Article Text

Download PDFPDF

EP153 Optimal view detection for ultrasound-guided supraclavicular block using deep learning approaches
  1. Jo Yumin1,
  2. Lee Dongheon2,
  3. Baek Donghyeon2,
  4. Choi Bo Kyung3,
  5. Aryal Nisan3,
  6. Shin Yong Sup2,
  7. Jung Jinsik2 and
  8. Hong Boohwi2
  1. 1Daejeon, Korea
  2. 2none, Daejeon, Korea
  3. 3MTEG Co., Ltd, Seoul, Korea


Background and Aims Successful ultrasound-guided supraclavicular block (SCB) requires the understanding of sonoanatomy and identification of the optimal view. Segmentation using a convolutional neural network (CNN) is limited in clearly determining the optimal view. The present study aims to develop a computer-aided diagnosis (CADx) system using a CNN that can determine the optimal view for complete SCB in real time.

Methods Ultrasound videos were retrospectively collected from 881 patients to develop the CADx system (600 to the training and validation set and 281 to the test set). The CADx system included classification and segmentation approaches, with Residual neural network (ResNet) and U-Net, respectively, applied as backbone networks. In the classification approach, an ablation study was performed to determine the optimal architecture and improve the performance of the model. In the segmentation approach, a cascade structure, in which U-Net is connected to ResNet, was implemented. The performance of the two approaches was evaluated based on a confusion matrix.

Results Using the classification approach, ResNet34 and gated recurrent units with augmentation showed the highest performance, with average accuracy 0.901, precision 0.613, recall 0.757, f1-score 0.677 and AUROC 0.936. Using the segmentation approach, U-Net combined with ResNet34 and augmentation showed poorer performance than the classification approach.

Abstract EP153 Figure 1

Overview of computer-aided diagnosis systems for determining the optimal view for ultrasound-guided supraclavicular block. (a) Classification approach. (b) Segmentation approach, in which the cascade structure of the segmentation model served as input to the classification model

Abstract EP153 Figure 2

Qualitative results of deep learning approaches for determining optimal views for ultrasound-guided supraclavicular block. The bar at the top-left represents the probability predicted by the convolutional neural network model. TE7, Venue Go, and X-Porte results are pictured in order from top to bottom. (a) Original ultrasound images. (b) Results predicted by the classification approach: gradient-weighted class activation mapping. (c) Results predicted by the segmentation approach

Abstract EP153 Figure 3

Comparative performances of the proposed deep learning approaches: (a, c, e) ROC curves of test sets (a) 1 (X-Porte), (c) 2 (Venue Go), and (e) 3 (TE7). (b, d, f) PR curves of test sets (b) 1 (X-Porte), (d) 2 (Venue Go), and (f) 3 (TE7)

Conclusions The CADx system described in this study showed high performance in determining the optimal view for SCB. This system could be expanded to include many anatomical regions and may have potential to aid clinicians in real-time setting.

  • convolutional neural network
  • computer aided diagnosis
  • deep learning
  • optimal view detection
  • regional anesthesia
  • supraclavicular block
  • ultrasound anatomy
  • UGRA

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.