RESUMEN
Glaucoma is a leading cause of irreversible vision impairment globally, and cases are continuously rising worldwide. Early detection is crucial, allowing timely intervention that can prevent further visual field loss. To detect glaucoma an examination of the optic nerve head via fundus imaging can be performed, at the center of which is the assessment of the optic cup and disc boundaries. Fundus imaging is noninvasive and low-cost; however, image examination relies on subjective, time-consuming, and costly expert assessments. A timely question to ask is: "Can artificial intelligence mimic glaucoma assessments made by experts?" Specifically, can artificial intelligence automatically find the boundaries of the optic cup and disc (providing a so-called segmented fundus image) and then use the segmented image to identify glaucoma with high accuracy? We conducted a comprehensive review on artificial intelligence-enabled glaucoma detection frameworks that produce and use segmented fundus images and summarized the advantages and disadvantages of such frameworks. We identified 36 relevant papers from 2011 to 2021 and 2 main approaches: 1) logical rule-based frameworks, based on a set of rules; and 2) machine learning/statistical modeling-based frameworks. We critically evaluated the state-of-art of the 2 approaches, identified gaps in the literature and pointed at areas for future research.
Asunto(s)
Glaucoma , Disco Óptico , Humanos , Inteligencia Artificial , Fondo de Ojo , Glaucoma/diagnóstico , Disco Óptico/diagnóstico por imagen , Aprendizaje AutomáticoRESUMEN
The knuckle creases present on the dorsal side of the human hand can play significant role in identifying the offenders of serious crime, especially when evidence images of more recognizable biometric traits, such as the face, are not available. These knuckle creases, if localized appropriately, can result in improved identification ability. This is attributed to ambient inclusion of the creases and minimal effect of background, which lead to quality and discerning feature extraction. This paper presents an ensemble approach, utilizing multiple object detector frameworks, to localize the knuckle regions in a functionally appropriate way. The approach leverages from the individual capabilities of the popular object detectors and provide a more comprehensive knuckle region localization. The investigations are completed with two large-scale public hand databases which consist of hand-dorsal images with varying backgrounds and finger positioning. In addition to that, effectiveness of the proposed approach is also tested with a novel proprietary unconstrained multi-ethnic hand dorsal dataset to evaluate its generalizability. Several novel performance metrics are tailored to evaluate the efficacy of the proposed knuckle localization approach. These metrics aim to measure the veracity of the detected knuckle regions in terms of their relation with the ground truth. The comparison of the proposed approach with individual object detectors and a state-of-the-art hand keypoint detector clearly establishes the outperforming nature of the proposed approach. The generalization of the proposed approach is also corroborated through the cross-dataset framework.
Asunto(s)
Mano , Articulación Metacarpofalángica , Biometría , Dedos , Mano/anatomía & histología , HumanosRESUMEN
Current research in automated disease detection focuses on making algorithms "slimmer" reducing the need for large training datasets and accelerating recalibration for new data while achieving high accuracy. The development of slimmer models has become a hot research topic in medical imaging. In this work, we develop a two-phase model for glaucoma detection, identifying and exploiting a redundancy in fundus image data relating particularly to the geometry. We propose a novel algorithm for the cup and disc segmentation "EffUnet" with an efficient convolution block and combine this with an extended spatial generative approach for geometry modelling and classification, termed "SpaGen" We demonstrate the high accuracy achievable by EffUnet in detecting the optic disc and cup boundaries and show how our algorithm can be quickly trained with new data by recalibrating the EffUnet layer only. Our resulting glaucoma detection algorithm, "EffUnet-SpaGen", is optimized to significantly reduce the computational burden while at the same time surpassing the current state-of-art in glaucoma detection algorithms with AUROC 0.997 and 0.969 in the benchmark online datasets ORIGA and DRISHTI, respectively. Our algorithm also allows deformed areas of the optic rim to be displayed and investigated, providing explainability, which is crucial to successful adoption and implementation in clinical settings.
RESUMEN
For any single anterior chamber cross-sectional (tomographic) imaging method, there is a practical compromise between image size and image resolution. In order to obtain large field-of-view cross-sectional images of the whole anterior chamber and high-resolution cross-sectional images of the fine corneal layers, measurements by multiple devices are currently required. This paper presents a novel raster scanning tomographic imaging device that acquires simultaneous large field-of-view Scheimpflug (12.5 mm image depth, 50 µm axial resolution in air) and high-resolution spectral domain optical coherence tomography (SD-OCT) (2 mm image depth, 3.7µm axial resolution in air) using the same illuminating photons. For the novel raster scanning 3D Scheimpflug imaging, a tunable lens system together with numerical methods for correcting refraction distortion were used. To demonstrate the capability of simultaneous measurement of both fine corneal layers and whole anterior chambers topology, ex vivo measurements on 12 porcine and 12 bovine eyes were carried out. There is a reasonable agreement in the overall central corneal thicknesses (CCT) obtained from the simultaneous SD-OCT and Scheimpflug measurements. In addition, because the same infrared light beam was used to illuminate the sample, both OCT and Scheimpflug images were taken at the exact same location of a sample simultaneously in a single measurement. This provides a unique method for measuring both the thickness and the refractive index of a sample.
RESUMEN
AIMS/HYPOTHESIS: Corneal confocal microscopy is a rapid non-invasive ophthalmic imaging technique that identifies peripheral and central neurodegenerative disease. Quantification of corneal sub-basal nerve plexus morphology, however, requires either time-consuming manual annotation or a less-sensitive automated image analysis approach. We aimed to develop and validate an artificial intelligence-based, deep learning algorithm for the quantification of nerve fibre properties relevant to the diagnosis of diabetic neuropathy and to compare it with a validated automated analysis program, ACCMetrics. METHODS: Our deep learning algorithm, which employs a convolutional neural network with data augmentation, was developed for the automated quantification of the corneal sub-basal nerve plexus for the diagnosis of diabetic neuropathy. The algorithm was trained using a high-end graphics processor unit on 1698 corneal confocal microscopy images; for external validation, it was further tested on 2137 images. The algorithm was developed to identify total nerve fibre length, branch points, tail points, number and length of nerve segments, and fractal numbers. Sensitivity analyses were undertaken to determine the AUC for ACCMetrics and our algorithm for the diagnosis of diabetic neuropathy. RESULTS: The intraclass correlation coefficients for our algorithm were superior to those for ACCMetrics for total corneal nerve fibre length (0.933 vs 0.825), mean length per segment (0.656 vs 0.325), number of branch points (0.891 vs 0.570), number of tail points (0.623 vs 0.257), number of nerve segments (0.878 vs 0.504) and fractals (0.927 vs 0.758). In addition, our proposed algorithm achieved an AUC of 0.83, specificity of 0.87 and sensitivity of 0.68 for the classification of participants without (n = 90) and with (n = 132) neuropathy (defined by the Toronto criteria). CONCLUSIONS/INTERPRETATION: These results demonstrated that our deep learning algorithm provides rapid and excellent localisation performance for the quantification of corneal nerve biomarkers. This model has potential for adoption into clinical screening programmes for diabetic neuropathy. DATA AVAILABILITY: The publicly shared cornea nerve dataset (dataset 1) is available at http://bioimlab.dei.unipd.it/Corneal%20Nerve%20Tortuosity%20Data%20Set.htm and http://bioimlab.dei.unipd.it/Corneal%20Nerve%20Data%20Set.htm.
Asunto(s)
Aprendizaje Profundo , Neuropatías Diabéticas/fisiopatología , Microscopía Confocal/métodos , Enfermedades Neurodegenerativas/fisiopatología , Algoritmos , Inteligencia Artificial , Córnea/metabolismo , Córnea/patología , Diabetes Mellitus Tipo 1/metabolismo , Diabetes Mellitus Tipo 1/fisiopatología , Humanos , Fibras Nerviosas/metabolismo , Fibras Nerviosas/patologíaRESUMEN
[This corrects the article DOI: 10.1371/journal.pone.0209409.].
RESUMEN
BACKGROUND: Glaucoma is the leading cause of irreversible blindness worldwide. It is a heterogeneous group of conditions with a common optic neuropathy and associated loss of peripheral vision. Both over and under-diagnosis carry high costs in terms of healthcare spending and preventable blindness. The characteristic clinical feature of glaucoma is asymmetrical optic nerve rim narrowing, which is difficult for humans to quantify reliably. Strategies to improve and automate optic disc assessment are therefore needed to prevent sight loss. METHODS: We developed a novel glaucoma detection algorithm that segments and analyses colour photographs to quantify optic nerve rim consistency around the whole disc at 15-degree intervals. This provides a profile of the cup/disc ratio, in contrast to the vertical cup/disc ratio in common use. We introduce a spatial probabilistic model, to account for the optic nerve shape, we then use this model to derive a disc deformation index and a decision rule for glaucoma. We tested our algorithm on two separate image datasets (ORIGA and RIM-ONE). RESULTS: The spatial algorithm accurately distinguished glaucomatous and healthy discs on internal and external validation (AUROC 99.6% and 91.0% respectively). It achieves this using a dataset 100-times smaller than that required for deep learning algorithms, is flexible to the type of cup and disc segmentation (automated or semi-automated), utilises images with missing data, and is correlated with the disc size (p = 0.02) and the rim-to-disc at the narrowest rim (p<0.001, in external validation). DISCUSSION: The spatial probabilistic algorithm is highly accurate, highly data efficient and it extends to any imaging hardware in which the boundaries of cup and disc can be segmented, thus making the algorithm particularly applicable to research into disease mechanisms, and also glaucoma screening in low resource settings.
Asunto(s)
Algoritmos , Diagnóstico por Computador/métodos , Técnicas de Diagnóstico Oftalmológico/estadística & datos numéricos , Glaucoma/diagnóstico por imagen , Diagnóstico por Computador/estadística & datos numéricos , Glaucoma/diagnóstico , Humanos , Modelos Estadísticos , Disco Óptico/diagnóstico por imagen , Nervio Óptico/diagnóstico por imagen , Análisis Espacial , Máquina de Vectores de SoporteRESUMEN
Automotive coating systems are designed to protect vehicle bodies from corrosion and enhance their aesthetic value. The number, size and orientation of small metallic flakes in the base coat of the paint has a significant effect on the appearance of automotive bodies. It is important for quality assurance (QA) to be able to measure the properties of these small flakes, which are approximately 10µm in radius, yet current QA techniques are limited to measuring layer thickness. We design and develop a time-domain (TD) full-field (FF) optical coherence tomography (OCT) system to scan automotive panels volumetrically, non-destructively and without contact. We develop and integrate a segmentation method to automatically distinguish flakes and allow measurement of their properties. We test our integrated system on nine sections of five panels and demonstrate that this integrated approach can characterise small flakes in automotive coating systems in 3D, calculating the number, size and orientation accurately and consistently. This has the potential to significantly impact QA testing in the automotive industry.
RESUMEN
We present in-line coating thickness measurements acquired simultaneously using 2 independent sensing modalities: terahertz pulsed imaging (TPI) and optical coherence tomography (OCT). Both techniques are sufficiently fast to resolve the coating thickness of individual pharmaceutical tablets in situ during the film coating operation, and both techniques are direct structural imaging techniques that do not require multivariate calibration. The TPI sensor is suitable to measure coatings greater than 50 µm and can penetrate through thick coatings even in the presence of pigments over a wide range of excipients. Due to the long wavelength, terahertz radiation is not affected by scattering from dust within the coater. In contrast, OCT can resolve coating layers as thin as 20 µm and is capable of measuring the intratablet coating uniformity and the intertablet coating thickness distribution within the coating pan. However, the OCT technique is less robust when it comes to the compatibility with excipients, dust, and potentially the maximum coating thickness that can be resolved. Using a custom-built laboratory scale coating unit, the coating thickness measurements were acquired independently by the TPI and OCT sensors throughout a film coating operation. Results of the in-line TPI and OCT measurements were compared against one another and validated with off-line TPI and weight gain measurements. Compared with other process analytical technology sensors, such as near-infrared and Raman spectroscopy, the TPI and OCT sensors can resolve the intertablet thickness distribution based on sampling a significant fraction of the tablet populations in the process. By combining 2 complementary sensing modalities, it was possible to seamlessly monitor the coating process over the range of film thickness from 20 µm to greater than 250 µm.
Asunto(s)
Química Farmacéutica/métodos , Comprimidos Recubiertos/síntesis química , Imágen por Terahertz/métodos , Tomografía de Coherencia Óptica/métodos , Química Farmacéutica/instrumentación , Comprimidos Recubiertos/análisis , Imágen por Terahertz/instrumentación , Tomografía de Coherencia Óptica/instrumentaciónRESUMEN
Optical coherence tomography (OCT) can monitor human donor corneas non-invasively during the de-swelling process following storage for corneal transplantation, but currently only resultant thickness as a function of time is extracted. To visualize and quantify the mechanism of de-swelling, we present a method exploiting the nanometer sensitivity of the Fourier phase in OCT data to image deformation velocities. The technique was demonstrated by non-invasively showing during de-swelling that osmotic flow through an intact epithelium is negligible and removing the endothelium approximately doubled the initial flow at that interface. The increased functional data further enabled the validation of a mathematical model of the cornea. Included is an efficient method of measuring high temporal resolution (1 minute demonstrated) corneal thickness, using automated collection and semi-automated graph search segmentation. These methods expand OCT capabilities to measure volume change processes for tissues and materials.
RESUMEN
We report the development of a Spectral Domain Line Field Optical Coherence Tomography (LF-OCT) system, using a broad bandwidth and spatial coherent Super-Continuum (SC) source. With conventional quasi-Continuous Wave (CW) setup we achieve axial resolutions up to 2.1 µm in air and 3D volume imaging speeds up to 213 kA-Scan/s. Furthermore, we report the use of a single SC pulse, of 2 ns duration, to temporally gate an OCT B-Scan image of 70 A-Scans. This is the equivalent of 35 GA-Scans/s. We apply the CW setup for high resolution imaging of the fine structures of a human cornea sample ex-vivo. The single pulse setup is applied to imaging of a coated pharmaceutical tablet. The fixed pattern noise due to spectral noise is removed by subtracting the median magnitude A-Scan. We also demonstrate that the Fourier phase can be used to remove aberration caused artefacts.