Objective Determination of Optimal Number of Spectral-Domain Optical Coherence Tomographic Images of Retina to Average

Shirasawa, Makoto; Sakamoto, Taiji; Terasaki, Hiroto; Yamashita, Takehiro; Uchino, Eisuke; Sonoda, Shozo

doi:10.1371/journal.pone.0110550

Cited by 5 publications

(6 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One method to compensate for these issues may be through en face image averaging, echoing the technique used to improve OCT image quality [33–36]. Unlike previous applications of registration and averaging in OCTA which were largely focused on correcting for eye motion artifacts [34, 37–39], our proposed method would require that a patient undergo multiple scans in a single session for offline image registration and averaging.…”

Section: Introductionmentioning

confidence: 99%

Visualization of Radial Peripapillary Capillaries Using Optical Coherence Tomography Angiography: The Effect of Image Averaging

Phillips

Krawitz

et al. 2017

PLoS ONE

View full text Add to dashboard Cite

ObjectivesTo assess the effect of image registration and averaging on the visualization and quantification of the radial peripapillary capillary (RPC) network on optical coherence tomography angiography (OCTA).MethodsTwenty-two healthy controls were imaged with a commercial OCTA system (AngioVue, Optovue, Inc.). Ten 10x10° scans of the optic disc were obtained, and the most superficial layer (50-μm slab extending from the inner limiting membrane) was extracted for analysis. Rigid registration was achieved using ImageJ, and averaging of each 2 to 10 frames was performed in five ~2x2° regions of interest (ROI) located 1° from the optic disc margin. The ROI were automatically skeletonized. Signal-to-noise ratio (SNR), number of endpoints and mean capillary length from the skeleton, capillary density, and mean intercapillary distance (ICD) were measured for the reference and each averaged ROI. Repeated measures analysis of variance was used to assess statistical significance. Three patients with primary open angle glaucoma were also imaged to compare RPC density to controls.ResultsQualitatively, vessels appeared smoother and closer to histologic descriptions with increasing number of averaged frames. Quantitatively, number of endpoints decreased by 51%, and SNR, mean capillary length, capillary density, and ICD increased by 44%, 91%, 11%, and 4.5% from single frame to 10-frame averaged, respectively. The 10-frame averaged images from the glaucomatous eyes revealed decreased density correlating to visual field defects and retinal nerve fiber layer thinning.ConclusionsOCTA image registration and averaging is a viable and accessible method to enhance the visualization of RPCs, with significant improvements in image quality and RPC quantitative parameters. With this technique, we will be able to non-invasively and reliably study RPC involvement in diseases such as glaucoma.

show abstract

Section: Introductionmentioning

confidence: 99%

Visualization of Radial Peripapillary Capillaries Using Optical Coherence Tomography Angiography: The Effect of Image Averaging

Phillips

Krawitz

et al. 2017

PLoS ONE

View full text Add to dashboard Cite

show abstract

“…The number of network input ranged from 1 to 20. One reason of choosing 1 to 20 is the consideration of computation power, another reason is that a previous detailed comparative study has indicated that the optimal number was 20 while minimizing the examination time and maximizing the image quality [ 8 ]. Finally, the network output, namely the enhanced OCT image, was obtained by performing a weighted sum of the averaged multiple inputs and the restored feature maps after the decoding module.…”

Section: Methodsmentioning

confidence: 99%

“…With subjective scoring, image averaging has been proved to be a statistically significant benefit for assessing the external limiting membrane (ELM) and outer nuclear layer (ONL) [ 7 ]. Then a detailed comparative experiment indicates that OCT image quality improves with an increase in the number of frames averaged, and the optimal number is 20 while minimizing the examination time and maximizing the image quality [ 8 ]. However, image averaging requires multiple repeated scans at the same location while the subject is instructed to maintain steady fixation.…”

Section: Introductionmentioning

confidence: 99%

“…Firstly, all existing studies perform image enhancement based on a single B-scan image and do not investigate whether image quality can improve continually while network inputs become multiple frames [ 12 – 15 ]. Secondly, performance evaluation is only performed with objective metrics rather than the subjective assessment from a clinical perspective, which has been conducted for image averaging [ 5 , 8 ].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Development and quantitative assessment of deep learning-based image enhancement for optical coherence tomography

Zhao

Lv²,

Meng

et al. 2022

BMC Ophthalmol

View full text Add to dashboard Cite

Purpose To develop a deep learning-based framework to improve the image quality of optical coherence tomography (OCT) and evaluate its image enhancement effect with the traditional image averaging method from a clinical perspective. Methods 359 normal eyes and 456 eyes with various retinal conditions were included. A deep learning framework with high-resolution representation was developed to achieve image quality enhancement for OCT images. The quantitative comparisons, including expert subjective scores from ophthalmologists and three objective metrics of image quality (structural similarity index measure (SSIM), peak signal-to-noise ratio (PSNR) and contrast-to-noise ratio (CNR)), were performed between deep learning method and traditional image averaging. Results With the increase of frame count from 1 to 20, our deep learning method always obtained higher SSIM and PSNR values than the image averaging method while importing the same number of frames. When we selected 5 frames as inputs, the local objective assessment with CNR illustrated that the deep learning method had more obvious tissue contrast enhancement than averaging method. The subjective scores of image quality were all highest in our deep learning method, both for normal retinal structure and various retinal lesions. All the objective and subjective indicators had significant statistical differences (P < 0.05). Conclusion Compared to traditional image averaging methods, our proposed deep learning enhancement framework can achieve a reasonable trade-off between image quality and scanning times, reducing the number of repeated scans.

show abstract

“…The optimum number of frames for image averaging during OCT imaging of the retina has been investigated based on objective and subjective determination of the improvement in a range of image quality parameters including image brightness, layer contrast and noise variance. Averaging 16–20 B‐scans significantly improves image quality metrics compared to a single non‐averaged image, with further increases in the number of frames for B‐scan averaging not significantly enhancing the image quality.…”

mentioning

confidence: 99%

Impact of image averaging on wide‐field choroidal thickness measurements using enhanced‐depth imaging optical coherence tomography

Hoseini-Yazdi

Vincent

Collins

et al. 2019

Clinical and Experimental Optometry

View full text Add to dashboard Cite

Background: The aim of this study was to examine the influence of B-scan averaging on choroidal thickness using wide-field enhanced-depth imaging optical coherence tomography. Methods: Six high-resolution trans-foveal horizontal enhanced-depth imaging line scans (spanning a 60 field) were acquired consecutively from the right eye of 10 healthy adults (mean age 30 AE 5 years), with each line scan an average of 10, 20, 30, 40, 50 or 100 B-scans, using the automated real-time image averaging and follow-up features of a Spectralis device. The impact of B-scan averaging on regional measures of wide-field choroidal thickness (across macular and peripheral regions) and their accuracy was investigated, assuming that averaging 100 B-scans would provide the most accurate estimate of choroidal thickness. Results: Regional estimates of wide-field choroidal thickness did not vary across the different B-scan averaging conditions (all p > 0.05). The mean choroidal thickness averaged across the full wide-field area exhibited the closest agreement to measures obtained with 100 averaged B-scans, when frame averaging exceeded 30 B-scans (95 per cent limits of agreement +10 to −7, +7 to −7 and +6 to −3 μm for 30, 40 and 50 averaged B-scans, respectively), compared to 10 and 20 averaged B-scans (95 per cent limits of agreement +13 to −8 and +13 to −6 μm, respectively; p < 0.01 and p < 0.02 compared to the accuracy of 50 averaged B-scans). Conclusion: Averaging 30 B-scans for an individual enhanced-depth imaging optical coherence tomography line scan provided accurate measures of choroidal thickness across a wide-field (60 ) area in young healthy eyes. This information can assist in designing the volumetric scan protocols required for detailed examination of the macular and peripheral choroid.

show abstract

Objective Determination of Optimal Number of Spectral-Domain Optical Coherence Tomographic Images of Retina to Average

Cited by 5 publications

References 16 publications

Visualization of Radial Peripapillary Capillaries Using Optical Coherence Tomography Angiography: The Effect of Image Averaging

Visualization of Radial Peripapillary Capillaries Using Optical Coherence Tomography Angiography: The Effect of Image Averaging

Development and quantitative assessment of deep learning-based image enhancement for optical coherence tomography

Impact of image averaging on wide‐field choroidal thickness measurements using enhanced‐depth imaging optical coherence tomography

Contact Info

Product

Resources

About