Cross-Shaped Windows Transformer with Self-supervised Pretraining for Clinically Significant Prostate Cancer Detection in Bi-parametric MRI
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2023
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917615423717376 |
|---|---|
| author | Li, Yuheng Wynne, Jacob Wang, Jing Qiu, Richard L. J. Roper, Justin Pan, Shaoyan Jani, Ashesh B. Liu, Tian Patel, Pretesh R. Mao, Hui Yang, Xiaofeng |
| author_facet | Li, Yuheng Wynne, Jacob Wang, Jing Qiu, Richard L. J. Roper, Justin Pan, Shaoyan Jani, Ashesh B. Liu, Tian Patel, Pretesh R. Mao, Hui Yang, Xiaofeng |
| contents | Biparametric magnetic resonance imaging (bpMRI) has demonstrated promising results in prostate cancer (PCa) detection using convolutional neural networks (CNNs). Recently, transformers have achieved competitive performance compared to CNNs in computer vision. Large scale transformers need abundant annotated data for training, which are difficult to obtain in medical imaging. Self-supervised learning (SSL) utilizes unlabeled data to generate meaningful semantic representations without the need for costly annotations, enhancing model performance on tasks with limited labeled data. We introduce a novel end-to-end Cross-Shaped windows (CSwin) transformer UNet model, CSwin UNet, to detect clinically significant prostate cancer (csPCa) in prostate bi-parametric MR imaging (bpMRI) and demonstrate the effectiveness of our proposed self-supervised pre-training framework. Using a large prostate bpMRI dataset with 1500 patients, we first pretrain CSwin transformer using multi-task self-supervised learning to improve data-efficiency and network generalizability. We then finetune using lesion annotations to perform csPCa detection. Five-fold cross validation shows that self-supervised CSwin UNet achieves 0.888 AUC and 0.545 Average Precision (AP), significantly outperforming four comparable models (Swin UNETR, DynUNet, Attention UNet, UNet). Using a separate bpMRI dataset with 158 patients, we evaluate our method robustness to external hold-out data. Self-supervised CSwin UNet achieves 0.79 AUC and 0.45 AP, still outperforming all other comparable methods and demonstrating good generalization to external data. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2305_00385 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Cross-Shaped Windows Transformer with Self-supervised Pretraining for Clinically Significant Prostate Cancer Detection in Bi-parametric MRI Li, Yuheng Wynne, Jacob Wang, Jing Qiu, Richard L. J. Roper, Justin Pan, Shaoyan Jani, Ashesh B. Liu, Tian Patel, Pretesh R. Mao, Hui Yang, Xiaofeng Image and Video Processing Computer Vision and Pattern Recognition Biparametric magnetic resonance imaging (bpMRI) has demonstrated promising results in prostate cancer (PCa) detection using convolutional neural networks (CNNs). Recently, transformers have achieved competitive performance compared to CNNs in computer vision. Large scale transformers need abundant annotated data for training, which are difficult to obtain in medical imaging. Self-supervised learning (SSL) utilizes unlabeled data to generate meaningful semantic representations without the need for costly annotations, enhancing model performance on tasks with limited labeled data. We introduce a novel end-to-end Cross-Shaped windows (CSwin) transformer UNet model, CSwin UNet, to detect clinically significant prostate cancer (csPCa) in prostate bi-parametric MR imaging (bpMRI) and demonstrate the effectiveness of our proposed self-supervised pre-training framework. Using a large prostate bpMRI dataset with 1500 patients, we first pretrain CSwin transformer using multi-task self-supervised learning to improve data-efficiency and network generalizability. We then finetune using lesion annotations to perform csPCa detection. Five-fold cross validation shows that self-supervised CSwin UNet achieves 0.888 AUC and 0.545 Average Precision (AP), significantly outperforming four comparable models (Swin UNETR, DynUNet, Attention UNet, UNet). Using a separate bpMRI dataset with 158 patients, we evaluate our method robustness to external hold-out data. Self-supervised CSwin UNet achieves 0.79 AUC and 0.45 AP, still outperforming all other comparable methods and demonstrating good generalization to external data. |
| title | Cross-Shaped Windows Transformer with Self-supervised Pretraining for Clinically Significant Prostate Cancer Detection in Bi-parametric MRI |
| topic | Image and Video Processing Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2305.00385 |