Automated Brain Segmentation on Computed Tomographic Images Using Perceptual Loss Based Convolutional Neural Networks
2
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Purpose: This study aimed to develop a new convolutional neural network-based deep learning (DL) technique for automated brain tissue segmentation from computed tomographic (CT) scans and to evaluate its performance in comparison to magnetic resonance imaging (MRI)-derived segmentations. Materials and Methods: This multicenter retrospective study collected paired CT and MRI data from 199 healthy individuals across two institutions. The data were divided into a training set (n = 100) and an internal test set (n = 50) from one institution, with additional datasets (n = 49) from the second institution for external validation. Ground truth masks for gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) were generated from T1-weighted MR images. A U-Net-based DL model was trained for each of the three brain regions, with a perceptual loss computed from VGG19. Model performance was evaluated by calculating continuous Dice coefficient (cDice), intersection-over-union (IOU), and 95th percentile Hausdorff distance (HD95). Volumetric estimates from CT-based segmentations were compared with MRI-derived volumes using the coefficient of determination (R2), intraclass correlation coefficients (ICC), and Bland-Altman analysis. Results: The DL network trained with the perceptual loss showed superior performance, compared with that trained without the perceptual loss. In internal tests, evaluation scores (without perceptual loss vs. with perceptual loss) were: cDice = 0.717 vs. 0.765 and HD95 = 6.641 mm vs. 6.314 mm in GM; cDice = 0.730 vs. 0.767 and HD95 = 5.841 mm vs. 5.644 mm in WM; and cDice = 0.600 vs. 0.630 and HD95 = 5.641 mm vs. 5.362 mm in CSF, respectively. Volumetric analyses revealed strong agreement between MRI-derived ground truth and CT-based segmentations with R2 = 0.83/0.90 and 0.85/0.87, and ICC = 0.91/0.94 and 0.92/0.93 for GM and WM, respectively, in internal/external tests. Conclusion: The proposed DL method, enhanced with perceptual loss, improves brain tissue segmentation from CT images. This approach shows promise as an alternative to MRIbased segmentation.
Abstract
Purpose: This study aimed to develop a new convolutional neural network-based deep learning (DL) technique for automated brain tissue segmentation from computed tomographic (CT) scans and to evaluate its performance in comparison to magnetic resonance imaging (MRI)-derived segmentations. Materials and Methods: This multicenter retrospective study collected paired CT and MRI data from 199 healthy individuals across two institutions. The data were divided into a training set (n = 100) and an internal test set (n = 50) from one institution, with additional datasets (n = 49) from the second institution for external validation. Ground truth masks for gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) were generated from T1-weighted MR images. A U-Net-based DL model was trained for each of the three brain regions, with a perceptual loss computed from VGG19. Model performance was evaluated by calculating continuous Dice coefficient (cDice), intersection-over-union (IOU), and 95th percentile Hausdorff distance (HD95). Volumetric estimates from CT-based segmentations were compared with MRI-derived volumes using the coefficient of determination (R2), intraclass correlation coefficients (ICC), and Bland-Altman analysis. Results: The DL network trained with the perceptual loss showed superior performance, compared with that trained without the perceptual loss. In internal tests, evaluation scores (without perceptual loss vs. with perceptual loss) were: cDice = 0.717 vs. 0.765 and HD95 = 6.641 mm vs. 6.314 mm in GM; cDice = 0.730 vs. 0.767 and HD95 = 5.841 mm vs. 5.644 mm in WM; and cDice = 0.600 vs. 0.630 and HD95 = 5.641 mm vs. 5.362 mm in CSF, respectively. Volumetric analyses revealed strong agreement between MRI-derived ground truth and CT-based segmentations with R2 = 0.83/0.90 and 0.85/0.87, and ICC = 0.91/0.94 and 0.92/0.93 for GM and WM, respectively, in internal/external tests. Conclusion: The proposed DL method, enhanced with perceptual loss, improves brain tissue segmentation from CT images. This approach shows promise as an alternative to MRIbased segmentation.
참고문헌 (26)
- Biondetti E, "Cerebral oxygen metabolism from MRI susceptibility", Neuroimage, 276, pp.120189~ (2023)
- Shin W, "Automated brain tissue segmentation based on fractional signal mapping from inversion recovery Look-Locker acquisition", Neuroimage, 52, pp.1347~1354 (2010)
- Deepa V, "Gray matter and white matter segmentation from MRI brain images using clustering methods", Int Res J Eng Tech(IRJET), 2, pp.913~921 (2015)
- Kim J, "Automated quantification of total cerebral blood flow from phase-contrast MRI and deep learning", J Imaging Inform Med, 37, pp.563~574 (2024)
- "SPM12"[URL]
- Zhang F, "Deep learning based segmentation of brain tissue from diffusion MRI", Neuroimage, 233, pp.117934~ (2021)
- Bawil MB, "Specialized gray matter segmentation via a generative adversarial network : application on brain white matter hyperintensities classification", Front Neurosci, 18, pp.1416174~ (2024)
- Srikrishna M, "Deep learning from MRI-derived labels enables automatic brain tissue classification on human brain CT", Neuroimage, 244, pp.118606~ (2021)
- "FreeSurfer"[URL]
- Luan X, "Rubik-Net : learning spatial information via rotation-driven convolutions for brain segmentation", IEEE J Biomed Health Inform, 26, pp.289~300 (2022)
- Zhuang Y, "APRNet : a 3D anisotropic pyramidal reversible network with multi-modal crossdimension attention for brain tissue segmentation in MR images", IEEE J Biomed Health Inform, 26, pp.749~761 (2022)
- Niu K, "P-ResUnet : segmentation of brain tissue with purified residual unet", Comput Biol Med, 151(Pt B), pp.106294~ (2022)
- Khalili N, "Automatic brain tissue segmentation in fetal MRI using convolutional neural networks", Magn Reson Imaging, 64, pp.77~89 (2019)
- Kumar P, "U-SegNet: fully convolutional neural network based automated brain tissue segmentation tool", 2018 25th IEEE International Conference on Image Processing (ICIP), pp.3503~3507 (2018), IEEE
- Mugler JP 3rd, "Three-dimensional magnetizationprepared rapid gradient-echo imaging(3D MP RAGE)", Magn Reson Med, 15, pp.152~157 (1990)
- Ronneberger O, "Medical image computing and computer-assisted intervention. MICCAI 2015", pp.234~241 (2015), Springer
- He K, "Deep residual learning for image recognition", 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.770~778 (2016), IEEE
- Johnson J, "Computer Vision – ECCV 2016", pp.694~711 (2016), Springer
- Deng J, "ImageNet: a large-scale hierarchical image database", 2009 IEEE Conference on Computer Vision and Pattern Recognition, pp.248~255 (2009), IEEE
- Sandor T, "Segmentation of brain CT images using the concept of region growing", Int J Biomed Comput, 29, pp.133~147 (1991)
- Irimia A, "Brain segmentation from computed tomography of healthy aging and geriatric concussion at variable spatial resolutions", Front Neuroinform, 13, pp.9~ (2019)
- Cai JC, "Fully automated segmentation of head CT neuroanatomy using deep learning", Radiol Artif Intell, 2, pp.e190183~ (2020)
- Wang T, "Deep learning-based automated segmentation of eight brain anatomical regions using head CT images in PET", CT, 2022, pp.22, 99~