Geographic atrophy (GA) is late-stage dry age-related macular degeneration (AMD). Improved predictors of GA progression would be useful in clinical trial design and may be relevant for clinical practice. The purpose of this study was to accurately predict GA progression over time from baseline fundus autofluorescence (FAF) images (Heidelberg Engineering, Inc., Germany) using deep learning. Study eyes of patients (n = 1312) enrolled in the Lampalizumab trials1, 2 (NCT02479386, NCT02247479, NCT02247531) were included. The dataset was split by patient into training (n = 1047) and holdout sets (n = 265). GA progression, defined as GA lesion growth rate, was derived by a linear fit on all available measurements of GA lesion area (mm2 , measured from manually graded FAF images). The model performance was evaluated using 5-fold cross-validation (CV). Coefficient of determination (R2 ) computed as the square of Pearson correlation coefficient was used as the performance metric. Multiple modeling approaches were implemented, and the best performance was observed using cascade learning. In this approach, pre-trained weights on ImageNet were finetuned to predict GA lesion area followed by further fine-tuning to predict GA growth rate. The 5-folds had an average CV R2 of 0.44, and the holdout showed R2 of 0.50 (95% confidence interval: 0.41 - 0.61). In comparison, a linear model using only baseline GA lesion area in the same holdout showed an R2 of 0.18. Further investigation with visualization techniques might help understand the pathophysiology behind the predictions. The predictions may be improved by combining with imaging modalities like near-infrared and/or optical coherence tomography.
A microfluidic chip with microchannels ranging from 8 to 96 μm was used to mimic blood vessels down to the capillary level. Blood flow within the microfluidic channels was analyzed with split-spectrum amplitude-decorrelation angiography (SSADA)-based optical coherence tomography (OCT) angiography. It was found that the SSADA decorrelation value was related to both blood flow speed and channel width. SSADA could differentiate nonflowing blood inside the microfluidic channels from static paper. The SSADA decorrelation value was approximately linear with blood flow velocity up to a threshold Vsat of 5.83±1.33 mm/s (mean±standard deviation over the range of channel widths). Beyond this threshold, it approached a saturation value Dsat. Dsat was higher for wider channels, and approached a maximum value Dsm as the channel width became much larger than the beam focal spot diameter. These results indicate that decorrelation values (flow signal) in capillary networks would be proportional to both flow velocity and vessel caliber but would be capped at a saturation value in larger blood vessels. These findings are useful for interpretation and quantification of clinical OCT angiography results.
Quantification of choroidal neovascularization (CNV) as visualized by optical coherence tomography angiography (OCTA) may have importance clinically when diagnosing or tracking disease. Here, we present an automated algorithm to quantify the vessel skeleton of CNV as vessel length. Initial segmentation of the CNV on en face angiograms was achieved using saliency-based detection and thresholding. A level set method was then used to refine vessel edges. Finally, a skeleton algorithm was applied to identify vessel centerlines. The algorithm was tested on nine OCTA scans from participants with CNV and comparisons of the algorithm’s output to manual delineation showed good agreement.
An automated algorithm was developed for total retinal blood flow (TRBF) using 70-kHz spectral optical coherence tomography (OCT). The OCT was calibrated for the transformation from Doppler shift to speed based on a flow phantom. The TRBF scan pattern contained five repeated volume scans (2×2 mm) obtained in 3 s and centered on central retinal vessels in the optic disc. The TRBF was calculated using an en face Doppler technique. For each retinal vein, blood flow was measured at an optimal plane where the calculated flow was maximized. The TRBF was calculated by summing flow in all veins. The algorithm tracked vascular branching so that either root or branch veins are summed, but never both. The TRBF in five repeated volumes were averaged to reduce variation due to cardiac cycle pulsation. Finally, the TRBF was corrected for eye length variation. Twelve healthy eyes and 12 glaucomatous eyes were enrolled to test the algorithm. The TRBF was 45.4±6.7 μl/min for healthy control and 34.7±7.6 μl/min for glaucomatous participants (p-value=0.01). The intravisit repeatability was 8.6% for healthy controls and 8.4% for glaucoma participants. The proposed automated method provided repeatable TRBF measurement.
Clinical diagnosis of cochlear dysfunction typically remains incomplete due to a lack of proper diagnostic methods.
Medical imaging modalities can only detect gross changes in the cochlea, and non-invasive in vivo cochlear
measurements are scarce. As a result, extensive efforts have been made to adapt optical coherence tomography (OCT)
techniques to analyze and study the cochlea. Herein, we detail the methods for measuring vibration using OCT. We
used spectral domain OCT with ~950 nm as the center wavelength and a bandwidth of ~80 nm. The custom
spectrometer used was based on a high speed line scan camera which is capable of line rates up to 28 kHz. The signal-to-
noise ratio of the system was ~90 dB. The data collection and processing software was written in LabVIEW and
MATLAB. We tested whether streaming directly from the camera, writing the data to multiple hard drives in the RAID-
0 configuration, and processing using the GPU shortened experiment times. We then analyzed the A-line phase noise
over several hundred milliseconds and growth curves from a piezoelectric element. We believe this is the first step
towards a diagnostic device which generates vibration information of cochlear structures.
Optical coherence tomography (OCT) is becoming a popular tool for imaging morphology in the
middle and inner ear. Vibratory measurements of the structures of the ear facilitate better
understanding of the function and limitations of the ear. We have developed an algorithm that
enables a standard spectrometer based OCT system to measure the full spectrum (90 kHz) frequency
response of the mouse ear by incorporating coherently interleaved sampling, increasing the effective
Nyquist rate of the system by a factor of 5+. The algorithm is evaluated by measuring the frequency
response of a mouse tympanic membrane to a pure tone stimulus.
Current medical imaging modalities, such as MRI and CT, do not provide high enough resolution to detect many
changes within the cochlea that cause hearing loss. We sought to develop the technique of optical coherence
tomography (OCT) to image the cochlea noninvasively and within its native environment. We used spectral domain
OCT with 950 nm as the center wavelength and a bandwidth of ~100 nm to image freshly excised normal mouse cochlea
at different developmental ages. The OCT system has an axial resolution of ~4 μm (in air) and a lateral resolution of
~10 μm. When we imaged normal adult mouse cochleae through the round window membrane, Reissner's membrane,
the basilar membrane, the tectorial membrane, the spiral ligament, the spiral limbus, and the modiolus could be clearly
identified. When we imaged intact adult cochleae, we were able to image through ~130 μm of bone and tissue to see up
to a depth of ~600 μm, and all of the previously identified structures were still visible. Imaging of early postnatal mice
during the timeline of cochlear development permitted visualization of the expected structural differences from adult
cochleae. Therefore, we conclude that spectral domain OCT is an effective technique for noninvasive imaging of the
murine cochlea.
Mice are an excellent model for studying mammalian hearing and transgenic mouse models of human hearing, loss are commonly available. However, the mouse cochlea is substantially smaller than other animal models routinely used to study cochlear physiology. This makes study of their hair cells difficult. We develop a novel methodology to optically image calcium within living hair cells left undisturbed within the excised mouse cochlea. Fresh cochleae are harvested, left intact within their otic capsule bone, and fixed in a recording chamber. The bone overlying the cochlear epithelium is opened and Reissner's membrane is incised. A fluorescent calcium indicator is applied to the preparation. A custom-built upright two-photon microscope was used to image the preparation using 3-D scanning. We are able to image about one third of a cochlear turn simultaneously, in either the apical or basal regions. Within one hour of animal sacrifice, we find that outer hair cells demonstrate increased fluorescence compared with surrounding supporting cells. This methodology is then used to visualize hair cell calcium changes during mechanotransduction over a region of the epithelium. Because the epithelium is left within the cochlea, dissection trauma is minimized and artifactual changes in hair cell physiology are expected to be reduced.
Mice are an excellent model for studying mammalian hearing and transgenic mouse models of human hearing loss
are commonly available for research. However, the mouse cochlea is substantially smaller than other animal models
routinely used to study cochlear physiology. This makes the study of their hair cells difficult. We developed a novel
methodology to optically image calcium within living hair cells left undisturbed within the excised mouse cochlea.
Fresh cochleae were harvested, left intact within their otic capsule bone, and glued upright in a recording chamber. The
bone overlying the region of the cochlear epithelium to be studied was opened and Reissner's membrane was incised. A
fluorescent indicator was applied to the preparation to image intracellular calcium. A custom-built upright two-photon
microscope was used to image the preparation using three dimensional scanning. We were able to image about 1/3 of a
cochlear turn simultaneously, in either the apical or basal regions. Within one hour of animal sacrifice, we found that
outer hair cells demonstrated increased fluorescence compared with surrounding supporting cells. Thus, this
methodology can be used to visualize hair cell calcium changes and mechanotransduction over a region of the epithelium.
Because the epithelium is left within the cochlea, dissection trauma is minimized and artifactual changes in hair cell
physiology are reduced.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.