Determination of probability of causative pathogen in infectious keratitis using deep learning algorithm of slit-lamp images
Ontology highlight
ABSTRACT:
SUBMITTER: Koyama A
PROVIDER: S-EPMC8608802 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:In this study, we aimed to develop a deep learning model for identifying bacterial keratitis (BK) and fungal keratitis (FK) by using slit-lamp images. We retrospectively collected slit-lamp images of patients with culture-proven microbial keratitis between 1 January 2010 and 31 December 2019 from two medical centers in Taiwan. We constructed a deep learning algorithm consisting of a segmentation model for cropping cornea images and a classification model that applies different convolutional neural networks (CNNs) to differentiate between FK and BK. The CNNs included DenseNet121, DenseNet161, DenseNet169, DenseNet201, EfficientNetB3, InceptionV3, ResNet101, and ResNet50. The model performance was evaluated and presented as the area under the curve (AUC) of the receiver operating characteristic curves. A gradient-weighted class activation mapping technique was used to plot the heat map of the model. By using 1330 images from 580 patients, the deep learning algorithm achieved the highest average accuracy of 80.0%. Using different CNNs, the diagnostic accuracy for BK ranged from 79.6% to 95.9%, and that for FK ranged from 26.3% to 65.8%. The CNN of DenseNet161 showed the best model performance, with an AUC of 0.85 for both BK and FK. The heat maps revealed that the model was able to identify the corneal infiltrations. The model showed a better diagnostic accuracy than the previously reported diagnostic performance of both general ophthalmologists and corneal specialists.
| S-EPMC8307675 | biostudies-literature
Project description:PURPOSE Corneal infections are a leading cause of visual impairment and blindness worldwide. Here we applied high-resolution transcriptomic profiling to assess the general and pathogen-specific molecular and cellular mechanisms during human corneal infection. METHODS Clinical diagnoses of herpes simplex virus (HSV) (n=5) and bacterial/fungal (n=5) keratitis were confirmed by histology. Healthy corneas (n=7) and keratoconus (n=4) samples served as controls. Formalin-fixed, paraffin-embedded (FFPE) human corneal specimens were analyzed using the 3' RNA sequencing method Massive Analysis of cDNA Ends (MACE RNA-seq). The cellular host response was investigated using comprehensive bioinformatic deconvolution (xCell) analyses and by integration with single cell RNA-seq data. RESULTS Our analysis identified 216 and 561 genes, that were specifically overexpressed in viral or bacterial/fungal keratitis, respectively, and allowed to distinguish the two etiologies. The virus-specific host response was driven by adaptive immunity and associated molecular signaling pathways, whereas the bacterial/fungal-specific host response mainly involved innate immunity signaling pathways and cell types. We identified several molecular mechanisms involved in the host response to infectious keratitis, including CXCL9, CXCR3, and MMP9 for viral, and S100A8/A9, MMP9, and the IL17 pathway for bacterial/fungal keratitis. CONCLUSIONS High-resolution molecular profiling provides new insights into the human corneal host response to viral and bacterial/fungal infection. Pathogen-specific molecular profiles may provide the foundation for novel diagnostic biomarker and therapeutic approaches that target inflammation-induced damage to corneal host cells with the goal to improve the outcome of infectious keratitis.
2024-01-30 | GSE241715 | GEO
Project description:PurposeTo investigate the sources of measurement variability when quantifying the morphology of microbial keratitis (MK) from slit-lamp photography (SLP) images using a semiautomated, image-analysis algorithm.MethodsProspectively enrolled patients with MK underwent SLP to obtain images of their epithelial defects (ED). Eyes were stained with fluorescein and imaged multiple times under blue light, at low and high magnifications. A masked research assistant chose the 3 best images and annotated each 3 times to provide seed regions corresponding to ED and healthy cornea. The algorithm returned the ED area for each seeded image. Eyes without EDs and algorithm failures were excluded. Variance components were estimated with a random effects model and intraclass correlation coefficients estimated with intragrader reliability.ResultsA total of 42 eyes from 42 MK participants were photographed. After excluding poor quality images, eyes with no EDs, and algorithm failures, 34 patients with 92 images and 274 seeds were analyzed. No significant differences in the average ED area were found between seedings or high- versus low-SLP magnifications (all P > 0.5, paired t tests). Minimal measurement variability was because of image (0.9%), magnification (0.2%), or seed (0.1%). Most variability was attributable to differences in ED sizes between patients (85.2%). 13.7% of variability was unexplained. Multiple iterations of the algorithm on the same image showed good consistency (intraclass correlation coefficient = 0.98, 95% confidence interval, 0.97-0.99).ConclusionsImage-analysis algorithms showed good reliability for measuring the ED area from SLP images. Most measurement variability was because of between-patient differences, not imaging settings or application of the algorithm by the user.
| S-EPMC7732187 | biostudies-literature
Project description:PURPOSE:To evaluate the efficacy of slit lamp breath shields to prevent droplet spray from a simulated sneeze. DESIGN:Experimental study to test effectiveness of personal protective equipment. METHODS:The nozzle of a spray gun was adjusted to angularly disperse a mist of colored dye that approximated a patient sneezing on a dimensionally accurate cardboard slit lamp model. We tested the designs of six commercially available breath shields, and one breath shield repurposed from a plastic container lid. Each breath shield was sprayed in a standardized fashion three times and the amount of overspray compared with no shield was quantified. The surface area that was sprayed was calculated using Adobe Photoshop's color range function. The average percentage of overspray of each breath shield was computed in comparison to the control. RESULTS:The breath shields ranged in surface area from 116-1254 cm2 and the amount of overspray varied from 54% to virtually none. Larger breath shields offered better protection than smaller ones. Breath shields attached to the objective lens arm were a better barrier than those hung by the oculars of comparable size. A repurposed plastic lid breath shield was 513cm2, was slightly curved toward the examiner's face, and allowed only 2% overspray. The largest breath shield (1254 cm2) hung near the oculars and prevented essentially all the overspray. CONCLUSION:The performance of different designs of breath shields is variable. Even high functioning shields should be used in conjunction with personal protective equipment including masks, goggles and gloves, and handwashing. Ideally patients should also wear a cloth mask during all slit lamp exams.
| S-EPMC7212996 | biostudies-literature
Project description:Ocular images play an essential role in ophthalmology. Current research mainly focuses on computer-aided diagnosis using slit-lamp images, however few studies have been done to predict the progression of ophthalmic disease. Therefore exploring an effective approach of prediction can help to plan treatment strategies and to provide early warning for the patients. In this study, we present an end-to-end temporal sequence network (TempSeq-Net) to automatically predict the progression of ophthalmic disease, which includes employing convolutional neural network (CNN) to extract high-level features from consecutive slit-lamp images and applying long short term memory (LSTM) method to mine the temporal relationship of features. First, we comprehensively compare six potential combinations of CNNs and LSTM (or recurrent neural network) in terms of effectiveness and efficiency, to obtain the optimal TempSeq-Net model. Second, we analyze the impacts of sequence lengths on model's performance which help to evaluate their stability and validity and to determine the appropriate range of sequence lengths. The quantitative results demonstrated that our proposed model offers exceptional performance with mean accuracy (92.22), sensitivity (88.55), specificity (94.31) and AUC (97.18). Moreover, the model achieves real-time prediction with only 27.6ms for single sequence, and simultaneously predicts sequence data with lengths of 3-5. Our study provides a promising strategy for the progression of ophthalmic disease, and has the potential to be applied in other medical fields.
| S-EPMC6067742 | biostudies-literature
Project description:BackgroundCorneal neovascularization (CoNV) is a common sign in anterior segment eye diseases, the level of which can indicate condition changes. Current CoNV evaluation methods are time-consuming and some of them rely on equipment which is not widely available in hospitals. Thus, a fast and efficient evaluation method is now urgently required. In this study, a deep learning (DL)-based model was developed to automatically segment and evaluate CoNV using anterior segment images from a slit-lamp microscope.MethodsA total of 80 cornea slit-lamp photographs (from 80 patients) with clinically manifested CoNV were collected from December 2021 to July 2022 at Tianjin Medical University Eye Hospital. Of these, 60 images were manually labelled by ophthalmologists using ImageJ software to train the vessel segmentation network IterNet. To evaluate the performance of this automated model, evaluation metrics including accuracy, precision, area under the receiver operating characteristic (ROC) curve (AUC), and F1 score were calculated between the manually labelled ground truth and the automatic segmentations of CoNV of 20 anterior segment images. Furthermore, the vessels pixel count was automatically calculated and compared with the manually labelled results to evaluate clinical usability of the automated segmentation network.ResultsThe IterNet model achieved an AUC of 0.989, accuracy of 0.988, sensitivity of 0.879, specificity of 0.993, area under precision-recall of 0.921, and F1 score of 0.879. The Bland-Altman plot between manually labelled ground truth and automated segmentation results produced a concordance correlation coefficient of 0.989, 95% limits of agreement between 865.4 and -562.4, and the vessels pixel count's Pearson coefficient of correlation was 0.981 (P<0.01).ConclusionsThe fully automated network model IterNet provides a time-saving and efficient method to make a quantitative evaluation of CoNV using slit-lamp anterior segment images. This method demonstrates great value and clinical application potential for patient care and future research.
| S-EPMC10585580 | biostudies-literature
Project description:Slit lamps are routinely used to examine large numbers of patients every day due to high throughput. Previous, cultivation-based results suggested slit lamps to be contaminated with bacteria, mostly coagulase-negative staphylococci, followed by micrococci, bacilli, but also Staphylococcus aureus. Our study aimed at obtaining a much more comprehensive, cultivation-independent view of the slit lamp bacteriota and its hygienic relevance, as regularly touched surfaces usually represent fomites, particularly if used by different persons. We performed extensive 16S rRNA gene sequencing to analyse the bacteriota, of 46 slit lamps from two tertiary care centers at two sampling sites, respectively. 82 samples yielded enough sequences for downstream analyses and revealed contamination with bacteria of mostly human skin, mucosa and probably eye origin, predominantly cutibacteria, staphylococci and corynebacteria. The taxonomic assignment of 3369 ASVs (amplicon sequence variants) revealed 19 bacterial phyla and 468 genera across all samples. As antibiotic resistances are of major concern, we screened all samples for methicillin-resistant Staphylococcus aureus (MRSA) using qPCR, however, no signals above the detection limit were detected. Our study provides first comprehensive insight into the slit lamp microbiota. It underlines that slit lamps carry a highly diverse, skin-like bacterial microbiota and that thorough cleaning and disinfection after use is highly recommendable to prevent eye and skin infections.
| S-EPMC8635730 | biostudies-literature
Project description:PurposeTo assess the usability and image quality of a smartphone adapter for direct slit lamp imaging.MethodsA single-center, prospective, clinical study conducted in the Department of Ophthalmology at the University Hospital Zurich, Switzerland. The smartphone group consisted of 26 medical staff (consultants, residents, and students). The control group consisted of one ophthalmic photographer. Both groups took images of the anterior and the posterior eye segment of the same proband. The control group used professional photography equipment. The participant group used an Apple iPhone 11 mounted on a slit lamp via a removable SlitREC smartphone adapter (Custom Surgical GmbH, Munich, Germany). The image quality was graded independently by two blinded ophthalmologists on a scale from 0 (low) to 10 (high quality). Images with a score ≥ 7.0/10 were considered as good as the reference images. The acquisition time was measured. A questionnaire on usability and experience in smartphone and slit lamp use was taken by all of the participants.ResultsEach participant had three attempts at the same task. The overall smartphone quality was 7.2/10 for the anterior and 6.4/10 for the posterior segment. The subjectively perceived difficulty decreased significantly over the course of three attempts (Kendall's W). Image quality increased as well but did not improve significantly from take 1 to take 3. However, the image quality of the posterior segment was significantly, positively correlated (Spearman's Rho) with work experience. The mean acquisition time for anterior segment imaging was faster in the smartphone group compared to the control group (156 vs. 206 s). It was vice versa for the posterior segment (180 vs. 151 s).ConclusionSlit lamp imaging with the presented smartphone adapter provides high-quality imaging of the anterior segment. Posterior segment imaging remains challenging in terms of image quality. The adapter constitutes a cost-effective, portable, easy-to-use solution for recording ophthalmic photos and videos. It can facilitate clinical documentation and communication among colleagues and with the patient especially outside normal consultation hours. Direct slit lamp imaging allows for time to be saved and increases the independence of ophthalmologists in terms of patient mobility and the availability of photographic staff.
| S-EPMC9913954 | biostudies-literature
Project description:PurposePhotobiomodulation (PBM) refers to therapeutic irradiation of tissue with low-energy, 630- to 1000-nm wavelength light. An increasing body of evidence supports a beneficial effect of PBM in retinal disorders. To date, most studies have utilized light-emitting diode irradiation sources. Slit-lamp-mounted retinal lasers produce a coherent beam that can be delivered with precisely defined dosages and predetermined target area; however, the use of retinal lasers raises safety concerns that warrant investigation prior to clinical application. In this study, we determined safe dosages of laser-delivered PBM to the retina.MethodsA custom-designed, slit-lamp-delivered, 670-nm, red/near-infrared laser was used to administer a range of irradiances to healthy pigmented and non-pigmented rat retinas. The effects of PBM on various functional and structural parameters of the retina were evaluated utilizing a combination of electroretinography, Spectral Domain Optical Coherence (SD-OCT), fluorescein angiography, histology and immunohistochemistry.ResultsIn non-pigmented rats, no adverse events were identified at any irradiances up to 500 mW/cm2. In pigmented rats, no adverse events were identified at irradiances of 25 or 100 mW/cm2; however, approximately one-third of rats that received 500 mW/cm2 displayed very localized photoreceptor damage in the peripapillary region, typically adjacent to the optic nerve head.ConclusionsA safety threshold exists for laser-delivered PBM in pigmented retinas and was identified as 500 mW/cm2 irradiance; therefore, caution should be exercised in the dosage of laser-delivered PBM administered to pigmented retinas.Translational relevanceThis study provides important data necessary for clinical translation of laser-delivered PBM for retinal diseases.
| S-EPMC7396177 | biostudies-literature
Project description:PurposeTo determine whether convolutional neural networks can detect morphological differences between images of microbiologically positive and negative corneal ulcers.MethodsA cross-sectional comparison of prospectively collected data consisting of bacterial and fungal cultures and smears from eyes with acute infectious keratitis at Aravind Eye Hospital. Two convolutional neural network architectures (DenseNet and MobileNet) were trained using images obtained from handheld cameras collected from culture-positive and negative images and smear-positive and -negative images. Each architecture was trained on two image sets: (1) one with labels assigned using only culture results and (2) one using culture and smear results. The outcome measure was area under the receiver operating characteristic curve for predicting whether an ulcer would be microbiologically positive or negative.ResultsThere were 1970 images from 886 patients were included. None of the models were better than random chance at predicting positive microbiologic results (area under the receiver operating characteristic curve ranged from 0.49 to 0.56; all confidence intervals included 0.5).ConclusionsThese two state-of-the-art deep convolutional neural network architectures could not reliably predict whether a corneal ulcer would be microbiologically positive or negative based on clinical photographs. This absence of detectable morphological differences informs the future development of computer vision models trained to predict the causative agent in infectious keratitis using corneal photography.Translational relevanceThese deep learning models were not able to identify morphological differences between microbiologically positive and negative corneal ulcers. This finding suggests that similar artificial intelligence models trained to identify the causative pathogen using only microbiologically positive cases may have potential to generalize well, including to cases with falsely negative microbiologic testing.
| S-EPMC9836011 | biostudies-literature