Abstract
Objectives
To examine the role of ADC threshold on agreement across observers and deep learning models (DLMs) plus segmentation performance of DLMs for acute ischemic stroke (AIS).
Methods
Twelve DLMs, which were trained on DWI-ADC-ADC combination from 76 patients with AIS using 6 different ADC thresholds with ground truth manually contoured by 2 observers, were tested by additional 67 patients in the same hospital and another 78 patients in another hospital. Agreement between observers and DLMs were evaluated by Bland-Altman plot and intraclass correlation coefficient (ICC). The similarity between ground truth (GT) defined by observers and between automatic segmentation performed by DLMs was evaluated by Dice similarity coefficient (DSC). Group comparison was performed using the Mann-Whitney U test. The relationship between the DSC and ADC threshold as well as AIS lesion size was evaluated by linear regression analysis. A p < .05 was considered statistically significant.
Results
Excellent interobserver agreement and intraobserver repeatability in the manual segmentation (all ICC > 0.98, p < .001) were achieved. The 95% limit of agreement was reduced from 11.23 cm2 for GT on DWI to 0.59 cm2 for prediction at an ADC threshold of 0.6 × 10−3 mm2/s combined with DWI. The segmentation performance of DLMs was improved with an overall DSC from 0.738 ± 0.214 on DWI to 0.971 ± 0.021 on an ADC threshold of 0.6 × 10−3 mm2/s combined with DWI.
Conclusions
Combining an ADC threshold of 0.6 × 10−3 mm2/s with DWI reduces interobserver and inter-DLM difference and achieves best segmentation performance of AIS lesions using DLMs.
Key Points
• Higher Dice similarity coefficient (DSC) in predicting acute ischemic stroke lesions was achieved by ADC thresholds combined with DWI than by DWI alone (all p < .05).
• DSC had a negative association with the ADC threshold in most sizes, both hospitals, and both observers (most p < .05) and a positive association with the stroke size in all ADC thresholds, both hospitals, and both observers (all p < .001).
• An ADC threshold of 0.6 × 10−3 mm2/s eliminated the difference of DSC at any stroke size between observers or between hospitals (p = .07 to > .99).
Similar content being viewed by others
Abbreviations
- ADC:
-
Apparent diffusion coefficient
- aGT:
-
ADC-gated ground truth
- AIS:
-
Acute ischemic stroke
- DLM:
-
Deep learning model
- DSC:
-
Dice similarity coefficient
- DSCg :
-
Dice similarity coefficient between two manually contoured ground truths
- DSCp :
-
Dice similarity coefficient between the prediction and the ground truth
- DWI:
-
Diffusion-weighted imaging
- ESM:
-
Electrical supplementary material
- GT:
-
Ground truth
- H1:
-
China Medical University Hospital
- H2:
-
China Medical University Hsinchu Hospital
- ICC:
-
Intraclass correlation coefficient
- LOA-95:
-
95% limits of agreement (i.e., 1.96 standard deviation)
- mGT:
-
Manually contoured ground truth
- sGT:
-
Semiautomatically defined ground truth
References
Nag MK, Koley S, China D et al (2017) Computer-assisted delineation of cerebral infarct from diffusion-weighted MRI using Gaussian mixture model. Int J Comput Assist Radiol Surg 12:539–552
Lee H, Jung K, Kang DW, Kim N (2020) Fully automated and real-time volumetric measurement of infarct core and penumbra in diffusion- and perfusion-weighted MRI of patients with hyper-acute stroke. J Digit Imaging 33:262–272
Lee H, Lee EJ, Ham S et al (2020) Machine learning approach to identify stroke within 4.5 hours. Stroke 51:860–866
Chen L, Bentley P, Rueckert D (2017) Fully automatic acute ischemic lesion segmentation in DWI using convolutional neural networks. Neuroimage Clin 15:633–643
Kim YC, Lee JE, Yu I et al (2019) Evaluation of diffusion lesion volume measurements in acute ischemic stroke using encoder-decoder convolutional network. Stroke 50:1444–1451
Prakash KNB, Gupta V, Bilello M, Beauchamp NJ, Nowinski WL (2006) Identification, segmentation, and image property study of acute infarcts in diffusion-weighted images by using a probabilistic neural network and adaptive Gaussian mixture model. Acad Radiol 13:1474–1484
Tsai JZ, Peng SJ, Chen YW et al (2014) Automatic detection and quantification of acute cerebral infarct by fuzzy clustering and histographic characterization on diffusion weighted MR imaging and apparent diffusion coefficient map. Biomed Res Int 2014:963032
Winzeck S, Mocking SJT, Bezerra R et al (2019) Ensemble of convolutional neural networks improves automated segmentation of acute ischemic lesions using multiparametric diffusion-weighted MRI. AJNR Am J Neuroradiol 40:938–945
Zhang R, Zhao L, Lou W et al (2018) Automatic segmentation of acute ischemic stroke from DWI using 3-D fully convolutional DenseNets. IEEE Trans Med Imaging 37:2149–2160
Woo I, Lee A, Jung SC et al (2019) Fully automatic segmentation of acute ischemic lesions on diffusion-weighted imaging using convolutional neural networks: comparison with conventional algorithms. Korean J Radiol 20:1275–1284
Boldsen JK, Engedal TS, Pedraza S et al (2018) Better diffusion segmentation in acute ischemic stroke through automatic tree learning anomaly segmentation. Front Neuroinform 12:21
Perez Malla CU, Valdes Hernandez MDC, Rachmadi MF, Komura T (2019) Evaluation of enhanced learning techniques for segmenting ischaemic stroke lesions in brain magnetic resonance perfusion images using a convolutional neural network scheme. Front Neuroinform 13:33
Mah YH, Jager R, Kennard C, Husain M, Nachev P (2014) A new method for automated high-dimensional lesion segmentation evaluated in vascular injury and applied to the human occipital lobe. Cortex 56:51–63
Xiong Y, Huang CC, Fisher M, Hackney DB, Bhadelia RA, Selim MH (2019) Comparison of automated CT perfusion softwares in evaluation of acute ischemic stroke. J Stroke Cerebrovasc Dis 28:104392
Maier O, Wilms M, von der Gablentz J, Kramer UM, Munte TF, Handels H (2015) Extra tree forests for sub-acute ischemic stroke lesion segmentation in MR sequences. J Neurosci Methods 240:89–100
Maier O, Schroder C, Forkert ND, Martinetz T, Handels H (2015) Classifiers for ischemic stroke lesion segmentation: a comparison study. PLoS One 10:e0145118
Kamnitsas K, Ledig C, Newcombe VFJ et al (2017) Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation. Med Image Anal 36:61–78
Wilke M, de Haan B, Juenger H, Karnath HO (2011) Manual, semi-automated, and automated delineation of chronic brain lesions: a comparison of methods. Neuroimage 56:2038–2046
Pustina D, Coslett HB, Turkeltaub PE, Tustison N, Schwartz MF, Avants B (2016) Automated segmentation of chronic stroke lesions using LINDA: lesion identification with neighborhood data analysis. Hum Brain Mapp 37:1405–1421
Ito KL, Kim H, Liew SL (2019) A comparison of automated lesion segmentation approaches for chronic stroke T1-weighted MRI data. Hum Brain Mapp 40:4669–4685
Ogata T, Christensen S, Nagakane Y et al (2013) The effects of alteplase 3 to 6 hours after stroke in the EPITHET-DEFUSE combined dataset: post hoc case-control study. Stroke 44:87–93
Katyal A, Bhaskar S (2021) CTP-guided reperfusion therapy in acute ischemic stroke: a meta-analysis. Acta Neurol Scand 143:355–366
Loubinoux I, Volk A, Borredon J et al (1997) Spreading of vasogenic edema and cytotoxic edema assessed by quantitative diffusion and T2 magnetic resonance imaging. Stroke 28:419–426 discussion 426-417
Yu Y, Xie Y, Thamm T et al (2021) Tissue at risk and ischemic core estimation using deep learning in acute stroke. AJNR Am J Neuroradiol. https://doi.org/10.3174/ajnr.A7081
Pistocchi S, Strambo D, Bartolini B et al (2021) MRI software for diffusion-perfusion mismatch analysis may impact on patients’ selection and clinical outcome. Eur Radiol. https://doi.org/10.1007/s00330-021-08211-2
Deutschmann H, Hinteregger N, Wiesspeiner U et al (2021) Automated MRI perfusion-diffusion mismatch estimation may be significantly different in individual patients when using different software packages. Eur Radiol 31:658–665
Yu Y, Xie Y, Thamm T et al (2020) Use of deep learning to predict final ischemic stroke lesions from initial magnetic resonance imaging. JAMA Netw Open 3:e200772
Juan CJ, Chang HC, Hsueh CJ et al (2009) Salivary glands: echo-planar versus PROPELLER diffusion-weighted MR imaging for assessment of ADCs. Radiology 253:144–152
Purushotham A, Campbell BC, Straka M et al (2015) Apparent diffusion coefficient threshold for delineation of ischemic core. Int J Stroke 10:348–353
Zhao B, Ding S, Wu H et al (2019) Automatic acute ischemic stroke lesion segmentation using semi-supervised learning. arXiv:1908.03735
Ronneberger O, Fischer P, Brox T (2015) U-Net: convolutional networks for biomedical image segmentation. Medical Image Computing and Computer-Assisted Intervention (MICCAI). Springer, LNCS 9351:234–241
Koo TK, Li MY (2016) A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med 15:155–163
Nazari-Farsani S, Nyman M, Karjalainen T, Bucci M, Isojarvi J, Nummenmaa L (2020) Automated segmentation of acute stroke lesions using a data-driven anomaly detection on diffusion weighted MRI. J Neurosci Methods 333:108575
Luby M, Bykowski JL, Schellinger PD, Merino JG, Warach S (2006) Intra- and interrater reliability of ischemic lesion volume measurements on diffusion-weighted, mean transit time and fluid-attenuated inversion recovery MRI. Stroke 37:2951–2956
Zhao B, Liu Z, Liu G et al (2021) Deep learning-based acute ischemic stroke lesion segmentation method on multimodal MR images using a few fully labeled subjects. Comput Math Methods Med 2021:3628179
Wu O, Winzeck S, Giese AK et al (2019) Big data approaches to phenotyping acute ischemic stroke using automated lesion segmentation of multi-center magnetic resonance imaging data. Stroke 50:1734–1741
Acknowledgements
The authors are grateful for the financial support from China Medical University Hsinchu Hospital and Taiwan Ministry of Science and Technology (Taiwan).
Funding
C.J.J. received financial support partly from China Medical University Hsinchu Hospital (CMUHCH-DMR-109-017 and CMUHCH-DMR-110-016). Y.J.L. received financial support partly from Taiwan Ministry of Science and Technology (107-2221-E-035 -033 -MY3).
Author information
Authors and Affiliations
Corresponding authors
Ethics declarations
Guarantor
The scientific guarantor of this publication is C.J.J.
Conflict of interest
The authors of this manuscript declare no relationships with any companies whose products or services may be related to the subject matter of the article.
Statistics and biometry
One of the authors has significant statistical expertise.
Informed consent
Written informed consent was waived by the institutional review board.
Ethical approval
Institutional review board approval was obtained.
Methodology
• retrospective
• diagnostic study
• multi-center study
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Ruey-Feng Chang and Yi-Jui Liu contribute equally to this work.
Supplementary Information
ESM 1
(DOCX 4342 kb)
Rights and permissions
About this article
Cite this article
Juan, CJ., Lin, SC., Li, YH. et al. Improving interobserver agreement and performance of deep learning models for segmenting acute ischemic stroke by combining DWI with optimized ADC thresholds. Eur Radiol 32, 5371–5381 (2022). https://doi.org/10.1007/s00330-022-08633-6
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00330-022-08633-6