Optical Coherence Tomography Image Data Research Articles

Identifying gastric cancer by optical coherence tomography based on an optimized depth-resolved attenuation estimation

Optical coherence tomography (OCT) is a reliable technique for cancer detection, with the potential to improve accuracy in identifying cancerous tissue through effective use of OCT image data. We proposed an optimized depth resolution estimation based optical attenuation coefficient (OAC) mapping method to reduce the error effect caused by discretization without considering whether the light is completely attenuated or not. The attenuation coefficient maps of gastric tissue were reconstructed using this method. Features were extracted from these maps using gray-level co-occurrence matrix and gray histogram, and a support vector machine was employed as a classifier for identifying gastric cancer tissues. The recognition accuracy achieved was 98.60%, which was higher than the 94.30% accuracy in the control group without OACs. This approach enhances the utilization of OCT image information and improves its practical application ability by increasing the recognition accuracy.

Federated learning for diagnosis of age-related macular degeneration.

This paper presents a federated learning (FL) approach to train deep learning models for classifying age-related macular degeneration (AMD) using optical coherence tomography image data. We employ the use of residual network and vision transformer encoders for the normal vs. AMD binary classification, integrating four unique domain adaptation techniques to address domain shift issues caused by heterogeneous data distribution in different institutions. Experimental results indicate that FL strategies can achieve competitive performance similar to centralized models even though each local model has access to a portion of the training data. Notably, the Adaptive Personalization FL strategy stood out in our FL evaluations, consistently delivering high performance across all tests due to its additional local model. Furthermore, the study provides valuable insights into the efficacy of simpler architectures in image classification tasks, particularly in scenarios where data privacy and decentralization are critical using both encoders. It suggests future exploration into deeper models and other FL strategies for a more nuanced understanding of these models' performance. Data and code are available at https://github.com/QIAIUNCC/FL_UNCC_QIAI.

Needle guidance with Doppler-tracked polarization-sensitive optical coherence tomography.

Optical coherence tomography (OCT) can be integrated into needle probes to provide real-time navigational guidance. However, unscanned implementations, which are the simplest to build, often struggle to discriminate the relevant tissues. We explore the use of polarization-sensitive (PS) methods as a means to enhance signal interpretability within unscanned coherence tomography probes. Broadband light from a laser centered at 1310nm was sent through a fiber that was embedded into a needle. The polarization signal from OCT fringes was combined with Doppler-based tracking to create visualizations of the birefringence properties of the tissue. Experiments were performed in (i)well-understood structured tissues (salmon and shrimp) and (ii)ex vivo porcine spine. The porcine experiments were selected to illustrate an epidural guidance use case. In the porcine spine, unscanned and Doppler-tracked PS OCT imaging data successfully identified the skin, subcutaneous tissue, ligament, and epidural spaces during needle insertion. PS imaging within a needle probe improves signal interpretability relative to structural OCT methods and may advance the clinical utility of unscanned OCT needle probes in a variety of applications.

Multimodal Deep Learning Model of Predicting Future Visual Field for Glaucoma Patients

Glaucoma is one of the most common reasons for blindness worldwide, especially in elderly people. Glaucoma can be monitored using visual field (VF) tests. Therefore, predicting the future VF to monitor progression of glaucoma is important. In this paper, we proposed a deep learning model to predict future VF based on previous VF and optical coherence tomography (OCT) images (including thickness map, vertical tomogram, and horizontal tomogram). The image data were analyzed using a ResNet-50 model. Image features and previous VFs were combined, and a long short-term memory (LSTM) network was used to predict future VF. A weighted method was used to detect noisy data. The proposed method was improved when applying weighted loss. The mean absolute error (MAE) was 3.31 ± 1.37, and the root mean square error (RMSE) was 4.58 ± 1.77. The model showed high performance when combining VF data and OCT image data. Furthermore, the model was useful for detecting and re-weighting noisy data.

Viscoelastic Biomechanical Properties of the Conventional Aqueous Outflow Pathway Tissues in Healthy and Glaucoma Human Eyes.

Background: Although the tissues comprising the ocular conventional outflow pathway have shown strong viscoelastic mechanical response to aqueous humor pressure dynamics, the viscoelastic mechanical properties of the trabecular meshwork (TM), juxtacanalicular connective tissue (JCT), and Schlemm’s canal (SC) inner wall are largely unknown. Methods: A quadrant of the anterior segment from two human donor eyes at low- and high-flow (LF and HF) outflow regions was pressurized and imaged using optical coherence tomography (OCT). A finite element (FE) model of the TM, the adjacent JCT, and the SC inner wall was constructed and viscoelastic beam elements were distributed in the extracellular matrix (ECM) of the TM and JCT to represent anisotropic collagen. An inverse FE-optimization algorithm was used to calculate the viscoelastic properties of the ECM/beam elements such that the TM/JCT/SC model and OCT imaging data best matched over time. Results: The ECM of the glaucoma tissues showed significantly larger time-dependent shear moduli compared to the heathy tissues. Significantly larger shear moduli were also observed in the LF regions of both the healthy and glaucoma eyes compared to the HF regions. Conclusions: The outflow tissues in both glaucoma eyes and HF regions are stiffer and less able to respond to dynamic IOP.

Epidural anesthesia needle guidance by forward-view endoscopic optical coherence tomography and deep learning

Epidural anesthesia requires injection of anesthetic into the epidural space in the spine. Accurate placement of the epidural needle is a major challenge. To address this, we developed a forward-view endoscopic optical coherence tomography (OCT) system for real-time imaging of the tissue in front of the needle tip during the puncture. We tested this OCT system in porcine backbones and developed a set of deep learning models to automatically process the imaging data for needle localization. A series of binary classification models were developed to recognize the five layers of the backbone, including fat, interspinous ligament, ligamentum flavum, epidural space, and spinal cord. The classification models provided an average classification accuracy of 96.65%. During puncture, it is important to maintain a safe distance between the needle tip and the dura mater. Regression models were developed to estimate that distance based on the OCT imaging data. Based on the Inception architecture, our models achieved a mean absolute percentage error of 3.05% ± 0.55%. Overall, our results validated the technical feasibility of using this novel imaging strategy to automatically recognize different tissue structures and measure the distances ahead of the needle tip during the epidural needle placement.

Transferability of an Artificial Intelligence Algorithm Predicting Rebubblings After Descemet Membrane Endothelial Keratoplasty.

To develop an artificial intelligence (AI) algorithm enabling corneal surgeons to predict the probability of rebubbling after Descemet membrane endothelial keratoplasty (DMEK) from images obtained using optical coherence tomography (OCT). Anterior segment OCT data of patients undergoing DMEK by 2 different DMEK surgeons (C.C. and B.B.; University of Cologne, Cologne, Germany) were extracted from the prospective Cologne DMEK database. An AI algorithm was trained by using a data set of C.C. to detect graft detachments and predict the probability of a rebubbling. The architecture of the AI model used in this study was called EfficientNet. This algorithm was applied to OCT scans of patients, which were operated by B.B. The transferability of this algorithm was analyzed to predict a rebubbling after DMEK. The algorithm reached an area under the curve of 0.875 (95% confidence interval: 0.880-0.929). The cutoff value based on the Youden index was 0.214, and the sensitivity and specificity for this value were 78.9% (67.6%-87.7%) and 78.6% (69.5%-86.1%). The development of AI algorithms allows good transferability to other surgeons reaching a high accuracy in predicting rebubbling after DMEK based on OCT image data.

Invited Session I: Artificial intelligence applications in ophthalmology and vision science: AI applications in glaucoma

Glaucoma is the second cause of blindness worldwide. Due to the complicated non-linear relationships between structural and functional assessment outcomes together with the large variabilities of disease progression patterns, accurate and reliable clinical assessment is a paramount importance for this slow but irreversible progressing disease. Recent advances in medical applications of artificial intelligence, especially deep learning (DL) approaches, have opened up unprecedented possibilities in computer aided clinical care. I will discuss a variety of DL studies on our large longitudinal glaucoma cohort data in collaborating IBM Watson Research Team. The topics include: 1) estimation of visual field parameters out of raw 3D optical coherence tomography (OCT) image data using a feature agnostic data driven approach, 2) forecasting functional measurements out of clinically available demographic information augmented by OCT feature analysis, 3) generating future 2D biomarker color mapping on OCT image data, and 4) identifying novel biomarkers and exploration of structure-function relationships using a group class activation mapping technique.

Benchmarking automated detection of the retinal external limiting membrane in a 3D spectral domain optical coherence tomography image dataset of full thickness macular holes

In this article, we present a new benchmark for the segmentation of the retinal external limiting membrane (ELM) using an image dataset of spectral domain optical coherence tomography (OCT) scans in a patient population with idiopathic full-thickness macular holes. Specifically, the dataset used contains OCT images from one eye of 107 patients with an idiopathic full-thickness macular hole. In total, the dataset contains 5243 individual 2-dimensional (2-D) OCT image slices, with each patient contributing 49 individual spectral-domain OCT tagged image slices. We display precise image-wise binary annotations to segment the ELM line. The OCT images present high variations in image contrast, motion, brightness, and speckle noise which can affect the robustness of applied algorithms, so we performed an extensive OCT imaging and annotation data quality analysis. Imaging data quality control included noise, blurriness and contrast scores, motion estimation, darkness and average pixel scores, and anomaly detection. Annotation quality was measured using gradient mapping of ELM line annotation confidence, and idiopathic full-thickness macular hole detection. Finally, we compared qualitative and quantitative results with seven state-of-the-art machine learning-based segmentation methods to identify the ELM line with an automated system.

Unsupervised domain adaptation model for lesion detection in retinal OCT images

Background and objective. Optical coherence tomography (OCT) is one of the most used retinal imaging modalities in the clinic as it can provide high-resolution anatomical images. The huge number of OCT images has significantly advanced the development of deep learning methods for automatic lesion detection to ease the doctor’s workload. However, it has been frequently revealed that the deep neural network model has difficulty handling the domain discrepancies, which widely exist in medical images captured from different devices. Many works have been proposed to solve the domain shift issue in deep learning tasks such as disease classification and lesion segmentation, but few works focused on lesion detection, especially for OCT images. Methods. In this work, we proposed a faster-RCNN based, unsupervised domain adaptation model to address the lesion detection task in cross-device retinal OCT images. The domain shift is minimized by reducing the image-level shift and instance-level shift at the same time. We combined a domain classifier with a Wasserstein distance critic to align the shifts at each level. Results. The model was tested on two sets of OCT image data captured from different devices, obtained an average accuracy improvement of more than 8% over the method without domain adaptation, and outperformed other comparable domain adaptation methods. Conclusion. The results demonstrate the proposed model is more effective in reducing the domain shift than advanced methods.

Inner retinal degeneration associated with optic nerve head drusen in pseudoxanthoma elasticum

Background/aimsTo determine the association of age, presence of optic nerve head drusen (ONHD) and number of previous intravitreal anti-vascular endothelial growth factor (anti-VEGF) injections with inner retinal layer thicknesses in...

Impact of stent malapposition on intracoronary flow dynamics: An optical coherence tomography-based patient-specific study

Percutaneous coronary intervention with stent implantation has emerged as a popular approach to treat coronary artery stenosis. Stent malapposition (SM), also referred as incomplete stent apposition, could reduce stent tissue coverage and hence increase the risk of late stent thrombosis. The objective of this study was to investigate the impact of SM on intracoronary flow dynamics by combining optical coherence tomography (OCT) image-based model reconstruction and computational analysis. Firstly, a stenosed coronary artery model was reconstructed from OCT and angiography imaging data of a patient. Two structural analyses were carried out to simulate two types of coronary artery stent implantations: a fully-apposed (FA) case and a SM case. Then, based on the two deformed coronary geometries, two computational fluid dynamics (CFD) analyses were performed to evaluate the differences of hemodynamic metrics between the FA and the SM cases, including wall shear stress (WSS), time-averaged WSS (TWSS), oscillatory shear index (OSI), WSS gradient (WSSG), time-averaged WSSG (TWSSG), and relative residence time (RRT). The results indicated that maximum flow velocity was higher in the SM case than that of the FA case, due to the incomplete expansion of the stent and artery. Moreover, the SM case had a lower percentage of areas of adverse WSS (< 0.5 Pa) and RRT (> 10/Pa) but a higher percentage of areas of adverse OSI (> 0.1) and WSSG (> 5000 Pa/m). Specifically, the differences of OSI, WSSG, and RRT between the two cases were relatively small. It was suggested that SM might not be responsible for negative hemodynamic metrics which would further result in stent thrombosis on the basis of the present specific model.

Generation of Augmented Capillary Network Optical Coherence Tomography Image Data of Human Skin for Deep Learning and Capillary Segmentation.

The segmentation of capillaries in human skin in full-field optical coherence tomography (FF-OCT) images plays a vital role in clinical applications. Recent advances in deep learning techniques have demonstrated a state-of-the-art level of accuracy for the task of automatic medical image segmentation. However, a gigantic amount of annotated data is required for the successful training of deep learning models, which demands a great deal of effort and is costly. To overcome this fundamental problem, an automatic simulation algorithm to generate OCT-like skin image data with augmented capillary networks (ACNs) in a three-dimensional volume (which we called the ACN data) is presented. This algorithm simultaneously acquires augmented FF-OCT and corresponding ground truth images of capillary structures, in which potential functions are introduced to conduct the capillary pathways, and the two-dimensional Gaussian function is utilized to mimic the brightness reflected by capillary blood flow seen in real OCT data. To assess the quality of the ACN data, a U-Net deep learning model was trained by the ACN data and then tested on real in vivo FF-OCT human skin images for capillary segmentation. With properly designed data binarization for predicted image frames, the testing result of real FF-OCT data with respect to the ground truth achieved high scores in performance metrics. This demonstrates that the proposed algorithm is capable of generating ACN data that can imitate real FF-OCT skin images of capillary networks for use in research and deep learning, and that the model for capillary segmentation could be of wide benefit in clinical and biomedical applications.

Deep convolutional neural networks: Outperforming established algorithms in the evaluation of industrial optical coherence tomography (OCT) images of pharmaceutical coatings

This paper presents a novel evaluation approach for optical coherence tomography (OCT) image analysis of pharmaceutical solid dosage forms based on deep convolutional neural networks (CNNs). As a proof of concept, CNNs were applied to image data from both, in- and at-line OCT implementations, monitoring film-coated tablets as well as single- and multi-layered pellets. CNN results were compared against results from established algorithms based on ellipse-fitting, as well as to human-annotated ground truth data. Performance benchmarks used include, efficiency (computation speed), sensitivity (number of detections from a defined test set) and accuracy (deviation from the reference method). The results were validated by comparing the output of several algorithms to data manually annotated by human experts and microscopy images of cross-sectional cuts of the same dosage forms as a reference method. In order to guarantee comparability for all results, the algorithms were executed on the same hardware. Since modern OCT systems must operate under real-time conditions in order to be implemented in-line into manufacturing lines, the necessary steps are discussed on how to achieve this goal without sacrificing the algorithmic performance and how to tailor a deep CNN to cope with the high amount of image noise and alterations in object appearance. The developed deep learning approach outperforms static algorithms currently available in pharma applications with respect to performance benchmarks, and represents the next level in real time evaluation of challenging industrial OCT image data.

Applications and future directions for optical coherence tomography in dermatology*

Optical coherence tomography (OCT) is a noninvasive optical imaging method that can generate high-resolution en face and cross-sectional images of the skin in vivo to a maximum depth of 2mm. While OCT holds considerable potential for noninvasive diagnosis and disease monitoring, it is poorly understood by many dermatologists. Here we aim to equip the practising dermatologist with an understanding of the principles of skin OCT and the potential clinical indications. We begin with an introduction to the technology and discuss the different modalities of OCT including angiographic (dynamic) OCT, which can image cutaneous blood vessels at high resolution. Next we review clinical applications. OCT has been most extensively investigated in the diagnosis of keratinocyte carcinomas, particularly basal cell carcinoma. To date, OCT has not proven sufficiently accurate for the robust diagnosis of malignant melanoma; however, the evaluation of abnormal vasculature with angiographic OCT is an area of active investigation. OCT, and in particular angiographic OCT, also shows promise in monitoring the response to therapy of inflammatory dermatoses, such as psoriasis and connective tissues disease. We additionally discuss a potential role for artificial intelligence in improving the accuracy of interpretation of OCT imaging data.

Flexible A-scan rate MHz-OCT: efficient computational downscaling by coherent averaging.

In order to realize adjustable A-scan rates of fast optical coherence tomography (OCT) systems, we investigate averaging of OCT image data acquired with a MHz-OCT system based on a Fourier Domain Mode Locked (FDML) laser. Increased system sensitivity and image quality can be achieved with the same system at the cost of lower imaging speed. Effectively, the A-scan rate can be reduced in software by a freely selectable factor. We demonstrate a detailed technical layout of the strategies necessary to achieve efficient coherent averaging. Since there are many new challenges specific to coherent averaging in swept source MHz-OCT, we analyze them point by point and describe the appropriate solutions. We prove that coherent averaging is possible at MHz OCT-speed without special interferometer designs or digital phase stabilization. We find, that in our system up to ∼100x coherent averaging is possible while achieving a sensitivity increase close to the ideal values. This corresponds to a speed reduction from 3.3 MHz to 33 kHz and a sensitivity gain of 20 dB. We show an imaging comparison between coherent and magnitude averaging of a human finger knuckle joint in vivo with 121dB sensitivity for the coherent case. Further, the benefits of computational downscaling in low sensitivity MHz-OCT systems are analyzed.

Unbiased identification of novel subclinical imaging biomarkers using unsupervised deep learning

Artificial intelligence has recently made a disruptive impact in medical imaging by successfully automatizing expert-level diagnostic tasks. However, replicating human-made decisions may inherently be biased by the fallible and dogmatic nature of human experts, in addition to requiring prohibitive amounts of training data. In this paper, we introduce an unsupervised deep learning architecture particularly designed for OCT representations for unbiased, purely data-driven biomarker discovery. We developed artificial intelligence technology that provides biomarker candidates without any restricting input or domain knowledge beyond raw images. Analyzing 54,900 retinal optical coherence tomography (OCT) volume scans of 1094 patients with age-related macular degeneration, we generated a vocabulary of 20 local and global markers capturing characteristic retinal patterns. The resulting markers were validated by linking them with clinical outcomes (visual acuity, lesion activity and retinal morphology) using correlation and machine learning regression. The newly identified features correlated well with specific biomarkers traditionally used in clinical practice (r up to 0.73), and outperformed them in correlating with visual acuity (hbox {R}^{2} = 0.46 compared to hbox {R}^{2} = 0.29 for conventional markers), despite representing an enormous compression of OCT imaging data (67 million voxels to 20 features). In addition, our method also discovered hitherto unknown, clinically relevant biomarker candidates. The presented deep learning approach identified known as well as novel medical imaging biomarkers without any prior domain knowledge. Similar approaches may be worthwhile across other medical imaging fields.

Otitis Media Middle Ear Effusion Identification and Characterization Using an Optical Coherence Tomography Otoscope.

To determine the feasibility of detecting and differentiating middle ear effusions (MEEs) using an optical coherence tomography (OCT) otoscope. Cross-sectional study. US tertiary care children's hospital. Seventy pediatric patients undergoing tympanostomy tube placement were preoperatively imaged using an OCT otoscope. A blinded reader quiz was conducted using 24 readers from 4 groups of tiered medical expertise. The primary outcome assessed was reader ability to detect presence/absence of MEE. A secondary outcome assessed was reader ability to differentiate serous vs nonserous MEE. OCT image data sets were analyzed from 45 of 70 total subjects. Blinded reader analysis of an OCT data subset for detection of MEE resulted in 90.6% accuracy, 90.9% sensitivity, 90.2% specificity, and intra/interreader agreement of 92.9% and 87.1%, respectively. Differentiating MEE type, reader identification of nonserous MEE had 70.8% accuracy, 53.6% sensitivity, 80.1% specificity, and intra/interreader agreement of 82.9% and 75.1%, respectively. Multivariate analysis revealed that age was the strongest predictor of OCT quality. The mean age of subjects with quality OCT was 5.01 years (n = 45), compared to 2.54 years (n = 25) in the remaining subjects imaged (P = .0028). The ability to capture quality images improved over time, from 50% to 69.4% over the study period. OCT otoscopy shows promise for facilitating accurate MEE detection. The imageability with the prototype device was affected by age, with older children being easier to image, similar to current ear diagnostic technologies.

Imaging of subendocardial adipose tissue and fiber orientation distributions in the human left atrium using optical coherence tomography.

BackgroundOptical coherence tomography (OCT) has the potential to provide real‐time imaging guidance for atrial fibrillation ablation, with promising results for lesion monitoring. OCT can also offer high‐resolution imaging of tissue composition, but there is insufficient cardiac OCT data to inform the use of OCT to reveal important tissue architecture of the human left atrium. Thus, the objective of this study was to define OCT imaging data throughout the human left atrium, focusing on the distribution of adipose tissue and fiber orientation as seen from the endocardium.Methods and ResultsHuman hearts (n = 7) were acquired for imaging the left atrium with OCT. A spectral‐domain OCT system with 1325 nm center wavelength, 6.5 μm axial resolution, 15 μm lateral resolution, and a maximum imaging depth of 2.51 mm in the air was used. Large‐scale OCT image maps of human left atrial tissue were developed, with adipose thickness and fiber orientation extracted from the imaging data. OCT imaging showed scattered distributions of adipose tissue around the septal and pulmonary vein regions, up to a depth of about 0.43 mm from the endocardial surface. The total volume of adipose tissue detected by OCT over one left atrium ranged from 1.42 to 28.74 mm3. Limited fiber orientation information primarily around the pulmonary veins and the septum could be identified.ConclusionOCT imaging could provide adjunctive information on the distribution of subendocardial adipose tissue, particularly around thin areas around the pulmonary veins and septal regions. Variations in OCT‐detected tissue composition could potentially assist ablation guidance.

Association Between Perifoveal Drusen Burden Determined by OCT and Genetic Risk in Early and Intermediate Age-Related Macular Degeneration.

PurposeThe purpose of this study was to determine associations between macular drusen parameters derived from an automatic optical coherence tomography (OCT) algorithm, nonadvanced age-related macular degeneration (AMD) stage, and genetic variants.MethodsEyes classified as early or intermediate AMD with OCT imaging and genetic data were selected (n = 239 eyes). Drusen area and volume measurements were estimated using the Zeiss Cirrus advanced retinal pigment epithelium analysis algorithm in a perifoveal zone centered on the fovea. Associations between drusen measurements and common genetic variants in the complement and high-density lipoprotein (HDL) lipid pathways and the ARMS2/HTRA1 variant were calculated using generalized estimating equations and linear mixed models adjusting for age, sex, smoking, body mass index, and education.ResultsDrusen area ≥ the median was independently associated with a higher number of risk alleles for CFH risk score and risk variants in C3 and ARMS2/HTRA1 compared with eyes with no measurable drusen. Similar results were obtained for drusen volume. When all genes were analyzed in the same model, only CFH score and ARMS2/HTRA1 were associated with drusen measurements. HDL pathway genes were not significantly related to drusen parameters. Nonadvanced AMD stages were associated with OCT-derived drusen area and volume.ConclusionsVariants in CFH and ARMS2/HTRA1, commonly associated with advanced AMD, were independently associated with an increase in drusen burden determined by OCT in an allele dose dependent manner, in eyes with early and intermediate AMD. Biomarkers such as a quantitative classification of nonadvanced AMD and other OCT-derived subphenotypes could provide earlier anatomic endpoints for clinical trials and facilitate the development of new therapies for AMD.