Related Experiment Video
Updated: Nov 7, 2025
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
Fully Automated Segmentation of Head CT Neuroanatomy Using Deep Learning
Jason C Cai1, Zeynettin Akkus1, Kenneth A Philbrick1
1Departments of Radiology (J.C.C., K.A.P., S.H., P.R., G.M.C., D.C.V., Q.H., B.J.E.) and Cardiovascular Science (Z.A.), Mayo Clinic Rochester, 200 First St. SW, RO_PB_02_RIL, Rochester, MN 55905; Department of Radiology, Khon Kaen University, Khon Kaen, Thailand (A.B.); Department of Health Sciences Research, Mayo Clinic Florida, Jacksonville, Fla (A.D.W.); and Department of Internal Medicine, Ascension St. John Hospital, Detroit, Mich (A.Z.).
A deep learning model accurately segments intracranial structures on head CT scans, showing high performance comparable to experts. This automated segmentation is feasible and generalizes well to various scans, including those with idiopathic normal pressure hydrocephalus (iNPH).
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Radiology
Background:
- Accurate segmentation of intracranial structures on CT scans is crucial for diagnosis and treatment planning.
- Manual segmentation is time-consuming and prone to interobserver variability.
- Deep learning offers a potential solution for automated and efficient segmentation.
Purpose of the Study:
- To develop and evaluate a deep learning model for segmenting eleven intracranial structures on head CT scans.
- To assess the model's accuracy and generalizability on primary and external datasets.
Main Methods:
- A deep learning model was trained on 62 normal head CT scans, with manual annotation of eleven intracranial structures.
- The model underwent rigorous evaluation on primary test datasets and two secondary datasets, including scans with idiopathic normal pressure hydrocephalus (iNPH).
- Interobserver variability was assessed, and techniques like within-network normalization and class weighting were employed.
Main Results:
- The model achieved an overall Dice coefficient of 0.84 ± 0.05 on the primary test dataset, with high performance for structures like the brainstem and cerebrum (0.96 ± 0.01).
- Segmentation accuracy was comparable to expert annotations and superior to existing methods.
- The model demonstrated robustness on external CT scans and those with ventricular enlargement, including iNPH cases.
Conclusions:
- Automated segmentation of CT neuroanatomy using deep learning is highly accurate and feasible.
- The developed model generalizes effectively to diverse CT scans, including those with idiopathic normal pressure hydrocephalus (iNPH).
- This technology has the potential to significantly improve the efficiency and consistency of neuroimaging analysis.
