cell exfoliation into urine is an active process, which may generate significant information regarding an individual's kidney status. In particular, proximal tubule cells (PTCs) make up majority of the kidney mass and represents the hallmark of CKD in reflecting the degree of tubulointerstitial fibrosis and atrophy. Multispectral assessment of native cell autofluorescence has been specifically shown to be sensitive to metabolic changes and oxidative stress, which are key factors involved in kidney function decline. Hence, we evaluated whether multispectral autofluorescence signals in urinary exfoliated PTCs reflect kidney function as determined by eGFR. Method Individuals aged 18 or above were included and informed consent was obtained from all participants involved in this study. Spot urine samples were collected and stored at -80°C. On assessment, exfoliated PTCs were extracted from thawed urine using a validated specific immunomagnetic separation method based on anti-CD13 and anti-SGLT2 antibodies. Multispectral autofluorescence imaging was performed using a custom-made autofluorescence microscopy system (standard Olympus iX83™ fluorescence microscope and multispectral excitation lamp from Quantitative™, AU). Images were captured by a Nüvü™ EMCCD camera HNü 1024 cooled to -65°C in our laboratory to reduce sensor-induced noise. For the 34 image channels taken per sample, image acquisition times of up to 5 seconds per channel, with multiple averaging (3 times) was applied to optimize image quality. Image analysis was conducted through a workflow encompassing data preprocessing, data normalization and outlier handling, followed by basic model training and evaluation. An entropy-based image feature selection methodology was initially completed by hand, with results subsequently compared to that generated from the AutoGluon machine-learning framework [1]. The models developed from this process were assessed using ROC (Receiver Operating Characteristic) curves and AUC (Area Under the Curve) values. Two discriminative studies were performed – an overall study between exfoliated PTCs of individuals eGFR≥60 and eGFR < 60 ml/min/1.73 m2 and a subgroup study between that of individuals eGFR≥90 and eGFR 60-90 ml/min/1.73 m2. Results 40 individuals were included in our study. 20 individuals had eGFR≥60 and 20 had eGFR < 60 ml/min/1.73 m2. Multispectral assessment of cell autofluorescence features differentiated between exfoliated PTCs of these 2 groups with AUC value of 0.97 (95% CI 0.94-1.00) (Fig. 1), model accuracy being 82%. Subgroup analysis included 20 individuals of which 10 had eGFR≥90 and 10 had eGFR 60-90 ml/min/1.73 m2. Multispectral assessment of cell autofluorescence features differentiated between exfoliated PTCs of these 2 groups with AUC value of 0.99 (95% CI 0.99-1.00) (Fig. 2), model accuracy being 85%. Conclusion An excellent degree of differentiation between urinary PTCs derived from those with normal or high eGFR and mild degrees of CKD is now reliably demonstrated, complementing conclusions from our group's original small pilot study which evaluated between urinary PTCs from CKD and non-CKD cohorts [2]. Further studies are required to determine whether our novel method can be used to non-invasively predict individuals at risk of progressive kidney disease and to monitor whether intervention is effective.

1243P Ocular surface squamous neoplasia early diagnosis using an AI-empowered autofluorescence multispectral imaging technique

1243P Ocular surface squamous neoplasia early diagnosis using an AI-empowered autofluorescence multispectral imaging technique

AI-Powered Biomolecular-Specific and Label-Free Multispectral Imaging Rapidly Detects Malignant Neoplasm in Surgically Excised Breast Tissue Specimens.

Repeated surgery is necessary for 20% to 40% of breast conservation surgeries owing to the unavailability of any adjunctive, accurate, and objective tool in the surgeon's hand for real-time margin assessment to achieve the desired balance of oncologic and cosmetic outcomes. To assess the feasibility of using a multispectral autofluorescence imaging device for discriminating malignant neoplasm from normal breast tissue in pathology as a critical step in the development of a device for intraoperative use, and to demonstrate the device's utility for use in processing and prioritizing specimens during frozen section and in the pathology grossing room. We performed a preliminary assessment of our device, called the TumorMAP system, on 172 fresh tissue blocks from 115 patients obtained from lumpectomy specimens at the time of initial gross examination and compared the device results with gold standard pathology evaluation. The preliminary results demonstrate the potential of our device in detecting breast cancer in fresh tissue samples with a sensitivity of 82%, a specificity of 91%, a positive predictive value of 84%, and a negative predictive value of 89%. Our results suggest that the TumorMAP system is suitable for the detection of malignant neoplasm in freshly excised breast specimens and has the potential to evaluate resection margins in real time.

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.

Exfoliated Kidney Cells from Urine for Early Diagnosis and Prognostication of CKD: The Way of the Future?

Chronic kidney disease (CKD) is a global health issue, affecting more than 10% of the worldwide population. The current approach for formal diagnosis and prognostication of CKD typically relies on non-invasive serum and urine biomarkers such as serum creatinine and albuminuria. However, histological evidence of tubulointerstitial fibrosis is the ‘gold standard’ marker of the likelihood of disease progression. The development of novel biomedical technologies to evaluate exfoliated kidney cells from urine for non-invasive diagnosis and prognostication of CKD presents opportunities to avoid kidney biopsy for the purpose of prognostication. Efforts to apply these technologies more widely in clinical practice are encouraged, given their potential as a cost-effective approach, and no risk of post-biopsy complications such as bleeding, pain and hospitalization. The identification of biomarkers in exfoliated kidney cells from urine via western blotting, enzyme-linked immunosorbent assay (ELISA), immunofluorescence techniques, measurement of cell and protein-specific messenger ribonucleic acid (mRNA)/micro-RNA and other techniques have been reported. Recent innovations such as multispectral autofluorescence imaging and single-cell RNA sequencing (scRNA-seq) have brought additional dimensions to the clinical application of exfoliated kidney cells from urine. In this review, we discuss the current evidence regarding the utility of exfoliated proximal tubule cells (PTC), podocytes, mesangial cells, extracellular vesicles and stem/progenitor cells as surrogate markers for the early diagnosis and prognostication of CKD. Future directions for development within this research area are also identified.

Application of multispectral imaging combined with machine learning models to discriminate special and traditional green coffee

Non-destructive techniques aided by machine learning models are widely implemented in food analysis. To discriminate between ‘special’ and ‘traditional’ classes of green coffee beans, an advanced multispectral imaging technique based on reflectance and autofluorescence data was employed in combination with four machine learning algorithms (SVM, RF, XGBoost, and CatBoost). Of the four algorithms, SVM showed superior accuracy (0.96) for the test dataset. Analysis using PCA and SVM algorithms showed that autofluorescence data from excitation/emission combination of 405/500 nm contributed most to the discrimination of special green coffee from the traditional class. Fluorophores that can be linked to green fluorescence, namely catechin, caffeine and 4-hydroxybenzoic, synapic and chlorogenic acids, were found to have a considerable influence on the differentiation of specialty and traditional coffees. Analysis based on multispectral autofluorescence imaging combined with SVM models was proven to be a valuable tool for future applications in the food industry for the non-destructive and real-time classification of special and traditional green coffee.

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

Simple SummaryCancer is able to damage the surface of the eye, especially in countries like Australia with high exposure to ultraviolet radiation from the sun. Such cancer (ocular surface squamous neoplasia or OSSN) is similar in appearance to a common and benign eye disease called pterygium. Currently, eye biopsy is the gold standard diagnostic method for OSSN, which is traumatic for the patient, carries risks, and has been potentially unnecessary in patients diagnosed with pterygium only. This research introduces an imaging-based method for OSSN screening, which will reduce pressure on health resources—and hopefully eliminate the need for eye biopsy in cases of suspected OSSN.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.

A Clinical Study to Evaluate Autofluorescence Imaging of Diabetic Foot Ulcers Using a Novel Artificial Intelligence Enabled Noninvasive Device.

Diabetic foot ulcers, with worldwide prevalence ranging from 12%-25%, are an important cause of nontraumatic lower limb amputation. Evidence-based assessment of early infection can help the clinician provide the right first line treatment thus helping improve the wound closure rate. Illuminate®, a novel point of care device working on multispectral autofluorescence imaging, helps in the rapid identification and classification of bacteria. This study was aimed to evaluate the diagnostic accuracy of the device in detecting bacterial gram type against standard culture methods. A total of 178 patients from a tertiary care center for diabetes was recruited and 203 tissue samples were obtained from the wound base by the plastic surgeon. The device was handled by the trained investigator to take wound images. The tissue samples were taken from the color-coded infected region as indicated by the device's Artificial Intelligence algorithm and sent for microbial assessment. The results were compared against the Gram type inferred by the device and the device was found to have an accuracy of 89.54%, a positive predictive value of 86.27% for detecting Gram-positive bacteria, 80.77% for Gram-negative bacteria, and 91.67% for no infection. The negative predictive value corresponded to 87.25% for Gram-positive, 92% for Gram-negative, and 96.12% for no infection. The Results exhibited the accuracy of this novel autofluorescence device in identifying and classifying the gram type of bacteria and its potential in significantly aiding clinicians towards early infection assessment and treatment.

Smartphone-enabled snapshot multispectral autofluorescence imaging and its application for bacteria assessments in skin and oral cavity

We propose a smartphone-enabled system and method to realize multispectral autofluorescence imaging for analyzing bacterial infection within skin and oral cavity. The system consists of an unmodified and intact commercial smartphone and a home-made cell phone case with built-in black light LEDs. We use Wiener estimation method to calibrate the RGB-mode smartphone camera and transform the acquired autofluorescence photographs into pseudo-multispectral data cubes with 15 wavebands ranging from 420 nm to 700 nm. Then, we extract and compare the spectral performance of emissions from bacteria-produced porphyrins and endogenous background tissue. Based on the extracted autofluorescence spectra, we apply weighted subtraction between wavebands of interest to realize bacteria targeting and feature mapping. We conduct analysis of autofluorescence on facial skin and dental plaques to demonstrate the performance of the proposed system and methods. Further, with this strategy, we realize quantitative analysis of the bacterial infection in the combination and oily types of skin. Compared to traditional bacteria assessment strategies, we provide a method with the features of real-time visualization, label-free molecular identification, and feature mapping. Meanwhile, differing from the most conventional multispectral imaging systems, the proposed smartphone-based system works in a snapshot mode, thus improving its immunity to motion artifacts. Considering the popularity of smartphones in today's world, it is expected a relatively easy acceptance of the proposed cost-effective method by the community, making an impact on skin and oral care in general and in rural areas with low resource settings in particular.

Non-invasive real-time imaging of reactive oxygen species (ROS) using auto-fluorescence multispectral imaging technique: A novel tool for redox biology

Detecting reactive oxygen species (ROS) that play a critical role as redox modulators and signalling molecules in biological systems currently requires invasive methods such as ROS -specific indicators for imaging and quantification. We developed a non-invasive, real-time, label-free imaging technique for assessing the level of ROS in live cells and thawed cryopreserved tissues that is compatible with in-vivo imaging. The technique is based on autofluorescence multispectral imaging (AFMI) carried out in an adapted fluorescence microscope with an expanded number of spectral channels spanning specific excitation (365 nm–495 nm) and emission (420 nm–700 nm) wavelength ranges. We established a strong quantitative correlation between the spectral information obtained from AFMI and the level of ROS obtained from CellROX staining. The results were obtained in several cell types (HeLa, PANC1 and mesenchymal stem cells) and in live kidney tissue. Additioanly,two spectral regimes were considered: with and without UV excitation (wavelengths > 400 nm); the latter being suitable for UV-sensitive systems such as the eye. Data were analyzed by linear regression combined with an optimization method of swarm intelligence. This allowed the calibration of AFMI signals to the level of ROS with excellent correlation (R = 0.84, p = 0.00) in the entire spectral range and very good correlation (R = 0.78, p = 0.00) in the limited, UV-free spectral range. We also developed a strong classifier which allowed us to distinguish moderate and high levels of ROS in these two regimes (AUC = 0.91 in the entire spectral range and AUC = 0.78 for UV-free imaging). These results indicate that ROS in cells and tissues can be imaged non-invasively, which opens the way to future clinical applications in conditions where reactive oxygen species are known to contribute to progressive disease such as in ophthalmology, diabetes, kidney disease, cancer and neurodegenerative diseases.

Oocyte and embryo evaluation by AI and multi-spectral auto-fluorescence imaging: Livestock embryology needs to catch-up to clinical practice

A highly accurate ‘non-invasive quantitative embryo assessment for pregnancy’ (NQEAP) technique that determines embryo quality has been an elusive goal. If developed, NQEAP would transform the selection of embryos from both Multiple Ovulation and Embryo Transfer (MOET), and even more so, in vitro produced (IVP) embryos for livestock breeding. The area where this concept is already having impact is in the field of clinical embryology, where great strides have been taken in the application of morphokinetics and artificial intelligence (AI); while both are already in practice, rigorous and robust evidence of efficacy is still required. Even the translation of advances in the qualitative scoring of human IVF embryos have yet to be translated to the livestock IVP industry, which remains dependent on the MOET-standardised 3-point scoring system. Furthermore, there are new ways to interrogate the biochemistry of individual embryonic cells by using new, light-based methodologies, such as FLIM and hyperspectral microscopy. Combinations of these technologies, in particular combining new imaging systems with AI, will lead to very accurate NQEAP predictive tools, improving embryo selection and recipient pregnancy success.

Novel automated non invasive detection of ocular surface squamous neoplasia using multispectral autofluorescence imaging

PurposeDiagnosing Ocular surface squamous neoplasia (OSSN) using newly designed multispectral imaging technique. MethodsEighteen patients with histopathological diagnosis of Ocular Surface Squamous Neoplasia (OSSN) were recruited. Their previously collected biopsy specimens of OSSN were reprocessed without staining to obtain auto fluorescence multispectral microscopy images. This technique involved a custom-built spectral imaging system with 38 spectral channels. Inter and intra-patient frameworks were deployed to automatically detect and delineate OSSN using machine learning methods. Different machine learning methods were evaluated, with K nearest neighbor and Support Vector Machine chosen as preferred classifiers for intra- and inter-patient frameworks, respectively. The performance of the technique was evaluated against a pathological assessment. ResultsQuantitative analysis of the spectral images provided a strong multispectral signature of a relative difference between neoplastic and normal tissue both within each patient (at p < 0.0005) and between patients (at p < 0.001). Our fully automated diagnostic method based on machine learning produces maps of the relatively well circumscribed neoplastic-non neoplastic interface. Such maps can be rapidly generated in quasi-real time and used for intraoperative assessment. Generally, OSSN could be detected using multispectral analysis in all patients investigated here. The cancer margins detected by multispectral analysis were in close and reasonable agreement with the margins observed in the H&E sections in intra- and inter-patient classification, respectively. ConclusionsThis study shows the feasibility of using multispectral auto-fluorescence imaging to detect and find the boundary of human OSSN. Fully automated analysis of multispectral images based on machine learning methods provides a promising diagnostic tool for OSSN which can be translated to future clinical applications.

Programmable LED-based integrating sphere light source for wide-field fluorescence microscopy

Wide-field fluorescence microscopy commonly uses a mercury lamp, which has limited spectral capabilities. We designed and built a programmable integrating sphere light (PISL) source which consists of nine LEDs, light-collecting optics, a commercially available integrating sphere and a baffle. The PISL source is tuneable in the range 365–490nm with a uniform spatial profile and a sufficient power at the objective to carry out spectral imaging. We retrofitted a standard fluorescence inverted microscope DM IRB (Leica) with a PISL source by mounting it together with a highly sensitive low- noise CMOS camera. The capabilities of the setup have been demonstrated by carrying out multispectral autofluorescence imaging of live BV2 cells.

110 - Multispectral autofluorescence imaging for cervical cancer screening

110 - Multispectral autofluorescence imaging for cervical cancer screening

Multispectral autofluorescence imaging for detection of cervical lesions: A preclinical study.

The aim of this study was to develop a novel optical imaging system for detecting protoporphyrin IX (PpIX) autofluorescence, to prove that PpIX autofluorescence is as useful as 5-aminolevulinic acid (5-ALA)-induced fluorescence for detecting and localizing cervical cancer, and to monitor the change in PpIX autofluorescence or induced PpIX fluorescence before, during, and after photodynamic therapy (PDT). TC-1 cells - highly tumorigenic cells immortalized using human papillomavirus type 16 proteins E6 and E7 - were subcutaneously grafted into the thighs of nude mice. The suspected tumor tissues were visualized using autofluorescence imaging and induced fluorescence imaging under 5-ALA administration. When the 5-ALA-induced PpIX was sufficiently accumulated in tumor tissues, PDT was performed using a 635-nm laser. We observed the change in fluorescence intensity during PDT. For 3 weeks after PDT, we monitored tumor remission by using white-light imaging and fluorescence imaging. The transplanted cells were visualized by PpIX autofluorescence, which was induced by heme synthesis. After 5-ALA administration, PpIX could be targeted by using PDT, which decreased PpIX autofluorescence. Photobleaching is useful for monitoring PDT dosimetry and for determining the photodynamic response to therapy. PpIX autofluorescence clearly differentiated the tumor from adjacent normal tissues. The results of PpIX autofluorescence imaging and 5-ALA-induced fluorescence imaging were identical. PpIX autofluorescence imaging is a simple and cost-effective cervical cancer screening method that could be performed during or after PDT to ensure effective treatment or remission as a change in fluorescence intensity can be observed in real time without a blinding effect.

Comparison of UV and visible autofluorescence of wheat grain tissues in macroscopic images of cross-sections and particles

Many plant tissues can be observed thanks to the autofluorescence of their cell wall components. Multispectral autofluorescence imaging at the macroscopic scale is a rapid efficient way of observing samples with a large field of view (>1cm2) and a nice resolution (<3μm per pixel). The objective of this work was to evaluate autofluorescence multispectral imaging at the macroscopic scale to identify the tissular origin of particles through their autofluorescence profiles. The pericarp and aleurone layer of wheat grain were used to compare contrasted autofluorescence profiles measured in cross sections and particles. Air and water mounting media were tested. Principal component analysis and variance analysis showed that the autofluorescence properties were retained from cross sections to particles. The mounting media modified the autofluorescence profiles and water was found to be better to observe both particles and sections. These results show that multispectral autofluorescence imaging at the macroscopic scale can be used to identify the histological origin of particles.

Assessment of advanced glycated end product accumulation in skin using auto fluorescence multispectral imaging

Several studies have shown that advanced glycation end products (AGE) play a role in both the microvascular and macrovascular complications of diabetes and are closely linked to inflammation and atherosclerosis. AGEs accumulate in skin and can be detected using their auto fluorescence (AF).A significant correlation exists between AGE AF and the levels of AGEs as obtained from skin biopsies. A commercial device, the AGE Reader, has become available to assess skin AF for clinical purposes but, while displaying promising results, it is limited to single-point measurements performed in contact to skin tissue. Furthermore, in vivo imaging of AGE accumulation is virtually unexplored.We proposed a non-invasive, contact-less novel technique for quantifying fluorescent AGE deposits in skin tissue using a multispectral imaging camera setup (MSI) during ultraviolet (UV) exposure. Imaging involved applying a region-of-interest mask, avoiding specular reflections and a simple calibration. Results of a study conducted on 16 subjects with skin types ranging from fair to deeply pigmented skin, showed that AGE measured with MSI in forearm skin was significantly correlated with the AGE reference method (AGE Reader on forearm skin, R=0.68, p=0.005). AGE measured in facial skin was borderline significantly related to AGE Reader on forearm skin (R=0.47, p=0.078). These results support the use of the technique in devices for non-touch measurement of AGE content in either facial or forearm skin tissue over time.