Deep learning-based automatic meibomian gland segmentation and morphology assessment in infrared meibography.
Ontology highlight
ABSTRACT:
SUBMITTER: Setu MAK
PROVIDER: S-EPMC8027879 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:BackgroundIn the present study, we evaluated the correlation between meibomian gland dropout and meibum quality in the same central 8 meibomian glands of the eyelid.MethodsNinety-nine eyes of 91 patients with dry eye were included in the study. Dropout of the 8 central meibomian glands of the eyelids was graded as 0, 1, 2, or 3, according to the dropout area. The meibum quality was graded as follows: grade 0, no secretion; 1, inspissated/toothpaste consistency; 2, cloudy liquid secretion; and 3, clear liquid secretion. For 68 eyes of 68 patients, correlation analysis between dropout and meibum quality was performed. To precisely analyze the direct correlation between meibomian gland dropout in meibography and meibum quality, we evaluated 31 eyes of 23 patients with focal dropout in meibography.ResultsThe median (interquartile range) meiboscore was 1.0 (2.0) in the upper eyelids and 0.0 (1.0) in the lower eyelids. The median (interquartile range) meibum quality grade was 3.0 (1.0) in the upper eyelids and 1.0 (1.0) in the lower eyelids. No significant correlation between the meiboscore and meibum quality grade was detected in the upper (p =0.746) or lower (p =0.551) eyelids. Analysis of the direct correlation between meibomian gland dropout in meibography and meibum quality in patients with focal dropout (loss of 1 or 2 adjacent meibomian glands), however, indicated that meibomian glands with dropout secreted little to no meibum.ConclusionsOverall analysis revealed no relationship between meibomian gland dropout and meibum quality, but more detailed investigation of each meibomian gland alone revealed that meibomian glands with dropout secrete little to no meibum.
| S-EPMC9252080 | biostudies-literature
Project description:Thermal pulsation (LipiFlow) has been advocated for meibomian gland dysfunction (MGD) treatment and was found useful. We aimed to evaluate the efficacy and safety of thermal pulsation in Asian patients with different grades of meibomian gland loss.A hospital-based interventional study comparing thermal pulsation to warm compresses for MGD treatment. Fifty patients were recruited from the dry eye clinic of a Singapore tertiary eye hospital. The ocular surface and symptom were evaluated before treatment, and one and three months after treatment. Twenty-five patients underwent thermal pulsation (single session), whereas 25 patients underwent warm compresses (twice daily) for 3 months. Meibomian gland loss was graded using infrared meibography, whereas function was graded using the number of glands with liquid secretion.The mean age (SD) of participants was 56.4 (11.4) years in the warm compress group and 55.6 (12.7) years in the thermal pulsation group. Seventy-six percent of the participants were female. Irritation symptom significantly improved over 3 months in both groups (P<0.01), whereas tear breakup time (TBUT) was modestly improved at 1 month in only the thermal pulsation group (P=0.048), without significant difference between both groups over the 3 months (P=0.88). There was also no significant difference in irritation symptom, TBUT, Schirmer test, and gland secretion variables between patients with different grades of gland loss or function at follow-ups.A single session of thermal pulsation was similar in its efficacy and safety profile to 3 months of twice daily warm compresses in Asians. Treatment efficacy was not affected by pretreatment gland loss.
| S-EPMC5098463 | biostudies-literature
Project description:PurposeTo evaluate the validity and reliability of alternative methods for evaluating meibomian gland (MG) dropout without using an infrared light system: red-filtered or isolated red-channel images (RCIs) of the everted eyelid.MethodsWe evaluated MG dropout in the everted upper and lower eyelids of 125 eyes of 64 patients with good-quality infrared meibography images (IMIs) and color digital photographs with and without a red filter. Red-filtered images (RFIs) were converted to black and white and adjusted for contrast/brightness [adjusted red-filtered images (aRFIs)]. RCIs were computationally isolated from color digital photographs obtained without a red filter. After randomization, the total meiboscore (0-6) was evaluated by 2 independent evaluators (interobserver reliability) masked to the image origin, and again after a 30-day interval (intraobserver reliability).ResultsThe meiboscores evaluated using the RFI, aRFI, and RCI were strongly positively correlated with those evaluated using the IMI (RFI: ρ = 0.788; aRFI: ρ = 0.735; RCI: ρ = 0.630; all P < 0.001, Spearman correlation analysis). Linear-weighted κ-values (κw) showed substantial agreement between the RFI and IMI (κw = 0.676, 95% CI = 0.594-0.759). The RFI had substantial intraobserver reliability (κw = 0.735, 95% CI = 0.685-0.785) and moderate interobserver reliability (κw = 0.467, 95% CI = 0.371-0.563). Computational adjustment of RFIs did not enhance the validity or reliability, and RCIs had limitations in some cases.ConclusionsMGs were successfully visualized using a red filter on a slit lamp and showed substantial agreement with visualization using the standard infrared method. Although interobserver reliability was only moderate, this alternative technique may be useful for evaluating MG dropout when an infrared meibography device is not available.
| S-EPMC6455092 | biostudies-literature
Project description:Automatic medical image segmentation plays a critical role in scientific research and medical care. Existing high-performance deep learning methods typically rely on large training datasets with high-quality manual annotations, which are difficult to obtain in many clinical applications. Here, we introduce Annotation-effIcient Deep lEarning (AIDE), an open-source framework to handle imperfect training datasets. Methodological analyses and empirical evaluations are conducted, and we demonstrate that AIDE surpasses conventional fully-supervised models by presenting better performance on open datasets possessing scarce or noisy annotations. We further test AIDE in a real-life case study for breast tumor segmentation. Three datasets containing 11,852 breast images from three medical centers are employed, and AIDE, utilizing 10% training annotations, consistently produces segmentation maps comparable to those generated by fully-supervised counterparts or provided by independent radiologists. The 10-fold enhanced efficiency in utilizing expert labels has the potential to promote a wide range of biomedical applications.
| S-EPMC8501087 | biostudies-literature
Project description:The analysis of the choroid in the eye is crucial for our understanding of a range of ocular diseases and physiological processes. Optical coherence tomography (OCT) imaging provides the ability to capture highly detailed cross-sectional images of the choroid yet only a very limited number of commercial OCT instruments provide methods for automatic segmentation of choroidal tissue. Manual annotation of the choroidal boundaries is often performed but this is impractical due to the lengthy time taken to analyse large volumes of images. Therefore, there is a pressing need for reliable and accurate methods to automatically segment choroidal tissue boundaries in OCT images. In this work, a variety of patch-based and fully-convolutional deep learning methods are proposed to accurately determine the location of the choroidal boundaries of interest. The effect of network architecture, patch-size and contrast enhancement methods was tested to better understand the optimal architecture and approach to maximize performance. The results are compared with manual boundary segmentation used as a ground-truth, as well as with a standard image analysis technique. Results of total retinal layer segmentation are also presented for comparison purposes. The findings presented here demonstrate the benefit of deep learning methods for segmentation of the chorio-retinal boundary analysis in OCT images.
| S-EPMC6746702 | biostudies-literature
Project description:The purpose of this study was to determine whether a deep-learning-based assessment system could facilitate preoperative grading of meningioma. This was a retrospective study conducted at two institutions covering 643 patients. The system, designed with a cascade network structure, was developed using deep-learning technology for automatic tumor detection, visual assessment, and grading prediction. Specifically, a modified U-Net convolutional neural network was first established to segment tumor images. Subsequently, the segmentations were introduced into rendering algorithms for spatial reconstruction and another DenseNet convolutional neural network for grading prediction. The trained models were integrated as a system, and the robustness was tested based on its performance on an external dataset from the second institution involving different magnetic resonance imaging platforms. The results showed that the segment model represented a noteworthy performance with dice coefficients of 0.920 ± 0.009 in the validation group. With accurate segmented tumor images, the rendering model delicately reconstructed the tumor body and clearly displayed the important intracranial vessels. The DenseNet model also achieved high accuracy with an area under the curve of 0.918 ± 0.006 and accuracy of 0.901 ± 0.039 when classifying tumors into low-grade and high-grade meningiomas. Moreover, the system exhibited good performance on the external validation dataset.
| S-EPMC8401675 | biostudies-literature
Project description:BackgroundMeibomian gland dysfunction (MGD), one of the most common ocular surface diseases, can induce dry eye and reduce patients' quality of life. Methodological limitations have resulted in contradictory interpretations of gland function. This study sought to investigate the correlation between meibography signal intensity (SI) and meibomian gland (MG) function and to validate an MGD classification strategy based on different levels of SI.MethodsA multicenter, cross-sectional analysis was conducted on 817 eyes from 361 patients with MGD and 52 healthy controls. Additionally, 78 eyes from 39 patients with MGD who had undergone LipiFlow treatment were recruited for longitudinal analyses. The SI value was obtained via meibography using an automated analyzer, and all participants underwent ocular surface examinations. A cross-sectional analysis was performed to determine SI distribution and its relationship to clinical characteristics via a generalized estimating equation model. Longitudinal analyses were conducted on the treatment cohort using a mixed-effects model to explore the outcome in different SI levels.ResultsRegression analysis revealed significant correlations between SI and lipid layer thickness (β=0.016), meibum expressibility (β=-0.676), meibum quality (β=-0.251), and fluorescein-stained tear-film break-up time (FBUT) (β=0.064) (all P values <0.001 for the above associations). Low-level SI MGD cases exhibited the most severe clinical signs, including the worst meibum expressibility (16% for level 3) and quality scores (19% for level 3), the shortest FBUT (3.82±0.13 s), and the thinnest lipid layer (65.68±2.58 nm), (all P values <0.05, respectively). Patients with medium SI showed the lowest ocular surface disease index (OSDI) value (26.64±1.06), the longest FBUT (4.56±0.08 s), and the thickest lipid layer (80.20±2.90 nm). After treatment, the high SI values reduced significantly at each follow-up point compared to baseline (all P values <0.05). The medium SI group demonstrated the greatest improvement in symptoms and signs, followed by the high SI group, and the low SI group.ConclusionsAutomated measurements of SI can effectively reflect MG secretory activity. The proposed low, medium, and high SI classifications represent different functional subtypes of MGD.
| S-EPMC11320511 | biostudies-literature
Project description:The study objective was to investigate the performance of a dedicated convolutional neural network (CNN) optimized for wrist cartilage segmentation from 2D MR images. CNN utilized a planar architecture and patch-based (PB) training approach that ensured optimal performance in the presence of a limited amount of training data. The CNN was trained and validated in 20 multi-slice MRI datasets acquired with two different coils in 11 subjects (healthy volunteers and patients). The validation included a comparison with the alternative state-of-the-art CNN methods for the segmentation of joints from MR images and the ground-truth manual segmentation. When trained on the limited training data, the CNN outperformed significantly image-based and PB-U-Net networks. Our PB-CNN also demonstrated a good agreement with manual segmentation (Sørensen-Dice similarity coefficient [DSC] = 0.81) in the representative (central coronal) slices with a large amount of cartilage tissue. Reduced performance of the network for slices with a very limited amount of cartilage tissue suggests the need for fully 3D convolutional networks to provide uniform performance across the joint. The study also assessed inter- and intra-observer variability of the manual wrist cartilage segmentation (DSC = 0.78-0.88 and 0.9, respectively). The proposed deep learning-based segmentation of the wrist cartilage from MRI could facilitate research of novel imaging markers of wrist osteoarthritis to characterize its progression and response to therapy.
| S-EPMC7784718 | biostudies-literature
Project description:The choroid layer is a vascular layer in human retina and its main function is to provide oxygen and support to the retina. Various studies have shown that the thickness of the choroid layer is correlated with the diagnosis of several ophthalmic diseases. For example, diabetic macular edema (DME) is a leading cause of vision loss in patients with diabetes. Despite contemporary advances, automatic segmentation of the choroid layer remains a challenging task due to low contrast, inhomogeneous intensity, inconsistent texture and ambiguous boundaries between the choroid and sclera in Optical Coherence Tomography (OCT) images. The majority of currently implemented methods manually or semi-automatically segment out the region of interest. While many fully automatic methods exist in the context of choroid layer segmentation, more effective and accurate automatic methods are required in order to employ these methods in the clinical sector. This paper proposed and implemented an automatic method for choroid layer segmentation in OCT images using deep learning and a series of morphological operations. The aim of this research was to segment out Bruch's Membrane (BM) and choroid layer to calculate the thickness map. BM was segmented using a series of morphological operations, whereas the choroid layer was segmented using a deep learning approach as more image statistics were required to segment accurately. Several evaluation metrics were used to test and compare the proposed method against other existing methodologies. Experimental results showed that the proposed method greatly reduced the error rate when compared with the other state-of-the-art methods.
| S-EPMC6395677 | biostudies-literature
Project description:Background: In the study of early cardiac development, it is essential to acquire accurate volume changes of the heart chambers. Although advanced imaging techniques, such as light-sheet fluorescent microscopy (LSFM), provide an accurate procedure for analyzing the heart structure, rapid, and robust segmentation is required to reduce laborious time and accurately quantify developmental cardiac mechanics. Methods: The traditional biomedical analysis involving segmentation of the intracardiac volume occurs manually, presenting bottlenecks due to enormous data volume at high axial resolution. Our advanced deep-learning techniques provide a robust method to segment the volume within a few minutes. Our U-net-based segmentation adopted manually segmented intracardiac volume changes as training data and automatically produced the other LSFM zebrafish cardiac motion images. Results: Three cardiac cycles from 2 to 5 days postfertilization (dpf) were successfully segmented by our U-net-based network providing volume changes over time. In addition to understanding each of the two chambers' cardiac function, the ventricle and atrium were separated by 3D erode morphology methods. Therefore, cardiac mechanical properties were measured rapidly and demonstrated incremental volume changes of both chambers separately. Interestingly, stroke volume (SV) remains similar in the atrium while that of the ventricle increases SV gradually. Conclusion: Our U-net-based segmentation provides a delicate method to segment the intricate inner volume of the zebrafish heart during development, thus providing an accurate, robust, and efficient algorithm to accelerate cardiac research by bypassing the labor-intensive task as well as improving the consistency in the results.
| S-EPMC8221393 | biostudies-literature