This paper presents a correction of a deep learning study for optic nerve diameter characterization and structure quantification in transorbital ultrasound images.[6][7] Optical nerve quantification serves as a biomarker for the non-invasive assessment of elevated intracranial pressure and other neuro-ophthalmic conditions.[1] Manual identification of the optic nerve is time-consuming and resource-intensive, while the accuracy of automated methods depends on the quality of ultrasound images.[1] The authors proposed a branched deep neural network for shared and specific feature extraction along with an uncertainty-aware loss function that enables robust object structure learning.[1] Experiments on a multicenter publicly available dataset showed a Dice score of 73.3% and an AUROC of 84.5% on the test dataset, outperforming the state of the art.[1][2] The model achieved a mean absolute error (MAE) of 0.26 mm for optic nerve diameter (OND) and 0.41 mm for optic nerve sheath diameter (ONSD), with intracellular correlation coefficients (ICC) of 0.67 for OND and 0.66 for ONSD.[2] Images were standardized to 256 × 256 pixels with expert manual annotations of the optic nerve and its sheath.[2]