Deep-Learning Based Automated Segmentation and Quantitative Volumetric Analysis of Orbital Muscle and Fat for Diagnosis of Thyroid Eye Disease.

Purpose: Thyroid eye disease (TED) is characterized by proliferation of orbital tissues and complicated by compressive optic neuropathy (CON). This study aims to utilize a deep-learning (DL)-based automated segmentation model to segment orbital muscle and fat volumes on computed tomography (CT) images and provide quantitative volumetric data and a machine learning (ML)-based classifier to distinguish between TED and TED with CON. Methods: Subjects with TED who underwent clinical evaluation and orbital CT imaging were included. Patients with clinical features of CON were classified as having severe TED, and those without were classified as having mild TED. Normal subjects were used for controls. A U-Net DL-model was used for automatic segmentation of orbital muscle and fat volumes from orbital CTs, and ensemble of Random Forest Classifiers were used for volumetric analysis of muscle and fat. Results: Two hundred eighty-one subjects were included in this study. Automatic segmentation of orbital tissues was performed. Dice coefficient was recorded to be 0.902 and 0.921 for muscle and fat volumes, respectively. Muscle volumes among normal, mild, and severe TED were found to be statistically different. A classification model utilizing volume data and limited patient data had an accuracy of 0.838 and an area under the curve (AUC) of 0.929 in predicting normal, mild TED, and severe TED. Conclusions: DL-based automated segmentation of orbital images for patients with TED was found to be accurate and efficient. An ML-based classification model using volumetrics and metadata led to high diagnostic accuracy in distinguishing TED and TED with CON. By enabling rapid and precise volumetric assessment, this may be a useful tool in future clinical studies.

Tackling prediction uncertainty in machine learning for healthcare.

Predictive machine-learning systems often do not convey the degree of confidence in the correctness of their outputs. To prevent unsafe prediction failures from machine-learning models, the users of the systems should be aware of the general accuracy of the model and understand the degree of confidence in each individual prediction. In this Perspective, we convey the need of prediction-uncertainty metrics in healthcare applications, with a focus on radiology. We outline the sources of prediction uncertainty, discuss how to implement prediction-uncertainty metrics in applications that require zero tolerance to errors and in applications that are error-tolerant, and provide a concise framework for understanding prediction uncertainty in healthcare contexts. For machine-learning-enabled automation to substantially impact healthcare, machine-learning models with zero tolerance for false-positive or false-negative errors must be developed intentionally.

Visual field and orbital computed tomography correlation in dysthyroid optic neuropathy due to thyroid eye disease.

Purpose: The pathogenesis of dysthyroid optic neuropathy (DON) in thyroid eye disease (TED) is thought to be compression of the apical optic nerve by hypertrophied extraocular muscles. We correlated worsening DON to the area occupied by extraocular muscles.Methods: Records of adults with TED DON evaluated from 1/1/2013 to 1/1/2018 were retrospectively reviewed. Each patient's visual field with the worst mean deviation (MD) was selected. Orbit CT scans were reviewed. Reformatted oblique coronal images were created perpendicular to the optic nerve. The cross-sectional area (CSA) of the orbit and each muscle group was measured and expressed as ratios of the CSA of the orbital apex. Univariate and multivariate analysis was performed for predictors of HVF MD.Results: 34 orbits with TED DON were analyzed. On orbital CT, the superior muscle complex occupied 15% of the apex (range 6-26%), inferior 18% (range 6-33%), lateral 10% (range 4-18%), medial 17% (range 8-27%), and all combined 61% (range 28-80%). Increasing total muscle area and superior complex area correlated with worsening MD. In multivariate linear regression, the superior muscle complex remained a significant predictor of MD (p = 0.01) over total muscle area (p = 0.25).Conclusions: Enlargement of extraocular muscles is common in TED, but DON occurs in only 6%. Our findings demonstrate that as DON worsens, as quantified by visual field MD, the superior muscle complex crowds the apex. This is consistent with the typical inferior visual field findings seen in TED DON. Hypertrophy of the superior rectus and levator palpabrae superioris complex may be predictive of worsening DON.

Clinical and radiographic features of hydrolyzed MIRAgel scleral buckles: A comparative analysis.

The MIRAgel (hydrogel) scleral buckle, introduced in the 1980s, was a novel material to repair retinal detachments. It was later discontinued due to the frequency of long-term complications related to buckle hydrolysis and expansion. These complications included pain, limited extraocular motility, and more serious complications such as infection or scleral perforation, which ultimately necessitated surgical extraction as late as 20-30 years after placement. Prompt and proper diagnosis and treatment is often delayed as these buckle-associated complications frequently mimic other orbital pathologies such as tumors or infections. The hydrolyzed MIRAgel buckle exhibits distinct radiographic features that are helpful in arriving at the correct diagnosis, particularly in cases of ambiguous clinical presentation or history. Here, we expand on the previously described radiographic features of hydrolyzed MIRAgel and compare them to features of common, mimicking orbital pathology.

Revision eDCR using a superior pedicled mucosal flap.

BACKGROUND: Endoscopic dacryocystorhinostomies (eDCRs) show patency rates between 81% and 94%. However, dacryocystorhinostomy (DCR) failure and the need for revision remain a significant challenge. One of the principal challenges in revision eDCR is the need to surgically identify the correct osteotomy site and maintain long-term patency in the setting of previously instrumented and potentially scarred tissue. At the same time, the surgeon must assume that the blood supply to the commonly described anterior and posteriorly pedicled flaps has been compromised. OBJECTIVE: The objective of the study is to describe a novel flap technique for revision eDCR. METHODS: The superior based mucosal flap is a novel technique that provides a vascularized mucosa preserving technique in revision eDCR despite previous instrumentation of the lacrimal system. This technique provides wide exposure of the revision osteotomy site while simultaneously allowing a viable mucosal flap to be replaced at the conclusion of the procedure, thereby minimizing bone exposure and cicatricial restenosis. RESULTS: The authors have utilized this technique in 13 procedures with 100% positive identification of the lacrimal sac, a 0% complication rate, and a 100% success rate after a mean follow-up of 26.93 ± 10.33 months (range 6-35 months). CONCLUSION: The eDCR using the superior pedicled mucosal flap provides excellent exposure of the maxillary bone and the lacrimal sac. This method preserves vascularity of the flap using a superiorly based pedicle which is typically inviolate during both open and endoscopic primary DCR. The mucosal flap can then be replaced, thereby minimizing bone exposure and optimizing patency.

Sonoporation enhances chemotherapeutic efficacy in retinoblastoma cells in vitro.

PURPOSE: To study the ability of ultrasound (US) and microbubbles (MB) to enhance chemotherapeutic efficacy against retinoblastoma Y79 cells in vitro. METHODS: The experiment was performed in three stages. The authors first compared cell viability of Y79 cells exposed to doxorubicin versus cells exposed to doxorubicin combined with low-intensity, low-frequency US + MB. They then evaluated enhanced cell permeability by studying the intensity of intracellular fluorescence in cells exposed to doxorubicin versus those exposed to doxorubicin with US + MB. Lastly they evaluated the morphologic characteristics of the cells by scanning electron microscopy (SEM) to identify the presence of pores. RESULTS: The Y79 cells exposed to doxorubicin with US + MB showed a significant decrease in cell viability at 72 hours compared with those exposed to doxorubicin alone (P = 0.02). Cells also showed immediate increased permeability to doxorubicin with the addition of US + MB compared with doxorubicin alone, which continued to increase over 60 minutes. SEM did not demonstrate physical pores at the lowest US + MB intensity shown to enhance intracellular doxorubicin fluorescence. CONCLUSIONS: US + MB facilitates the uptake of chemotherapy in retinoblastoma Y79 cells in vitro. This occurs in the absence of visible pores, suggesting a possible secondary mechanism for increased drug delivery. This experiment is the first step toward enhancing chemotherapy with sonoporation in the treatment of intraocular tumors. This technique may lead to more effective chemotherapy treatments with less collateral damage to ocular tissues and may allow reduced systemic dosage and systemic side effects.