Open-Source Automatic Segmentation of Ocular Structures and Biomarkers of Microbial Keratitis on Slit-Lamp Photography Images Using Deep Learning.

We propose a fully-automatic deep learning-based algorithm for segmentation of ocular structures and microbial keratitis (MK) biomarkers on slit-lamp photography (SLP) images. The dataset consisted of SLP images from 133 eyes with manual annotations by a physician, P1. A modified region-based convolutional neural network, SLIT-Net, was developed and trained using P1's annotations to identify and segment four pathological regions of interest (ROIs) on diffuse white light images (stromal infiltrate (SI), hypopyon, white blood cell (WBC) border, corneal edema border), one pathological ROI on diffuse blue light images (epithelial defect (ED)), and two non-pathological ROIs on all images (corneal limbus, light reflexes). To assess inter-reader variability, 75 eyes were manually annotated for pathological ROIs by a second physician, P2. Performance was evaluated using the Dice similarity coefficient (DSC) and Hausdorff distance (HD). Using seven-fold cross-validation, the DSC of the algorithm (as compared to P1) for all ROIs was good (range: 0.62-0.95) on all 133 eyes. For the subset of 75 eyes with manual annotations by P2, the DSC for pathological ROIs ranged from 0.69-0.85 (SLIT-Net) vs. 0.37-0.92 (P2). DSCs for SLIT-Net were not significantly different than P2 for segmenting hypopyons (p > 0.05) and higher than P2 for WBCs (p < 0.001) and edema (p < 0.001). DSCs were higher for P2 for segmenting SIs (p < 0.001) and EDs (p < 0.001). HDs were lower for P2 for segmenting SIs (p = 0.005) and EDs (p < 0.001) and not significantly different for hypopyons (p > 0.05), WBCs (p > 0.05), and edema (p > 0.05). This prototype fully-automatic algorithm to segment MK biomarkers on SLP images performed to expectations on an exploratory dataset and holds promise for quantification of corneal physiology and pathology.

Patient Reported Outcomes in Microbial Keratitis.

PURPOSE: The aim of this investigation was to study the patient-reported outcomes of patients with microbial keratitis (MK) using the 9-item National Eye Institute-Visual Function Questionnaire (NEI VFQ-9). METHODS: Using the Sight Outcomes Research Collaborative ophthalmology electronic health record repository, patients with MK and control patients who completed the NEI VFQ-9 within 7 days of their appointment were identified. The questionnaire is scored as a mean of the 9 items on a scale from 0 to 100, with higher scores indicating better functioning. Composite and individual item scores were compared between groups using the analysis of variance. RESULTS: In total, 916 questionnaires were completed from patients with acute MK (n = 84), nonacute MK (n = 30), MK with a corneal transplant (n = 21), from controls seen in a satellite comprehensive ophthalmology clinic (n = 528), and controls seen at a subspecialty ophthalmology clinic (n = 253). The mean NEI VFQ-9 composite scores per group were 66.6 (SD = 26.8), 78.1 (SD = 17.1), 58.6 (SD =21.6), 88.0 (SD = 10.2), and 83.5 (SD = 13.0), respectively (P < 0.0001). Both patients with acute MK and patients with MK requiring transplant reported significantly worse function than nonacute MK, comprehensive, and specialty patients. Patients with nonacute MK reported significantly worse function than comprehensive control patients (all Tukey-adjusted P < 0.05). DISCUSSION: Patients who had or eventually require corneal transplant for management of their MK report worse visual function than patients with nonacute MK. This may be important in helping physicians counsel their patients.

Stereotactic Navigation Improves Outcomes of Orbital Decompression Surgery for Thyroid Associated Orbitopathy.

PURPOSE: To test whether intraoperative stereotactic navigation during orbital decompression surgery resulted in quantifiable surgical benefit. METHODS: This retrospective cohort study examined all consecutive patients who underwent primary orbital decompression surgery for thyroid associated orbitopathy performed by a single surgeon (A.K.) during the periods of 2012-2014 (non-navigated), and 2017-2018 (navigated). The study was HIPAA-compliant, was approved by the Institutional Review Board, and adhered to the tenets of the Helsinki declaration. Recorded parameters included patient age, sex, race, decompression technique (side of operation and walls decompressed), estimated blood loss (EBL), intraoperative complications, times that patient entered and exited the operating room (OR), times of surgical incision and dressing completion, pre- and postoperative best corrected visual acuity (BCVA), proptosis, diplopia, postoperative change in strabismus deviation, and need for subsequent strabismus surgery. Recorded times were used to calculate operating time (initial incision to dressing) and maintenance time (time between OR entry and initial incision and time between dressings and OR exit). The total maintenance time was averaged over total number of operations. Student t test was used to compare surgical times, maintenance times, EBL, and proptosis reduction. Fisher exact test was used to compare BCVA change, strabismus deviation change, resolution or onset of diplopia, and need for corrective strabismus surgery. RESULTS: Twenty-two patients underwent primary orbital decompression surgery without navigation, and 23 patients underwent navigation-guided primary orbital decompression surgery. There were no intraoperative complications in either group. The average operative time was shorter in the navigated group for a unilateral balanced decompression (n = 10 vs. 19; 125.8 ± 13.6 vs. 141.3 ± 19.4 min; p-value = 0.019), and a unilateral lateral wall only decompression (n = 13 vs. 3; 80.5 ± 12.8 vs. 93.0 ± 6.1 min; p-value = 0.041). The average maintenance time per surgery was not significantly different between the non-navigated group (51.3 ± 12.7 min) and the navigated group (50.5 ± 6.4 min). There was no significant difference between the navigated and non-navigated groups in average EBL per surgery. There was no significant difference in BCVA change. Average proptosis reduction was larger in the navigated group, but this was not significant. There was a significantly lower proportion of patients who required corrective strabismus surgery following decompression in the navigated group than in the non-navigated group (39.1% vs. 77.3%, p-value = 0.012). CONCLUSIONS: Intraoperative stereotactic navigation during orbital decompression surgery has the potential to provide the surgeon with superior spatial awareness to improve patient outcomes. This study found that use of intraoperative navigation reduced operative time (even without factoring in a resident teaching component) while also reducing the need for subsequent strabismus surgery. This study is limited by its size but illustrates that use of intraoperative navigation guidance has substantive benefits in orbital decompression surgery.

Algorithm Variability in Quantification of Epithelial Defect Size in Microbial Keratitis Images.

PURPOSE: To investigate the sources of measurement variability when quantifying the morphology of microbial keratitis (MK) from slit-lamp photography (SLP) images using a semiautomated, image-analysis algorithm. METHODS: Prospectively enrolled patients with MK underwent SLP to obtain images of their epithelial defects (ED). Eyes were stained with fluorescein and imaged multiple times under blue light, at low and high magnifications. A masked research assistant chose the 3 best images and annotated each 3 times to provide seed regions corresponding to ED and healthy cornea. The algorithm returned the ED area for each seeded image. Eyes without EDs and algorithm failures were excluded. Variance components were estimated with a random effects model and intraclass correlation coefficients estimated with intragrader reliability. RESULTS: A total of 42 eyes from 42 MK participants were photographed. After excluding poor quality images, eyes with no EDs, and algorithm failures, 34 patients with 92 images and 274 seeds were analyzed. No significant differences in the average ED area were found between seedings or high- versus low-SLP magnifications (all P > 0.5, paired t tests). Minimal measurement variability was because of image (0.9%), magnification (0.2%), or seed (0.1%). Most variability was attributable to differences in ED sizes between patients (85.2%). 13.7% of variability was unexplained. Multiple iterations of the algorithm on the same image showed good consistency (intraclass correlation coefficient = 0.98, 95% confidence interval, 0.97-0.99). CONCLUSIONS: Image-analysis algorithms showed good reliability for measuring the ED area from SLP images. Most measurement variability was because of between-patient differences, not imaging settings or application of the algorithm by the user.