No abstract

No abstract

Purpose The objective of the study is to develop deep learning models using synthetic fundus images to assess the direction (intorsion versus extorsion) and amount (physiologic versus pathologic) of static ocular torsion. Static ocular torsion assessment is an important clinical tool for classifying vertical ocular misalignment; however, current methods are time-intensive with steep learning curves for frontline providers. Methods We used a dataset ( n = 276) of right eye fundus images. The disc-foveal angle was calculated using ImageJ to generate synthetic images via image rotation. Using synthetic datasets ( n = 12,740 images per model) and transfer learning (the reuse of a pretrained deep learning model on a new task), we developed a binary classifier (intorsion versus extorsion) and a multiclass classifier (physiologic versus pathologic intorsion and extorsion). Model performance was evaluated on unseen synthetic and nonsynthetic data. Results On the synthetic dataset, the binary classifier had an accuracy and area under the receiver operating characteristic curve (AUROC) of 0.92 and 0.98, respectively, whereas the multiclass classifier had an accuracy and AUROC of 0.77 and 0.94, respectively. The binary classifier generalized well on the nonsynthetic data (accuracy = 0.94; AUROC = 1.00). Conclusions The direction of static ocular torsion can be detected from synthetic fundus images using deep learning methods, which is key to differentiate between vestibular misalignment (skew deviation) and ocular muscle misalignment (superior oblique palsies). Translational Relevance Given the robust performance of our models on real fundus images, similar strategies can be adopted for deep learning research in rare neuro-ophthalmologic diseases with limited datasets.

No abstract

PurposeTo quantify the T1 and T2 values of CSF in the subarachnoid space (SAS) at 3 T and interpret them in the context of water exchange between CSF and brain tissues.MethodsCSF T1 was measured using inversion recovery, and CSF T2 was assessed using T2‐preparation. T1 and T2 values in the SAS were compared with those in the frontal horns of lateral ventricles, which have less brain‐CSF exchange. Phantom experiments were performed to examine whether there were spatial variations in T1 and T2 that were unrelated to brain‐CSF exchange. Simulations were conducted to investigate the relationship between the brain‐CSF exchange rate and the apparent T1 and T2 values of SAS CSF.ResultsThe CSF T1 and T2 values were 4308.7 ± 146.9 ms and 1885.5 ± 67.9 ms, respectively, in the SAS and were 4454.0 ± 187.9 ms and 2372.9 ± 72.0 ms in the frontal horns. The SAS CSF had shorter T1 (p = 0.006) and T2 (p < 0.0001) than CSF in the frontal horns. Phantom experiments showed negligible (< 6 ms for T1; < 1 ms for T2) spatial variations in T1 and T2, suggesting that the T1 and T2 differences between SAS and frontal horns were largely attributed to physiological reasons. Simulations revealed that faster brain‐CSF exchange rates lead to shorter apparent T1 and T2 of SAS CSF. However, the experimentally observed T2 difference between SAS and frontal horns was greater than that attributable to typical exchange effect, suggesting that the T2 shortening in SAS may reflect a combined effect of exchange and deoxyhemoglobin susceptibility.ConclusionQuantification of SAS CSF relaxation times may be useful to assess the brain‐CSF exchange.