Acute ischemic stroke lesion segmentation in non-contrast CT images using 3D convolutional neural networks
In this paper, an automatic algorithm aimed at volumetric segmentation of acute ischemic stroke lesion in non-contrast computed tomography brain 3D images is proposed. Our deep-learning approach is based on the popular 3D U-Net convolutional neural network architecture, which was modified by adding...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Samara National Research University
2023-10-01
|
| Series: | Компьютерная оптика |
| Subjects: | |
| Online Access: | https://www.computeroptics.ru/eng/KO/Annot/KO47-5/470511e.html |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832591902753423360 |
|---|---|
| author | A.V. Dobshik S.K. Verbitskiy I.A. Pestunov K.M. Sherman Yu.N. Sinyavskiy A.A. Tulupov V.B. Berikov |
| author_facet | A.V. Dobshik S.K. Verbitskiy I.A. Pestunov K.M. Sherman Yu.N. Sinyavskiy A.A. Tulupov V.B. Berikov |
| author_sort | A.V. Dobshik |
| collection | DOAJ |
| description | In this paper, an automatic algorithm aimed at volumetric segmentation of acute ischemic stroke lesion in non-contrast computed tomography brain 3D images is proposed. Our deep-learning approach is based on the popular 3D U-Net convolutional neural network architecture, which was modified by adding the squeeze-and-excitation blocks and residual connections. Robust pre-processing methods were implemented to improve the segmentation accuracy. Moreover, a special patches sampling strategy was used to address the large size of medical images and class imbalance and to stabilize neural network training. All experiments were performed using five-fold cross-validation on the dataset containing non-contrast computed tomography volumetric brain scans of 81 patients diagnosed with acute ischemic stroke. Two radiology experts manually segmented images independently and then verified the labeling results for inconsistencies. The quantitative results of the proposed algorithm and obtained segmentation were measured by the Dice similarity coefficient, sensitivity, specificity and precision metrics. The suggested pipeline provides a Dice improvement of 12.0 %, sensitivity of 10.2 % and precision 10.0 % over the baseline and achieves an average Dice of 62.8 ± 3.3 %, sensitivity of 69.9 ± 3.9 %, specificity of 99.7 ± 0.2 % and precision of 61.9 ± 3.6 %, showing promising segmentation results. |
| format | Article |
| id | doaj-art-4838e698a40a4a6f8ab4e37deeb08c32 |
| institution | Kabale University |
| issn | 0134-2452 2412-6179 |
| language | English |
| publishDate | 2023-10-01 |
| publisher | Samara National Research University |
| record_format | Article |
| series | Компьютерная оптика |
| spelling | doaj-art-4838e698a40a4a6f8ab4e37deeb08c322025-01-22T05:21:01ZengSamara National Research UniversityКомпьютерная оптика0134-24522412-61792023-10-0147577077710.18287/2412-6179-CO-1233Acute ischemic stroke lesion segmentation in non-contrast CT images using 3D convolutional neural networksA.V. Dobshik0S.K. Verbitskiy1I.A. Pestunov2K.M. Sherman3Yu.N. Sinyavskiy4A.A. Tulupov5V.B. Berikov6Novosibirsk State UniversityNovosibirsk State UniversityNovosibirsk State University; Federal Research Center for Information and Computational TechnologiesInternational Tomography Center SB RASFederal Research Center for Information and Computational TechnologiesInternational Tomography Center SB RASNovosibirsk State University; Sobolev Institute of Mathematics SB RASIn this paper, an automatic algorithm aimed at volumetric segmentation of acute ischemic stroke lesion in non-contrast computed tomography brain 3D images is proposed. Our deep-learning approach is based on the popular 3D U-Net convolutional neural network architecture, which was modified by adding the squeeze-and-excitation blocks and residual connections. Robust pre-processing methods were implemented to improve the segmentation accuracy. Moreover, a special patches sampling strategy was used to address the large size of medical images and class imbalance and to stabilize neural network training. All experiments were performed using five-fold cross-validation on the dataset containing non-contrast computed tomography volumetric brain scans of 81 patients diagnosed with acute ischemic stroke. Two radiology experts manually segmented images independently and then verified the labeling results for inconsistencies. The quantitative results of the proposed algorithm and obtained segmentation were measured by the Dice similarity coefficient, sensitivity, specificity and precision metrics. The suggested pipeline provides a Dice improvement of 12.0 %, sensitivity of 10.2 % and precision 10.0 % over the baseline and achieves an average Dice of 62.8 ± 3.3 %, sensitivity of 69.9 ± 3.9 %, specificity of 99.7 ± 0.2 % and precision of 61.9 ± 3.6 %, showing promising segmentation results.https://www.computeroptics.ru/eng/KO/Annot/KO47-5/470511e.htmlischemic strokebrainnon-contrast ctsegmentationcnn3d u-net |
| spellingShingle | A.V. Dobshik S.K. Verbitskiy I.A. Pestunov K.M. Sherman Yu.N. Sinyavskiy A.A. Tulupov V.B. Berikov Acute ischemic stroke lesion segmentation in non-contrast CT images using 3D convolutional neural networks Компьютерная оптика ischemic stroke brain non-contrast ct segmentation cnn 3d u-net |
| title | Acute ischemic stroke lesion segmentation in non-contrast CT images using 3D convolutional neural networks |
| title_full | Acute ischemic stroke lesion segmentation in non-contrast CT images using 3D convolutional neural networks |
| title_fullStr | Acute ischemic stroke lesion segmentation in non-contrast CT images using 3D convolutional neural networks |
| title_full_unstemmed | Acute ischemic stroke lesion segmentation in non-contrast CT images using 3D convolutional neural networks |
| title_short | Acute ischemic stroke lesion segmentation in non-contrast CT images using 3D convolutional neural networks |
| title_sort | acute ischemic stroke lesion segmentation in non contrast ct images using 3d convolutional neural networks |
| topic | ischemic stroke brain non-contrast ct segmentation cnn 3d u-net |
| url | https://www.computeroptics.ru/eng/KO/Annot/KO47-5/470511e.html |
| work_keys_str_mv | AT avdobshik acuteischemicstrokelesionsegmentationinnoncontrastctimagesusing3dconvolutionalneuralnetworks AT skverbitskiy acuteischemicstrokelesionsegmentationinnoncontrastctimagesusing3dconvolutionalneuralnetworks AT iapestunov acuteischemicstrokelesionsegmentationinnoncontrastctimagesusing3dconvolutionalneuralnetworks AT kmsherman acuteischemicstrokelesionsegmentationinnoncontrastctimagesusing3dconvolutionalneuralnetworks AT yunsinyavskiy acuteischemicstrokelesionsegmentationinnoncontrastctimagesusing3dconvolutionalneuralnetworks AT aatulupov acuteischemicstrokelesionsegmentationinnoncontrastctimagesusing3dconvolutionalneuralnetworks AT vbberikov acuteischemicstrokelesionsegmentationinnoncontrastctimagesusing3dconvolutionalneuralnetworks |