Premium
Automated delineation of orbital abscess depicted on CT scan using deep learning
Author(s) -
Fu Roxana,
Leader Joseph K.,
Pradeep Tejus,
Shi Junli,
Meng Xin,
Zhang Yanchun,
Pu Jiantao
Publication year - 2021
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1002/mp.14907
Subject(s) - jaccard index , artificial intelligence , context (archaeology) , hausdorff distance , convolutional neural network , deep learning , transfer of learning , abscess , computer science , medicine , pattern recognition (psychology) , surgery , geology , paleontology
Objectives To develop and validate a deep learning algorithm to automatically detect and segment an orbital abscess depicted on computed tomography (CT). Methods We retrospectively collected orbital CT scans acquired on 67 pediatric subjects with a confirmed orbital abscess in the setting of infectious orbital cellulitis. A context‐aware convolutional neural network (CA‐CNN) was developed and trained to automatically segment orbital abscess. To reduce the requirement for a large dataset, transfer learning was used by leveraging a pre‐trained model for CT‐based lung segmentation. An ophthalmologist manually delineated orbital abscesses depicted on the CT images. The classical U‐Net and the CA‐CNN models with and without transfer learning were trained and tested on the collected dataset using the 10‐fold cross‐validation method. Dice coefficient, Jaccard index, and Hausdorff distance were used as performance metrics to assess the agreement between the computerized and manual segmentations. Results The context‐aware U‐Net with transfer learning achieved an average Dice coefficient and Jaccard index of 0.78 ± 0.12 and 0.65 ± 0.13, which were consistently higher than the classical U‐Net or the context‐aware U‐Net without transfer learning ( P < 0.01). The average differences of the abscess between the computerized results and the experts in terms of volume and Hausdorff distance were 0.10 ± 0.11 mL and 1.94 ± 1.21 mm, respectively. The context‐aware U‐Net detected all orbital abscess without false positives. Conclusions The deep learning solution demonstrated promising performance in detecting and segmenting orbital abscesses on CT images in strong agreement with a human observer.