Poor Quality Photographs Research Articles

Deep Learning Identifies High-Quality Fundus Photographs and Increases Accuracy in Automated Primary Open Angle Glaucoma Detection.

To develop and evaluate a deep learning (DL) model to assess fundus photograph quality, and quantitatively measure its impact on automated POAG detection in independent study populations. Image quality ground truth was determined by manual review of 2815 fundus photographs of healthy and POAG eyes from the Diagnostic Innovations in Glaucoma Study and African Descent and Glaucoma Evaluation Study (DIGS/ADAGES), as well as 11,350 from the Ocular Hypertension Treatment Study (OHTS). Human experts assessed a photograph as high quality if of sufficient quality to determine POAG status and poor quality if not. A DL quality model was trained on photographs from DIGS/ADAGES and tested on OHTS. The effect of DL quality assessment on DL POAG detection was measured using area under the receiver operating characteristic (AUROC). The DL quality model yielded an AUROC of 0.97 for differentiating between high- and low-quality photographs; qualitative human review affirmed high model performance. Diagnostic accuracy of the DL POAG model was significantly greater (P < 0.001) in good (AUROC, 0.87; 95% CI, 0.80-0.92) compared with poor quality photographs (AUROC, 0.77; 95% CI, 0.67-0.88). The DL quality model was able to accurately assess fundus photograph quality. Using automated quality assessment to filter out low-quality photographs increased the accuracy of a DL POAG detection model. Incorporating DL quality assessment into automated review of fundus photographs can help to decrease the burden of manual review and improve accuracy for automated DL POAG detection.

Crowdsourcing fungal biodiversity: revision of iNaturalist observations in Northwestern Siberia

The paper presents the first analysis of crowdsourcing data of all observations of fungi (including lichens) and myxomycetes in Northwestern Siberia uploaded to iNaturalist.org to date (24.02.2022). The Introduction presents an analysis of fungal diversity crowdsourcing globally, in Russia, and in the region of interest. Materials and methods describe the protocol of uploading data to iNaturalist.org, the structure of the crowdsourcing community, initiative to revise the accumulated data, procedures of data analysis, and compilation of a dataset of revised crowdsourced data. The Results present the analysis of accumulated data by several parameters: temporal, geographical and taxonomical scope, observation and identification efforts, identifiability of various taxa, species novelty and Red Data Book categories and the protection status of registered observations. The Discussion provides data on usability of crowdsourcing data for biodiversity research and conservation of fungi, including pros and contras. The Electronic Supplements to the paper include an annotated checklist of observations of protected species with information on Red Data Book categories and the protection status, and an annotated checklist of regional records of new taxa. The paper is supplemented with a dataset of about 15 000 revised and annotated records available through Global Biodiversity Information Facility (GBIF). The tradition of crowdsourcing is rooted in mycological societies around the world, including Russia. In Northwestern Siberia, a regional mycological club was established in 2018, encouraging its members to contribute observations of fungi on iNaturalist.org. A total of about 15 000 observations of fungi and myxomycetes were uploaded so far, by about 200 observers, from three administrative regions (Yamalo-Nenetsky Autonomous Okrug, Khanty-Mansi Autonomous Okrug, and Tyumen Region). The geographical coverage of crowdsourcing observations remains low. However, the observation activity has increased in the last four years. The goal of this study consisted of a collaborative effort of professional mycologists invited to help with the identification of these observations and analysis of the accumulated data. As a result, all observations were reviewed by at least one expert. About half of all the observations have been identified reliably to the species level and received Research Grade status. Of those, 90 species (195 records) represented records of taxa new to their respective regions; 876 records of 53 species of protected species provide important data for conservation programmes. The other half of the observations consists of records still under-identified for various reasons: poor quality photographs, complex taxa (impossible to identify without microscopic or molecular study), or lack of experts in a particular taxonomic group. The Discussion section summarises the pros and cons of the use of crowdsourcing for the study and conservation of regional fungal diversity, and summarises the dispute on this subject among mycologists. Further research initiatives involving crowdsourcing data must focus on an increase in the quality of observations and strive to introduce the habit of collecting voucher specimens among the community of amateurs. The timely feedback from experts is also important to provide quality and the increase of personal involvement.

Effects of Positioning Errors Onto the Ratio and Angle Measurements in Photographs of Patients With Unilateral Cleft Lip and Palate.

Nasolabial esthetics is 1 of the most important issues in the evaluation of treatment outcomes in patients with unilateral cleft lip and palate (CLP). Frontal and basal view photographs are commonly used to evaluate nasolabial esthetics in patients with unilateral CLP. High-quality photography plays an important role in evaluating the treatment outcomes and poor quality photographs can result in errors, whereas evaluating the treatment results. Facial proportions and angles are important in the esthetic perception of the face. Hence, this study aimed to determine the effects of positioning errors on the ratio and angle measurements in photographs of patients with unilateral CLP. In the clinical and computer environment, a mechanism was set up to take photos at standard and at different angles. A real-size unilateral CLP silicone doll was constructed for the photographs to be taken in the clinical setting. A test-retest was performed to determine the reliability of the measured values. The correlation coefficients for the first measurement and remeasurement values were statistically significant; positive and strong correlations were obtained. Based on the data obtained from our study, the angle and ratio measurements obtained from the two-dimensional photographs provided reliable results.

Life Style Intervention Improves Retinopathy Status-The Finnish Diabetes Prevention Study.

The aim of the study was to find out whether participation in earlier intervention had an effect on the occurrence of retinopathy in study participants. We also examined risk factors (age, sex, weight, fasting and 2 h glucose, fasting insulin, blood pressure, serum lipids) for early retinal changes. The study included 522 individuals (mean 55 years old, range 40–64 years) with impaired glucose tolerance who were randomized into intervention (weight loss, healthy diet, and physical activity, N = 265) and control groups (N = 257). Intervention lasted for median of four years in 1993–2000, after which annual follow-up visits at study clinics were conducted. In the years 2002–2006 (at least five years after stopping intervention), fundus photography was offered for all study participants in four of five study clinics. Photographs were assessed by two experienced ophthalmologists (A.A. and K.K.), masked for the group assignment. After exclusion of poor quality photographs, the data of 211 individuals (N = 113 for intervention and N = 98 for control group) were included in the present study. The occurrence of microaneurysms was significantly higher in the control (37/98, 38%) than in the intervention group (27/113, 24%; p = 0.029). In the model, including age, sex, diabetes diagnosis before the retinal assessment, body mass index (BMI), and treatment group, the odds ratio for microaneurysms was markedly lower in intervention group (OR 0.52; 0.28–0.97, p = 0.039). The only risk factor that predicted the occurrence of microaneurysms was serum triglycerides at baseline (mean ± SD 1.9 ± 0.9 vs. 1.6 ± 0.7, mmol/L, with and without microaneurysms, respectively, p = 0.003). Triglycerides associated with decreased microaneurysms in regression analysis for age, sex, fasting glucose, and intervention group (OR 1.92, p = 0.018). Lifestyle intervention in overweight and obese individuals with impaired glucose tolerance showed decreased occurrence of retinal microaneurysms. Elevated serum triglycerides were associated to the development of early diabetic microangiopathy.

Establishing Base Elements of Perspective in Order to Reconstruct Architectural Buildings from Photographs

The use of perspective images, especially historical photographs for retrieving information about presented architectural environment is a fast developing field recently. The photography image is a perspective image with secure geometrical connection with reality, therefore it is possible to reverse this process. The aim of the herby study is establishing requirements which a photographic perspective representation should meet for a reconstruction purpose, as well as determination of base elements of perspective such as a horizon line and a circle of depth, which is a key issue in any reconstruction. The starting point in the reconstruction process is geometrical analysis of the photograph, especially determination of the kind of perspective projection applied, which is defined by the building location towards a projection plane. Next, proper constructions can be used. The paper addresses the problem of establishing base elements of perspective on the basis of the photograph image in the case when camera calibration is impossible to establish. It presents different geometric construction methods selected dependently on the starting assumptions. Therefore, the methods described in the paper seem to be universal. Moreover, they can be used even in the case of poor quality photographs with poor perspective geometry. Such constructions can be realized with computer aid when the photographs are in digital form as it is presented in the paper. The accuracy of the applied methods depends on the photography image accuracy, as well as drawing accuracy, however, it is sufficient for further reconstruction. Establishing base elements of perspective presented in the paper is especially useful in difficult cases of reconstruction, when one lacks information about reconstructed architectural form and it is necessary to lean on solid geometry.

The Importance of Quality for the Accommodation of Amateur Photographs in the Professional News Media

Amateur photographs must fulfill specific standards if they are to be used in professional news media. These standards include news value, credibility, ethics and taste, as well as the technical and aesthetic quality of images. The research presented in this article explores the value of poor-quality photographs for the news industry, and the view that they can be good enough for certain news content. At the same time, this article reveals strong reservations regarding the incorporation of amateur photographs. The article asks whether such findings manifest necessary quality standards, or whether they constitute claims to defend the position of the professional news media. The data for this article were collected in 2010/2011 using observation, interviews and textual analysis.

Characteristics of cases submitted to a statewide system of child abuse experts

BackgroundThe Texas Department of Family and Protective Services (DFPS) created a statewide network of child abuse pediatricians working at 6 academic centers in 2006. MethodsCase data for the period 9/1/2012–8/31/2014 were reviewed for child demographics, risk factors, physician determinations, injury descriptors, and information made available to the physician. Results8061 of 410,315 DFPS cases (2%) were referred during the period. Network utilization rates ranged from 7.5% in counties served by San Antonio to 1.4% in Austin-served counties. Cases from outlying counties were less likely to involve older children. African American children were over-represented in referrals, and Caucasian and Asian children were underrepresented. Prevalent risk factors included prior DFPS history, single parent-household, mental illness, substance abuse, and domestic violence. Network providers found no or only nonspecific evidence for maltreatment in 45% of cases. Determinations were hindered by poor quality photographs in 188 cases and inadequate skeletal surveys in 121. Insufficient information precluded case completion for 65 cases of children <2years. ConclusionsOur data indicate that child abuse centers in Texas are underutilized, and there are practice differences with respect to medical referrals across different regions. Our findings concerning ethnic differences and risk factors are consistent with national trends. The problems of inadequate photographs and case information may improve with better training of workers; the issue of poor quality skeletal surveys requires further study. Standard DFPS criteria for referrals to child abuse pediatricians should mitigate some of the problems identified by this study.

Sensitivity and specificity of nonmydriatic digital imaging in screening diabetic retinopathy in Indian eyes.

Background:Nonmydriatic digital imaging (NMDI) is ideal for screening diabetic retinopathy (DR), but its use in Indian eyes has not been evaluated.Aim:The aim was to evaluate the sensitivity and specificity of NMDI as a screening tool in detecting DR in Indian eyes.Design:A prospective, nonrandomized, noncomparative, noninterventional study.Materials and Methods:A total of 500 diabetic patients visiting the endocrinology clinic (September 2008-June 2010) underwent NMDI (Zeiss Procam), followed by routine dilated fundus photography (FP; Zeiss Visupac 450+) of 345° retinal fields (1) optic disc and macula, (2) superotemporal, and (3) nasal to optic disc. Two-masked retina specialists graded the images for quality and severity of DR, and compared between NMDI and dilated FP.Statistical Analysis:SPSS Windows 17 for version.Results:Mean age was 52.97 ± 13.46 years (306 males: 194 females). The rate of ungradable images was 30.6% and 31% by the two observers. By observer 1, the sensitivity and specificity of detecting any DR was 58.8% and 69.1%, respectively, (κ = 0.608) and sight-threatening DR (STDR) was 63.1% and 68.9%, respectively, (κ = 0.641). By observer 2, the sensitivity and specificity was 57.3% and 68.3%, respectively, for any DR (κ = 0.593) and 62.8% and 68.3%, respectively, for STDR (κ = 0.637). The level of agreement between two observers was high (κ = 0.96).Conclusion:A high rate of poor quality photographs and low sensitivity limited the use of NMDI as a perfect screening system, particularly in dark iris population with diabetes as seen in Indian eyes.

The muranga teleophthalmology study: Comparison of virtual (teleglaucoma) with in-person clinical assessment to diagnose glaucoma

Purpose:While the effectiveness of teleophthalmology is generally accepted, its ability to diagnose glaucomatous eye disease remains relatively unknown. This study aimed to compare a web-based teleophthalmology assessment with clinical slit lamp examination to screen for glaucoma among diabetics in a rural African district.Materials and Methods:Three hundred and nine diabetic patients underwent both the clinical slit lamp examination by a comprehensive ophthalmologist and teleglaucoma (TG) assessment by a glaucoma subspecialist. Both assessments were compared for any focal glaucoma damage; for TG, the quality of photographs was assessed, and vertical cup-to-disk ratio (VCDR) was calculated in a semi-automated manner. In patients with VCDR > 0.7, the diagnostic precision of the Frequency Doubling Technology (FDT) C-20 screening program was assessed.Results:Of 309 TG assessment photos, 74 (24%) were deemed unreadable due to media opacities, patient cooperation, and unsatisfactory photographic technique. While the identification of individual optic nerve factors showed either fair or moderate agreement, the ability to diagnose glaucoma based on the overall assessment showed moderate agreement (Kappa [κ] statistic 0.55% and 95% confidence interval [CI]: 0.48-0.62). The use of FDT to detect glaucoma in the presence of disc damage (VCDR > 0.7) showed substantial agreement (κ statistic of 0.84 and 95% CI 0.79-0.90). A positive TG diagnosis of glaucoma carried a 77.5% positive predictive value, and a negative TG diagnosis carried an 82.2% negative predicative value relative to the clinical slit lamp examination.Conclusion:There was moderate agreement between the ability to diagnose glaucoma using TG relative to clinical slit lamp examination. Poor quality photographs can severely limit the ability of TG assessment to diagnose optic nerve damage and glaucoma. Although further work and validation is needed, the TG approach provides a novel, and promising method to diagnose glaucoma, a major cause of ocular morbidity throughout the world.

Diagnostic Value and Cost-Effectiveness of Good Quality Digital Images Accompanying Electronic Referrals for Suspected Skin Malignancies

The aim of this study was to investigate the outcome and cost-effectiveness of good and poor quality photographs accompanying the electronic referrals for suspected skin malignancies. A retrospective study of 100 patients, divided into 2 groups, 50 with good quality photographs and 50 with poor quality photographs. Patients with no digital images, or who failed to attend, or patients with incomplete notes were excluded from the study. The treatment pathway, waiting times, and estimated cost between the 2 groups were compared. Good photographs were more likely to be treated at the 1-Stop Clinic (P = 0.05). Good images had a better positive predictive value than poor quality images (62.55% vs. 42.86%). Good quality images are more accurate than poor quality images in triaging of patients, and thus more effective in facilitating the treatment of malignant lesions timely. Good quality photographs allow a delayed appropriate treatment of benign lesions. This increases the safety for patients in a queue in a rationed health care system, and improves patient flow.

Bayesian analysis to correct false-negative errors in capture–recapture photo-ID abundance estimates

Capture–recapture methods are largely used for estimating the size of some cetacean populations. The application of those methods for photo-identification data of recognizable individuals is very common. Poor quality photographs may lead the analyst to identify two sightings of the same individual as being different (false-negative errors). This kind of matching error inflates population size estimates. We develop a Bayesian approach to obtain bias corrected estimates of the population size N. The method can be used for Mt type capture–recapture models (Otis et al. Wildlife Monographs 62 (1978) 1–135) involving two or more sampling occasions. We used the methodology for simulated data.

Colonoscopic Features of Sessile Serrated Polyps

Sessile serrated polyps (SSPs) are one of the precursors for colon cancer, because of a high BRAF mutation rate. However, the colonoscopic appearances of SSP are still unclear. This study aimed to clarify colonoscopic features of SSPs. Methods: One experienced colonoscopist performed chromocolonoscopy for 190 patients in Australia with the aim to remove all polyps. 166 serrated polyps were detected, of which 54 serrated polyps measuring 5 mm or greater were enrolled. Polyp size was assessed based on resected specimens. One independent colonoscopist reviewed printed photographs of the serrated polyps, blinded to histology. Polyp morphology (flat, protruded), polyp surface (smooth, rough), polyp hue (pale, pink), adherent mucus (thin, thick) and pit pattern (asteroid, elliptic, gyrus-like) were evaluated. Histology was assessed by a single pathologist, and serrated polyps were subclassified as goblet cell (GC) type, microvesicular (MV) type, SSP, traditional serrated adenoma (TSA) and mixed polyp. Results: Ten serrated polyps were excluded because of poor quality photographs, leaving 44 serrated polyps which were studied. The mean size ( ± SD) of the 44 serrated polyps was 8.8 ± 4.0 mm, and 61% were located in the proximal colon. The mean size, location and colonoscopic features of each histological type are shown in the table. All 5 mixed polyps included a TSA component on histology, so these types were combined. The mean size of polyps in the TSA/mixed polyp category was significantly larger than those of MV type and SSP (Bonferroni correction), but there was no significant difference between mean size of MV type and SSP. Except for proximal location, SSPs were very similar in endoscopic appearance to the MV type. Both types of serrated polyps were pale, smooth and flat in appearance with an asteroid pit pattern. The majority of GC type located in the proximal colon showed protruded morphology. The TSA and mixed polyps were distributed evenly and there was no specific morphology. Conclusion: At colonoscopy, GC type can be discriminated from MV type and SSP. However, both MV type and SSP were pale, smooth and flat in appearance with an asteroid pit pattern, making differentiation difficult. The only difference was the anatomical distribution.

Diabetic retinopathy: the unmet needs for screening and a review of potential solutions

Diabetic retinopathy remains the leading cause of severe vision loss and blindness in the developed world, in spite of recognized ocular treatments that are successful at reducing the rate of vision impairment. Retinal photography appears a promising method to perform screening in such a setting utilizing new 45°+ retinal cameras that do not require pupil dilation and can be operated by a trained, nonophthalmic technician. Certain developments may make the photography more successful including the conversion to electronic chip camera sensors that allow each picture as it is taken to be immediately projected onto a monitor for evaluation and assessment. Utilizing a nonmydriatic camera, studies of single-field photography through a dilated pupil have demonstrated superior or equal sensitivity to fundus examination by an ophthalmologist in a number of studies. However, photography without pupil dilation, especially in the older age group may result in poor-quality photographs owing to intense bilateral pupil constriction after the first images and also due to the presence of cataracts. Computer analysis of the retinal images allows extraction of quantitative data, not only of the diabetic lesions but also of vascular changes that, up until now, have been impossible by human grading and potentially allows a much more detailed and quantitative evaluation of the progression of retinopathy over time. When success of image processing algorithms is demonstrated for a large number of images taken under screening conditions, the benefits of retinal photography and image processing to provide timely, reliable, quantitative and cost-effective results, will make this the preferred method over physician examination or human grader evaluation of the images.

Using slides to test for changes in crown defoliation assessment methods part II: Application of the image analysis system CROCO

We tested whether the semi-automatic program CROCO can replace visual assessments of slides to detect changes in defoliation assessment methods. We randomly selected a series of slides of 24 Norway spruce trees with 220 field assessments made between 1986 and 1995. The slides had been randomly arranged and assessed by three experts without knowledge of the tree number or the year when the slide was taken. Defoliation scores were computed with CROCO. Each tree had thus three different defoliation scores, field assessments, photo assessments and CROCO scores. CROCO scores were less correlated with the field assessments (Spearman's rank correlation: 0.67) than were the slide assessments with the field assessments (0.79-0.83). However, CROCO was not biased against the field scores, while slide assessments systematically underestimated defoliation. In a multi-variate mixed effect model none of the variables tree overlap, tree visibility and light conditions was significant in explaining differences between slide assessors and CROCO scores. The same model applied for the differences from the field scores yielded significant effects for poor light conditions (CROCO and all assessors), for crown overlap (CROCO and one assessor) and for visibility (one assessor). We conclude, therefore, that CROCO can be used to detect past and future changes in assessment methods without bias if poor quality photographs are avoided.

The Diagnosis of Diabetic Retinopathy: Ophthalmoscopy versus Fundus Photography

To compare fundus photography with ophthalmoscopy in the detection of diabetic retinopathy. Ophthalmoscopy and fundus photographs with a nonmydriatic camera, both performed through dilated pupils, were compared to diagnose retinopathy in a cohort of 410 Oklahoma Indians with noninsulin-dependent diabetes mellitus. A total of 795 eyes were examined using both methods. The mean age of participants was 60.3 years, with a mean duration of diabetes of 17.3 years. An overall agreement of 86.3% with a kappa statistic kappa of 0.74 was found between ophthalmoscopy and fundus photography with a nonmydriatic camera. For the diagnosis of proliferative diabetic retinopathy, kappa = 0.84 with an agreement of 98.1%. With a total of 61 cases of proliferative retinopathy diagnosed by either method in our study, ophthalmoscopy alone detected 88.5% and fundus photography, 78.7%. When compared on a lesion-by-lesion basis, agreement between the two diagnostic methods was highest for nonproliferative retinopathy, as well as fibrous proliferation. The fundus photography with a nonmydriatic camera, performed with mydriasis, is comparable to ophthalmoscopy for the detection of retinopathy. It may prove to be a suitable, cost-effective method for routine screening in diabetes clinics, provided ophthalmologic referral is ensured for those with a diagnosis of any form of retinopathy, questionable retinopathy, nondiabetic retinopathy, those with poor quality photographs, as well as those with acute changes in visual acuity.

Detection of Diabetic Retinopathy in the Community Using a Non-mydriatic Camera

The role of the non-mydriatic fundus camera in detection of diabetic retinopathy was evaluated as part of a comprehensive screening programme for diabetic complications offered to all diabetic patients in a rural town. Retinopathy was demonstrated in 124/358 (35%) of patients screened. Forty-eight patients (13%) were judged to have sight-threatening retinopathy, of whom 29 patients (8% of the total) were not already under the care of an ophthalmologist. However, in only 66% of patients were photographs of both eyes of adequate quality to assess for retinopathy. The percentage of poor quality photographs increased with age in those aged greater than 50 years. It is concluded that the non-mydriatic camera can increase the detection of sight-threatening retinopathy in the community. Although this method of screening is not perfect, because of the number of poor quality photographs, it may be as good as or better than existing screening practices in unselected diabetic populations.