Subconjunctival hemorrhage (SCH) is a prevalent ocular condition characterized by the accumulation of blood beneath the conjunctiva, resulting in a visible red patch on the eye’s surface. The appearance of SCH and the limited understanding of its progression can cause significant anxiety for patients. To address this issue and enhance ocular health management, we develop a deep learning-enabled monitoring approach that quantitatively tracks the SCH healing process through spectral reconstruction. Our approach comprises two key technical components. Firstly, automatic white balance algorithms are employed to estimate the light source’s color temperature and adjust image colors, minimizing the impact of varying lighting conditions. The SCH segmentation achieves an accuracy of 96.2 %, effectively avoiding interference from skin and eyelashes. Secondly, our monitoring approach evaluates SCH color changes, which are crucial for determining the stage of recovery. By learning a complex mapping function, the approach generates 31 hyperspectral bands (400–700 nm) by recovering the lost spectral information from a given RGB image. This process allows for a more detailed spectroscopic assessment of the affected area. The rich spectral signatures obtained from these hyperspectral images enable the classification of SCH into three distinct stages, reflecting the blood reabsorption process. This study is the first to apply deep learning-based spectral reconstruction to SCH determination, enabling evaluation of the recovery process through spectroscopic and quantitative analysis. This approach has the potential to improve daily patient care and promote better eye health control by offering more comprehensive monitoring of SCH progression.
Meibomian gland dysfunction (MGD) is a significant cause of evaporative dry eye disease, occurring when the meibomian glands (MGs) in the eyelids produce abnormal lipid amounts. MG morphological features are crucial indicators of MG function and dry eye symptoms. However, the relationship between MG morphological irregularities and MGD remains unclear. To address this, we develop an integrated deep-learning-enabled monitoring system within a portable meibography device, enabling early identification and quantification of irregularly-shaped MGs. Our approach comprises two key technical components. First, a customized model is fine-tuned to classify MG irregularities into four types: overlapping, shortening, thickening, and tortuosity. We then quantitatively analyze MG irregularity ratios among four meiboscore groups of varying MG atrophy degrees and examine their connection to Ocular Surface Disease Index (OSDI) indexes from a subjective symptom perspective. From meiboscore 0 to 3, the overlapping MG ratio decreases by 17 %, and the shortening MG ratio increases by 12 %. Furthermore, we’ve built a handheld device equipped with infrared (IR) LED arrays and a USB camera to facilitate long-term and dynamic assessment. This meibography technology is compatible with common operating systems and can be integrated into a smartphone. The high-resolution images captured by this device can be used to assess various types of irregularities. This intelligent portable system offers an automatic and efficient quantitative evaluation of MG morphological irregularities, enabling home inspection and reducing costs. It has the potential to be applied in diagnosing and monitoring MG conditions, facilitating the management of MGD.
The assessment of microplastics (MPs) pollution and water quality monitoring raise a lot of attention in recent years. Discriminative methods are highly needed for quick and accurate in situ MP detections. Digital holography records the wavefront information of the objects and contains the morphology, refractive index, and roughness information. Polarization imaging inspects the optical anisotropy of MPs, which is related to their birefringence and material characteristics. In this work, we explore the capability of holographic and polarization imaging for the identification of MPs. The computed features, such as the angle of polarization (AoP) and degree of linear polarization (DoLP), show distinguishable characteristics of MPs. We inspect the method feasibility on MP classification as well as biological and natural particles. The proposed method shows potential use in real-time, non-contact in situ MPs detection and water pollution monitoring.
Detecting and quantifying microplastic particles have become important problems in environmental monitoring in recent years. In the natural environment, microplastic and nanoplastic particles are often mixed with large pieces of plastic, microalgae, microorganisms, and leaf fragments, etc., making them difficult to be distinguished. In addition, the microplastics themselves are made of different materials and have various shapes. As a result, the conventional classification methods based mostly on morphological characteristics cannot accurately classify microplastics in a complex environment, which brings great challenges to their detection and analysis. We have developed a classification and detection method based on digital holographic imaging and deep learning, which effectively classifies the types of microplastic particles by using the holographic interference fringe features of microplastic particles. With heterogeneous samples containing microplastic particles, microalgae and other substances, we are able to demonstrate the strength of our technique in the detection and characterization of the microplastics. Indeed, the results show that the deep learning network can automatically extract the features of holographic images of different particles in such samples, and delineate with good sensitivity the feature differences in the digital holograms that are caused by optical path differences introduced by various kinds of particles. Furthermore, this holographic feature-based classification is not affected by material morphological characteristics and has good robustness.
Dynamic speckle analysis (DSA) is a non-invasive method to detect movements of the inspected objects. By illuminating the observed sample using a coherent light source, motion information can be obtained from a series of reflecting speckle patterns. Conventional DSA methods record the intensity of the speckle patterns using a frame-based imaging sensor. Here, we propose a novel implementation of DSA using the event sensor which captures the brightness changes of the dynamic speckle patterns with high temporal resolution and low latency. Our method is based on block matching algorithm in which the captured event stream is divided into many non-overlapping blocks and motion information can be computed by searching for the most likely blocks. The experiment results demonstrate the feasibility of our proposed method in different dynamic levels and this work will be beneficial for various applications such as biomedical imaging and material science.
Water scattering is a significant limiting factor for underwater imaging quality. It changes the transportation direction of the original light path, causes the attenuation of light intensity, and so on. In this work, we use a synthetic polarizing camera to capture the images with different polarization states and reduce the impact of water scattering in one step with the underwater light propagation model and the Stokes vector. In addition, an untrained deep network is designed to complete the image descattering processing. Compared with the methods based on deep learning or physical model prior, it is more efficient. This technology is suitable for use in portable underwater imaging optical systems for real-time imaging and detecting particulate matter such as microplastics and microbial particles. It also broadens the application of underwater polarization imaging.
Microplastics, which are a major source of pollution in the ocean, need to be accurately detected and monitored. However, the current detection approaches often require complex optical instrumentation and a long time for image processing. Furthermore, because of the difficulties of particle sampling, it is hard to collect a dataset with sufficient images and a balanced distribution. Digital holography, which is a non-destructive imaging method, is suitable for the in situ imaging. In this work, we propose a novel digital holography microplastics classification system which combines deep learning and generative adversarial networks. We experimentally show that our method yields a higher accuracy for microplastics classification and can efficiently reduce the imbalance ratio of the dataset. This method can be modified for other in situ image classification tasks that likewise suffer from a small and imbalanced distribution dataset.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.