Author Contributions
Conceptualization, O.C.-C., B.P.G.-S. and V.P.; methodology, B.P.G.-S., J.A.A.-D. and O.C.-C.; software, B.P.G.-S., J.A.A.-D. and O.C.-C.; validation, B.P.G.-S., J.A.A.-D. and V.P.; formal analysis, B.P.G.-S. and J.A.A.-D.; investigation, O.C.-C., B.P.G.-S. and J.A.A.-D.; resources, V.P., R.R.-R., C.C.-R. and S.S.; data curation, B.P.G.-S.; writing—original draft preparation, O.C.-C., J.A.A.-D. and V.P.; writing—review and editing, B.P.G.-S., V.P. and J.A.A.-D.; visualization, B.P.G.-S.; supervision, B.P.G.-S., J.A.A.-D., V.P., R.R.-R. and C.C.-R.; project administration, V.P., R.R.-R., C.C.-R. and S.S.; funding acquisition, V.P., R.R.-R. and C.C.-R. All authors have read and agreed to the published version of the manuscript.
Figure 1.
Scheme of the proposed model.
Figure 1.
Scheme of the proposed model.
Figure 2.
Distribution of segmentation masks’ sizes (in pixels) with annotation of the first quartile (Q1), median (Q2), and third quartile (Q3): (a) ISLES 2015, (b) ISLES 2022.
Figure 2.
Distribution of segmentation masks’ sizes (in pixels) with annotation of the first quartile (Q1), median (Q2), and third quartile (Q3): (a) ISLES 2015, (b) ISLES 2022.
Figure 3.
F1-Scores resulting from changing key hyperparameters , . The combination leading to the best results is highlighted in orange. (a) Experiments performed on FLAIR sequences of ISLES 2015. (b) Experiments performed on DWI modality of ISLES 2022.
Figure 3.
F1-Scores resulting from changing key hyperparameters , . The combination leading to the best results is highlighted in orange. (a) Experiments performed on FLAIR sequences of ISLES 2015. (b) Experiments performed on DWI modality of ISLES 2022.
Figure 4.
Learning curves comparison: (a) Proposed model, (b) SGD, (c) W/O A.M., (d) CBAM, (e) Dice Loss, (f) Focal Loss.
Figure 4.
Learning curves comparison: (a) Proposed model, (b) SGD, (c) W/O A.M., (d) CBAM, (e) Dice Loss, (f) Focal Loss.
Figure 5.
Visual comparison of the model’s versions results: Ground truth masks are displayed in the first column (a,g,m,s). Results of Proposed model are given in the second column (b,h,n,t), of Dice Loss model in the third column (c,i,o,u), of Focal Loss model in the fourth column (d,j,p,v), of W/O A.M. model in the fifth column (e,k,q,w), and of CBAM model in the sixth column (f,l,r,x).
Figure 5.
Visual comparison of the model’s versions results: Ground truth masks are displayed in the first column (a,g,m,s). Results of Proposed model are given in the second column (b,h,n,t), of Dice Loss model in the third column (c,i,o,u), of Focal Loss model in the fourth column (d,j,p,v), of W/O A.M. model in the fifth column (e,k,q,w), and of CBAM model in the sixth column (f,l,r,x).
Figure 6.
Violin plot of the proposed model’s results on FLAIR images using the axial view, where the dot localizes the median, and the white line represents the mean: (a) IoU scores by mask’s size category, (b) F1-Scores by mask’s size category.
Figure 6.
Violin plot of the proposed model’s results on FLAIR images using the axial view, where the dot localizes the median, and the white line represents the mean: (a) IoU scores by mask’s size category, (b) F1-Scores by mask’s size category.
Figure 7.
Performance of the proposed model in segmenting small lesions on different MRI modalities using the ISLES 2015 dataset (dot and white line represent the median and the mean values): (a) IoU scores by MRI modality, (b) F1-Scores by MRI modality.
Figure 7.
Performance of the proposed model in segmenting small lesions on different MRI modalities using the ISLES 2015 dataset (dot and white line represent the median and the mean values): (a) IoU scores by MRI modality, (b) F1-Scores by MRI modality.
Figure 8.
Overall performance of the proposed model on different MRI modalities using the ISLES 2015 dataset (dot and white line represent the median and the mean values): (a) IoU scores in the coronal plane, (b) IoU scores in the sagittal plane.
Figure 8.
Overall performance of the proposed model on different MRI modalities using the ISLES 2015 dataset (dot and white line represent the median and the mean values): (a) IoU scores in the coronal plane, (b) IoU scores in the sagittal plane.
Figure 9.
Examples of segmented FLAIR images in the coronal plane by the proposed method (second row) and their corresponding ground truth mask (first row) for mask categories Small (a,e), Medium Down (b,f), Medium Up (c,g), and Large (d,h).
Figure 9.
Examples of segmented FLAIR images in the coronal plane by the proposed method (second row) and their corresponding ground truth mask (first row) for mask categories Small (a,e), Medium Down (b,f), Medium Up (c,g), and Large (d,h).
Figure 10.
Examples of segmented FLAIR images in the sagittal plane by the proposed method (second row) and their corresponding ground truth mask (first row) for mask categories Small (a,e), Medium Down (b,f), Medium Up (c,g), and Large (d,h).
Figure 10.
Examples of segmented FLAIR images in the sagittal plane by the proposed method (second row) and their corresponding ground truth mask (first row) for mask categories Small (a,e), Medium Down (b,f), Medium Up (c,g), and Large (d,h).
Figure 11.
Violin plot of the proposed model’s results on DWI and ADC images using the axial view, where the dot localizes the median, and the white line represents the mean: (a) F1-Scores by mask size category using DWI and configuration A (FL = 0.7, GDL = 0.3), (b) F1-Scores by mask size category using DWI and configuration B (FL = 0.9, GDL = 0.1), (c) F1-Scores by mask size category using ADC and configuration A (FL = 0.7, GDL = 0.3), (d) F1-Scores by mask size category using ADC and configuration B (FL = 0.9, GDL = 0.1).
Figure 11.
Violin plot of the proposed model’s results on DWI and ADC images using the axial view, where the dot localizes the median, and the white line represents the mean: (a) F1-Scores by mask size category using DWI and configuration A (FL = 0.7, GDL = 0.3), (b) F1-Scores by mask size category using DWI and configuration B (FL = 0.9, GDL = 0.1), (c) F1-Scores by mask size category using ADC and configuration A (FL = 0.7, GDL = 0.3), (d) F1-Scores by mask size category using ADC and configuration B (FL = 0.9, GDL = 0.1).
Figure 12.
Violin plot of the non-segmented images’ mask size in pixels. Mean value is marked as a white line.
Figure 12.
Violin plot of the non-segmented images’ mask size in pixels. Mean value is marked as a white line.
Figure 13.
Examples of ground truth mask of DWI images in the axial plane (first row) and the segmentation results by the proposed method using , (second row) and , (third row) for mask categories Small (a,e,i), Medium Down (b,f,j), Medium Up (c,g,k), and Large (d,h,l).
Figure 13.
Examples of ground truth mask of DWI images in the axial plane (first row) and the segmentation results by the proposed method using , (second row) and , (third row) for mask categories Small (a,e,i), Medium Down (b,f,j), Medium Up (c,g,k), and Large (d,h,l).
Table 1.
Number of samples for training, validation, and testing using ISLES 2015 in different planes.
Table 1.
Number of samples for training, validation, and testing using ISLES 2015 in different planes.
Plane | Mask Size | Training | Validation | Testing | Total |
---|
Axial | Small | 190 | 48 | 98 | 336 |
Medium Down | 189 | 48 | 105 | 342 |
Medium Up | 190 | 47 | 102 | 339 |
Large | 190 | 47 | 102 | 339 |
Total | 759 | 190 | 407 | 1356 |
Coronal | Small | 237 | 59 | 127 | 423 |
Medium Down | 239 | 60 | 128 | 427 |
Medium Up | 238 | 60 | 127 | 425 |
Large | 239 | 59 | 128 | 426 |
Total | 953 | 238 | 510 | 1701 |
Sagittal | Small | 142 | 36 | 76 | 254 |
Medium Down | 143 | 36 | 76 | 255 |
Medium Up | 142 | 36 | 76 | 254 |
Large | 143 | 35 | 77 | 255 |
Total | 570 | 143 | 305 | 1018 |
Table 2.
Number of samples for training, validation, and testing using ISLES 2022 in different planes.
Table 2.
Number of samples for training, validation, and testing using ISLES 2022 in different planes.
Plane | Mask Size | Training | Validation | Testing | Total |
---|
Axial | Small | 670 | 168 | 359 | 1197 |
Medium Down | 681 | 170 | 365 | 1216 |
Medium Up | 675 | 169 | 362 | 1206 |
Large | 677 | 169 | 362 | 1208 |
Total | 2703 | 676 | 1448 | 4827 |
Coronal | Small | 1060 | 264 | 568 | 1892 |
Medium Down | 1166 | 292 | 625 | 2083 |
Medium Up | 1126 | 282 | 603 | 2011 |
Large | 1121 | 280 | 601 | 2002 |
Total | 4473 | 1118 | 2397 | 7988 |
Sagittal | Small | 854 | 213 | 458 | 1525 |
Medium Down | 849 | 213 | 455 | 1517 |
Medium Up | 867 | 217 | 464 | 1548 |
Large | 857 | 214 | 459 | 1530 |
Total | 3427 | 857 | 1836 | 6120 |
Table 3.
Features description of the model’s versions used for the ablation test.
Table 3.
Features description of the model’s versions used for the ablation test.
Notation | Attention Module | Loss Function | Optimizer |
---|
Proposed | AM | GDFL | AdamW |
W/O A. M. | NO | GDFL | AdamW |
CBAM | CBAM | GDFL | AdamW |
Dice Loss | AM | Dice loss | AdamW |
Focal Loss | AM | Focal loss | AdamW |
SGD | AM | GDFL | SGD |
Table 4.
Performance comparison of the model’s version used for the ablation experiment in format mean ± standard deviation.
Table 4.
Performance comparison of the model’s version used for the ablation experiment in format mean ± standard deviation.
Metric | Proposed | W/O A. M. | CBAM | Dice Loss | Focal Loss | SGD |
---|
IoU | 0.8596 ± 0.1598 | 0.8239 ± 0.1593 | 0.8553 ± 0.1601 | 0.8258 ± 0.1798 | 0.8300 ± 0.1797 | 0.4004 ± 0.2796 |
F1-Score | 0.9129 ± 0.1362 | 0.8917 ± 0.1364 | 0.9106 ± 0.1345 | 0.8893 ± 0.1570 | 0.8922 ± 0.1549 | 0.5079 ± 0.3244 |
HD | 4.09 ± 7.67 | 5.19 ± 8.27 | 4.24 ± 8.17 | 4.94 ± 8.97 | 3.77 ± 6.59 | 31.24 ± 23.98 |
N. S. | 4.10 ± 2.70 | 4.30 ± 2.65 | 3.20 ± 2.27 | 6.10 ± 4.04 | 5.80 ± 1.78 | 86.10 ± 2.98 |
N. S. Small | 3.80 ± 2.52 | 4.00 ± 2.68 | 2.70 ± 2.28 | 5.80 ± 3.99 | 5.40 ± 1.91 | 70.60 ± 1.50 |
N. S. M. D. | 0.30 ± 0.90 | 0.30 ± 0.64 | 0.40 ± 0.80 | 0.30 ± 0.90 | 0.40 ± 0.92 | 15.50 ± 2.62 |
Table 5.
Performance of the proposed model in different MRI modalities and planes using the ISLES 2015 dataset.
Table 5.
Performance of the proposed model in different MRI modalities and planes using the ISLES 2015 dataset.
Plane | Metric | FLAIR | DWI | T1 | T2 |
---|
Axial | IoU | 0.8596 ± 0.1598 | 0.8524 ± 0.1562 | 0.8355 ± 0.1926 | 0.8441 ± 0.1959 |
F1-Score | 0.9129 ± 0.1362 | 0.9096 ± 0.1290 | 0.8916 ± 0.1783 | 0.8966 ± 0.1809 |
HD | 4.09 ± 7.67 | 4.13 ± 6.94 | 5.12 ± 9.24 | 4.98 ± 8.96 |
N. S. | 4.10 ± 2.70 | 2.70 ± 2.05 | 11.20 ± 6.35 | 10.70 ± 4.05 |
Coronal | IoU | 0.8550 ± 0.1658 | 0.8491 ± 0.1660 | 0.8447 ± 0.1846 | 0.8532 ± 0.1973 |
F1-Score | 0.9094 ± 0.1425 | 0.9058 ± 0.1418 | 0.8995 ± 0.1651 | 0.9017 ± 0.1816 |
HD | 4.74 ± 6.88 | 5.58 ± 9.34 | 5.95 ± 9.67 | 5.20 ± 8.13 |
N. S. | 5.00 ± 2.94 | 5.00 ± 2.16 | 7.67 ± 2.49 | 12.33 ± 2.49 |
Sagittal | IoU | 0.8350 ± 0.2087 | 0.8238 ± 0.1978 | 0.8040 ± 0.2206 | 0.8038 ± 0.2366 |
F1-Score | 0.8885 ± 0.1933 | 0.8847 ± 0.1752 | 0.8667 ± 0.2030 | 0.8623 ± 0.2235 |
HD | 8.20 ± 14.44 | 8.63 ± 14.02 | 10.38 ± 17.49 | 7.60 ± 12.10 |
N. S. | 8.33 ± 0.94 | 4.33 ± 0.47 | 8.33 ± 2.87 | 13.33 ± 4.19 |
Table 6.
Performance of the proposed model in different MRI modalities and planes using the ISLES 2022 dataset.
Table 6.
Performance of the proposed model in different MRI modalities and planes using the ISLES 2022 dataset.
Plane | Metric | DWI | ADC |
---|
Axial | IoU | 0.6961 ± 0.3414 | 0.5505 ± 0.3666 |
F1-Score | 0.7517 ± 0.3470 | 0.6192 ± 0.3884 |
HD | 9.90 ± 17.56 | 16.73 ± 23.36 |
N. S. | 121 ± 9 | 253 ± 13 |
Coronal | IoU | 0.5833 ± 0.4000 | 0.4954 ± 0.3918 |
F1-Score | 0.6318 ± 0.4187 | 0.5535 ± 0.4193 |
HD | 10.73 ± 19.32 | 18.08 ± 26.32 |
N. S. | 188 ± 19 | 364 ± 27 |
Sagittal | IoU | 0.5636 ± 0.3926 | 0.4554 ± 0.3740 |
F1-Score | 0.6176 ± 0.4147 | 0.5211 ± 0.4103 |
HD | 14.02 ± 23.21 | 23.73 ± 32.06 |
N. S. | 133 ± 25 | 318 ± 8 |
Table 7.
Comparison of the proposed method with the reported results of state-of-the-art methods on the ISLES 2015 dataset. The best values are given in bold text.
Table 7.
Comparison of the proposed method with the reported results of state-of-the-art methods on the ISLES 2015 dataset. The best values are given in bold text.
Method | MRI Modality | F1-Score | HD | Accuracy | Precision | Sensitivity | Specificity |
---|
Liu et al. [9] | FLAIR | 0.7178 | 3.36 | - | - | - | - |
FLAIR-DWI | 0.7639 | 3.19 | - | - | - | - |
Zhang et al. [23] | DWI | 0.58 | 38.98 | - | 0.60 | 0.68 | - |
Zhang et al. [24] | DWI | 0.6220 | - | 0.9998 | - | 0.7322 | 0.9997 |
Karthik et al. [17] | Multi | 0.7008 | - | - | - | - | - |
Shah et al. [20] | Multi | 0.7156 | - | - | - | - | - |
Mahmood and Basit [12] | Multi | 0.54 | - | - | 0.67 | 0.5 | - |
Aboudi et al. [38] | Multi | 0.5577 | - | 0.9996 | 0.9977 | - | - |
Aboudi et al. [19] | Multi | 0.7960 | - | 0.9956 | 0.9712 | - | - |
Abdmouleh et al. [18] | FLAIR | 0.8135 | - | 0.9673 | - | 0.8007 | 0.9962 |
DWI | 0.6928 | - | 0.9649 | - | 0.6069 | 0.9967 |
T1 | 0.7000 | - | 0.9651 | - | 0.6301 | 0.9961 |
T2 | 0.7072 | - | 0.965 | - | 0.6443 | 0.9961 |
Kumar et al. [39] | FLAIR | 0.8289 | - | - | - | - | - |
DWI | 0.7029 | - | - | - | - | - |
T1 | 0.7015 | - | - | - | - | - |
T2 | 0.7368 | - | - | - | - | - |
Proposed | FLAIR | 0.9129 | 4.09 | 0.9986 | 0.9152 | 0.9240 | 0.9992 |
DWI | 0.9096 | 4.13 | 0.9985 | 0.9084 | 0.9263 | 0.9991 |
T1 | 0.8916 | 5.12 | 0.9984 | 0.8870 | 0.9088 | 0.999 |
T2 | 0.8966 | 4.98 | 0.9984 | 0.8968 | 0.9092 | 0.9991 |
Table 8.
Comparison of the proposed method with the reported results of state-of-the-art methods on the ISLES 2022 dataset.
Table 8.
Comparison of the proposed method with the reported results of state-of-the-art methods on the ISLES 2022 dataset.
Criterion | Wu et al. [25] | Werdiger et al. [26] | Jeong et al. [27] | Proposed |
---|
F1-Score | 0.8560 | 0.6940 | 0.7869 | 0.7641 | 0.7517 | 0.6192 |
MRI Modality | Multi | Multi | Multi | DWI | DWI | ADC |
Table 9.
Complexity comparison with state-of-the-art models.
Table 9.
Complexity comparison with state-of-the-art models.
Method | Parameters (M) |
---|
Wu et al. [25] | 119.0 |
Werdiger et al. [26] | 22.3 |
Proposed | 7.9 |
Table 10.
Average training time of the proposed model in different datasets and planes.
Table 10.
Average training time of the proposed model in different datasets and planes.
Dataset | GPU Model | Axial | Coronal | Sagittal |
---|
ISLES 2015 | GeForce RTX 3070 | 34 m 32 s | 41 m 08 s | 24 m 55 s |
GeForce RTX 3090 | 16 m 26 s | 20 m 07 s | 11 m 59 s |
ISLES 2022 | GeForce RTX 3090 | 53 m 26 s | 1 h 16 m 37 s | 1 h 16 m 39 s |