A machine vision based frailty index for mice

1AbstractChronological aging is uniform, but biological aging is heterogeneous. Clinically, this heterogeneity manifests itself in health status and mortality, and it distinguishes healthy from unhealthy aging. Clinical frailty indexes (FIs) serve as an important tool in gerontology to capture health status. FIs have been adapted for use in mice and are an effective predictor of mortality risk. To accelerate our understanding of biological aging, high-throughput approaches to pre-clinical studies are necessary. Currently, however, mouse frailty indexing is manual and relies on trained scorers, which imposes limits on scalability and reliability. Here, we introduce a machine learning based visual frailty index (vFI) for mice that operates on video data from an open field assay. We generate a large mouse FI datasets comprising 256 males and 195 females. From video data on these same mice, we use neural networks to extract morphometric, gait, and other behavioral features that correlate with manual FI score and age. We use these features to train a regression model that accurately predicts frailty within 1.03 ± 0.08 (3.9% ± 0.3%) of the pre-normalized FI score in terms of median absolute error. We show that features of biological aging are encoded in open-field video data and can be used to construct a vFI that can complement or replace current manual FI methods. We use the vFI data to examine sex-specific aspects of aging in mice. This vFI provides increased accuracy, reproducibility, and scalability, that will enable large scale mechanistic and interventional studies of aging in mice.

Accurate Epigenetic Aging in Bottlenose Dolphins (Tursiops Truncatus), an Essential Step in the Conservation of at-Risk Dolphins.

Epigenetics, specifically DNA methylation, allows for estimation of animal age from blood or remotely sampled skin. This multi tissue epigenetic aging clock uses 110 longitudinal samples from 34 Navy bottlenose dolphins (Tursiops truncatus), identifying 195 cytosine-phosphate-guanine sites associated with chronological aging via leave-one-individual-out-cross-validation (R2=0.95). With a median absolute error of 2.5 years this clock improves age estimation capacity in wild dolphins, expanding conservation efforts, enabling better understanding of population demographics.

Corneal topographic ring vs zone in trifocal IOL calculation

Purpose. To compare the results of trifocal IOL calculation using various corneal topographic data (ring and zone). Methods. This retrospective study involved 35 patients (40 eyes), underwent cataract surgery (FLACS) with trifocal IOL implantation (AcrySof IQ PanOptix). The calculation was performed using IOL-Master 500 according to 4 formulas (Haigis, HofferQ, Holladay 1, SRK / T) and Tomey OA-2000 according to 2 formulas (Barrett II Universal, Olsen). Topographic values included Km collected from Pentacam HR Power Distribution Apex map with diameter of 3.0 and 5.0 mm on a ring and zone. Predicted and actual refraction were compared after surgery. Results. Mean Km value on 3 mm zone and ring was: 42.75±1,46 D and 42.91±1.43 D, respectively (p<0.0001). Mean Km value on 5 mm zone and ring was: 43.09±1.5 D and 43.55±1.48 D, respectively (p<0.0001). According to 6 formulas Mean Absolute Error (MAE) calculated using 3 mm zone data was significantly less then on 3mm ring: 0.3± 0.28; 0.48±0.3 and Median Absolute Error (MedAE) was 0.225 (0.3); 0.465 (0.397) respectively (p<0.01). The same data were obtained on 5mm zone and ring: MAE was 0.29±0.28; 0.35±0.29 and MedAE amounted to 0.225 (0.3); 0.29 (0.38) respectively (p=0.02). Conclusion. Mean Km value on Power Distribution Apex map according to ring is significantly greater then according to zone. 1) Predicted refraction using corneal topographic ring data deviates towards hyperopia relative to the actual postoperative refraction. 2) The use of topographic data on zone allows to obtain more accurate calculation of trifocal IOL than when using the data on the ring. Key words: IOL calculation, Trifocal IOL, corneal topography.

Lenstar LS 900 versus Pentacam-AXL: analysis of refractive outcomes and predicted refraction

Analysis of refractive outcomes, using biometry data collected with a new biometer (Pentacam-AXL, OCULUS, Germany) and a reference biometer (Lenstar LS 900, HAAG-STREIT AG, Switzerland), in order to assess differences in the predicted and actual refraction using different formulas. Prospective, institutional study, in which intraocular lens (IOL) calculation was performed using the Haigis, SRK/T and Hoffer Q formulas with the two systems in patients undergoing cataract surgery between November 2016 and August 2017. Four to 6 weeks after surgery, the spherical equivalent (SE) was derived from objective refraction. Mean prediction error (PE), mean absolute error (MAE) and the median absolute error (MedAE) were calculated. The percentage of eyes within ± 0.25, ± 0.50, ± 1.00, and ± 2.00 D of MAE was determined. 104 eyes from 76 patients, 35 males (46.1%), underwent uneventful phacoemulsification with IOL implantation. Mean SE after surgery was − 0.29 ± 0.46 D. Mean prediction error (PE) using the SRK/T, Haigis and Hoffer Q formulas with the Lenstar was significantly different (p > 0.0001) from PE calculated with the Pentacam in all three formulas. Percentage of eyes within ± 0.25 D MAE were larger with the Lenstar device, using all three formulas. The difference between the actual refractive error and the predicted refractive error is consistently lower when using Lenstar. The Pentacam-AXL user should be alert to the critical necessity of constant optimization in order to obtain optimal refractive results.

Reducing Errors in Velocity–Azimuth Display (VAD) Wind and Deformation Retrievals from Airborne Doppler Radars in Convective Environments

The present study describes methods to reduce the uncertainty of velocity–azimuth display (VAD) wind and deformation retrievals from downward-pointing, conically scanning, airborne Doppler radars. These retrievals have important applications in data assimilation and real-time data processing. Several error sources for VAD retrievals are considered here, including violations to the underlying wind field assumptions, Doppler velocity noise, data gaps, temporal variability, and the spatial weighting function of the VAD retrieval. Specific to airborne VAD retrievals, we also consider errors produced due to the radar scans occurring while the instrument platform is in motion. While VAD retrievals are typically performed using data from a single antenna revolution, other strategies for selecting data can be used to reduce retrieval errors. Four such data selection strategies for airborne VAD retrievals are evaluated here with respect to their effects on the errors. These methods are evaluated using the second hurricane nature run numerical simulation, analytic wind fields, and observed Doppler radar radial velocities. The proposed methods are shown to reduce the median absolute error of the VAD wind retrievals, especially in the vicinity of deep convection embedded in stratiform precipitation. The median absolute error due to wind field assumption violations for the along-track and for the across-track wind is reduced from 0.36 to 0.08 m s−1 and from 0.35 to 0.24 m s−1, respectively. Although the study focuses on Doppler radars, the results are equally applicable to conically scanning Doppler lidars as well.

A 3D map of englacial attenuation rate from radar reflections at Law Dome, East Antarctica

The East Antarctic Ice Sheet (EAIS) is the largest source of potential sea-level rise, containing approximately 52 m of sea-level equivalent. To constrain estimates of sea-level rise into the future requires knowledge of ice-sheet properties and geometry and ice-penetrating radar offers a means to estimate these properties (e.g. ice thickness, englacial temperatures). One of the regions that have been extensively surveyed using ice-penetrating radar from the Investigating the Cryospheric Evolution of the Central Antarctic Plate (ICECAP) project in East Antarctica is Law Dome, a small independent ice cap situated to the west of Totten Ice Shelf. The ice cap is slow-moving, has a low melt-rate at the surface and moderate wind speeds, making it a useful study site for estimating the radar attenuation. A new method is proposed for the estimation of attenuation rate from radar data which is mathematically modelled as a constrained regularised l2 minimisation problem. In the proposed method, only radar data is required and the englacial reflectors are automatically detected from the radar data itself. To validate our methodology, attenuation differences at flight crossover points are calculated and statistical analyses performed to assess the accuracy of the results. For spatial analyses, the errors are of the order 22.6 %, 15.2 %, and 32.8 % for mean absolute error, median absolute error, and root mean square error respectively. Also, for the depth analyses, up to the depth of 800 m, the errors are under 29.9 %, 24.2 %, and 38.8 % for mean absolute error, median absolute error, and root mean square error respectively. A final product of 3D attenuation rates and uncertainty estimates is provided. The generated dataset is publicly available at https://doi.org/10.25959/5e6851e266f4a (Abdul Salam, 2020).

Accuracy of common IOL power formulas in 611 eyes based on axial length and corneal power ranges

AimsTo provide clinical guidance on the use of intraocular lens (IOL) power calculation formulas according to the biometric parameters.Methods611 eyes that underwent cataract surgery were retrospectively analysed in subgroups according to the axial length (AL) and corneal power (K). The predicted residual refractive error was calculated and compared to evaluate the accuracy of the following formulas: Haigis, Hoffer Q, Holladay 1 and SRK/T. Furthermore, the percentages of eyes with ≤±0.25, ≤±0.5 and 1 dioptres (D) of the prediction error were recorded.ResultsThe Haigis formula showed the highest percentage of cases with ≤0.5 D in eyes with a short AL and steep K (90%), average AL and steep cornea (73.2%) but also in long eyes with a flat and average K (65% and 72.7%, respectively). The Hoffer Q formula delivered the lowest median absolute error (MedAE) in short eyes with an average K (0.30 D) and Holladay 1 in short eyes with a steep K (Holladay 1 0.24 D). SRK/T presented the highest percentage of cases with ≤0.5 D in average long eyes with a flat and average K (80.5% and 68.1%, respectively) and the lowest MedAE in long eyes with an average K (0.29 D).ConclusionOverall, the Haigis formula shows accurate results in most subgroups. However, attention must be paid to the axial eye length as well as the corneal power when choosing the appropriate formula to calculate an IOL power, especially in eyes with an unusual biometry.

T2 formula in a highly myopic population, comparison with other methods and description of an improved approach for estimating corneal height

Background To determine the accuracy of the T2 formula as applied to highly myopic eyes, to compare the T2 formula to the SRK/T and Holladay 1 formulas, and to describe possible ways to improve the estimation of corneal height and prediction error in two settings, the Hadassah Hospital, Ophthalmology Department, Jerusalem, Israel and Clínica Barraquer, Bogotá, Colombia. Methods In this retrospective case series, optical biometer measurements were taken for 63 highly myopic patients (> 25 mm) undergoing uneventful crystalline lens phacoemulsification and insertion of an acrylic intraocular lens. Prediction errors were obtained, with estimations of ±0.50 D, ± 1.00 D, and greater than ±2.00 D. A method to improve the corneal height calculation is described. Results The SRK/T formula (mean absolute error [MAE] = 0.418; median absolute error [MedAE] = 0.352) was the most accurate, followed by the T2 (MAE = 0.435; MedAE = 0.381) and Holladay 1 (MAE = 0.455; MedAE = 0.389) formulas. Both the SRK/T and T2 formulas overestimated corneal height, but values were higher with the T2 formula. Corneal height was more precisely estimated using an alternative method that, when combined with axial length optimization, resulted in lower MAE (0.425) and MedAE (0.365) values than when applying the T2 formula alone. Conclusions The T2 formula seems to be less accurate than the SRK/T formula in highly myopic eyes. An improved corneal height estimation method is described for the the T2 formula.

Comparison of using Galilei Dual Scheimpflug Analyzer G4 and Barrett formula in predicting low cylinder preoperatively for cataract surgeries

Purpose: To compare corneal astigmatism estimation from Barrett toric calculator, with measurement from Galilei Dual Scheimpflug Analyzer G4 in low corneal cylinder patients. Methods: Preoperative corneal astigmatism was measured using Auto Kerato-Refractometer (AutoKM), IOL Master, and Galilei G4 (combined Placido-dual Scheimpflug analyzer) and was processed by Barrett toric calculator with measurements obtained from Auto Kerato-Refractometer and from IOL Master. A total of 42 eyes undergoing cataract surgery with nontoric intraocular lens implantation were included. Corneal astigmatism was calculated based on manifest refractive astigmatism with implications of surgically induced astigmatism. Errors in predicted residual astigmatism were calculated by the difference between postoperative manifest cylindrical refractive error and preoperative corneal cylinder using vector analysis. Results: Centroid error in predicted residual astigmatism was with-the-rule 0.36 D for AutoKM and 0.48 D for IOL Master, was lower at 0.24 D for the Barrett–IOL Master, and was lowest at 0.21 D for the Barrett–AutoKM ( p < .001). The Galilei G4 demonstrated the highest centroid error for SimK (0.53 D) and lower for total corneal power (0.49 D). The Barrett toric calculator obtained the lowest median absolute error in predicted residual astigmatism for AutoKM (0.43 D) and IOL Master (0.54 D). The Barrett–IOL Master demonstrated that 61% and 76% of eyes were within 0.50 and 0.75 D of the predicted residual astigmatism, respectively. Conclusion: The Barrett–IOL Master had more accurate prediction of residual astigmatism for low astigmatism eyes before cataract surgery compared to Galilei Dual Scheimpflug Analyzer G4 in this study.