No abstract

Background/aimsFundus fluorescein angiography (FFA) is an important technique to evaluate diabetic retinopathy (DR) and other retinal diseases. The interpretation of FFA images is complex and time-consuming, and the ability of diagnosis is uneven among different ophthalmologists. The aim of the study is to develop a clinically usable multilevel classification deep learning model for FFA images, including prediagnosis assessment and lesion classification.MethodsA total of 15 599 FFA images of 1558 eyes from 845 patients diagnosed with DR were collected and annotated. Three convolutional neural network (CNN) models were trained to generate the label of image quality, location, laterality of eye, phase and five lesions. Performance of the models was evaluated by accuracy, F-1 score, the area under the curve and human-machine comparison. The images with false positive and false negative results were analysed in detail.ResultsCompared with LeNet-5 and VGG16, ResNet18 got the best result, achieving an accuracy of 80.79%–93.34% for prediagnosis assessment and an accuracy of 63.67%–88.88% for lesion detection. The human-machine comparison showed that the CNN had similar accuracy with junior ophthalmologists. The false positive and false negative analysis indicated a direction of improvement.ConclusionThis is the first study to do automated standardised labelling on FFA images. Our model is able to be applied in clinical practice, and will make great contributions to the development of intelligent diagnosis of FFA images.

BackgroundWe modified and reconstructed a high image quality portable non-mydriatic fundus camera and compared it with the tabletop fundus camera to evaluate the efficacy of the new camera in detecting retinal diseases.MethodsWe designed and built a novel portable handheld fundus camera with telemedicine system. The image quality of fundus cameras was compared to that of existing commercial tabletop cameras by taking photographs of 364 eyes from the 254 patients. In all 800 fundus images taken by two camera types, 400 images per camera, were graded with the four image clarity classifications.ResultsUsing the portable fundus camera, 63% (252/400) images were graded as excellent overall quality, 20.5% (82/400) were good, 11.75% (47/400) were fair, and 4.75% (19/400) were inadequate. Using the tabletop fundus camera, 70.75% (283/400) images were graded as excellent overall quality, 20.4% (51/400) were good, 13.25% (53/400) were fair, and 3.25% (13/400) were inadequate. Common retinal diseases were easily identified from fundus images obtained from the portable fundus camera.ConclusionThe new type of non-mydriatic portable fundus camera was qualified to have professional quality of fundus images. The revolutionary screening camera provides a foundational platform which can potentially improve the accessibility of retinal screening programmes.

Simultaneous IFVA and ICGA are generally safe procedures with an acceptable incidence of an adverse reaction. However, patients with drug allergy history may have a higher incidence and greater severity of an adverse reaction.

PurposeTo investigate the clinical features and cytochrome P450 family 4 subfamily V polypeptide 2 (CYP4V2) gene mutations in 14 Chinese families with Bietti crystalline dystrophy (BCD).MethodsSeventeen patients from 14 unrelated Chinese families with BCD were recruited for complete clinical ophthalmic examination and genetic study. The 11 exons of CYP4V2 were amplified from genomic DNA of all patients and their family members by polymerase chain reaction (PCR) and then sequenced. Exons of TIMP3 were also sequenced in BCD patient associated with choroidal neovascularization (CNV). One hundred and seventy unrelated healthy Chinese subjects were screened for mutations in CYP4V2.ResultsAll 17 patients with BCD had mutations in CYP4V2; one of these mutations was novel (c.219T>A, p.F73L) and four other mutations had been reported. The p.F73L mutation was a commonly detected mutation in our study (seven out of 34 alleles), either in the homozygous state or in the heterozygous state. Among the patients, considerable phenotypic variability was detected, both within and between families. Screening of TIMP3 did not find any mutation in the BCD patient associated with CNV.ConclusionThe novel CYP4V2 c.219T>A (p.F73L) mutation may be another recurrent mutation in Chinese patients with BCD. Our study expands the mutation spectrum of CYP4V2 and characterizes novel genotype–phenotype associations in Chinese patients with BCD.