Clinician-Guided Deep Learning Segmentation of Skull Base Pneumatization on Computed Tomography Using 3D Slicer and MONAI Label.
Authors
Affiliations (9)
Affiliations (9)
- Department of Neurosurgery, County Emergency Clinical Hospital of Târgu Mureș, 50 Gheorghe Marinescu Street, 540136 Târgu Mureș, Romania.
- Doctoral School, George Emil Palade University of Medicine, Pharmacy, Science and Technology of Târgu Mureș, 38 Gheorghe Marinescu Street, 540139 Târgu Mureș, Romania.
- Department of Anatomy, George Emil Palade University of Medicine, Pharmacy, Science and Technology of Târgu Mureș, 38 Gheorghe Marinescu Street, 540139 Târgu Mureș, Romania.
- Department of Radiology, County Emergency Clinical Hospital of Târgu Mureș, 50 Gheorghe Marinescu Street, 540136 Târgu Mureș, Romania.
- Department of Radiology ME2, George Emil Palade University of Medicine, Pharmacy, Science and Technology of Târgu Mureș, 38 Gheorghe Marinescu Street, 540139 Târgu Mureș, Romania.
- Department of Ophthalmology, Mureș County Clinical Hospital, 6 Bernády György Street, 540072 Târgu Mureș, Romania.
- Department of Ophthalmology, George Emil Palade University of Medicine, Pharmacy, Science and Technology of Târgu Mureș, 38 Gheorghe Marinescu Street, 540139 Târgu Mureș, Romania.
- Department of ENT, Pius Brînzeu County Emergency Clinical Hospital of Timișoara, 156 Liviu Rebreanu Boulevard, 300723 Timișoara, Romania.
- Department of ENT George Emil Palade University of Medicine, Pharmacy, Science and Technology of Târgu Mureș, 38 Gheorghe Marinescu Street, 540139 Târgu Mureș, Romania.
Abstract
Skull base pneumatization is anatomically variable and clinically relevant to temporal bone and transsphenoidal surgical corridors, but manual volumetric segmentation is time-consuming. This retrospective pilot study evaluated a clinician-guided deep learning workflow for mastoid and sphenoid sinus compartment segmentation on bone computed tomography. Images were curated and annotated in 3D Slicer using MONAI Label and separate three-dimensional SegResNet models. The mastoid development dataset comprised 122 side-cases, with 28 reserved side-cases; the sphenoid dataset comprised 117 development and 17 reserved examinations. The best internal-validation Dice scores were 0.8660 for the mastoid model and 0.8750 for the sphenoid model. In AI-assisted correction cohorts, mean Dice ranged from 0.9539 to 0.9691 for mastoid and from 0.9347 to 0.9426 for sphenoid. In independently annotated subsets, AI-to-expert Dice was 0.8083-0.8097 for mastoid and 0.8339-0.8473 for sphenoid, while interobserver Dice was 0.8003 and 0.9136, respectively. AI assistance reduced mean segmentation time by 86.5% for mastoid and 81.4% for sphenoid. These findings support clinician-supervised AI segmentation as an efficient starting point for volumetric assessment.