Autonomous Tissue Retraction in Robotic Assisted Minimally Invasive Surgery – A Feasibility Study

Attanasio, Aleks; Scaglioni, Bruno; Leonetti, Matteo; Frangi, Alejandro F.; Cross, William; Biyani, Chandra Shekhar; Valdastri, Pietro

doi:10.1109/lra.2020.3013914

Cited by 50 publications

(39 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to train the networks, DMs must be associated with labels highlighting the areas of the image covered by tissue flaps and by the surgical tools. In a previous work [2], our group developed FlapNet: a dataset of 1080 DMs extracted from images collected during a robotic surgery course, performed with a DaVinci Xi at the University of Leeds, on Thiel-embalmed cadavers [35] by experienced surgeons. Starting from the full stereo video stream of a lobectomy, the most relevant frames of the stream are extracted and labelled: for each DM, a binary mask is created, classifying each pixel as background (0) or tissue (1).…”

Section: A Data Setupmentioning

confidence: 99%

“…As show in Figure 4, the network comprises two symmetric encoding and decoding branches, with parallel connections linking the encoders to the decoders. The standard U-Net architecture is suitable for segmenting single images in endoscopic scenarios, as demonstrated by our previous work [2], but cannot correlate consecutive frames (e.g. a video stream) and therefore has limited robustness.…”

Section: B Neural Network Developmentmentioning

confidence: 99%

“…In previous work [2], we proposed a feasibility study on autonomous tissue retraction, developed on a DaVinci Research Kit (DVRK). To detect a candidate flap of tissue for the retraction, a single endoscopic Depth Map was segmented with deep-learning techniques, and the system was autonomously executing the retraction, based on the analysis of the image.…”

Section: Introductionmentioning

confidence: 99%

“…The performances of these networks are compared to our previous work, the process of parameters optimisation is discussed in detail and the effectiveness of a pre-training stage is evaluated. Additionally, a dataset, based on the FlapNet [2], is developed to train and verify the performances of the networks. The dataset, comprising labels and training images, and the code are publicly available for the research community at https://github.com/ Stormlabuk/dvrk ULSTM.…”

Section: Introductionmentioning

confidence: 99%

“…Then, in Section III the performances of the three architectures are discussed. Additionally, a comparison with a pre-trained model [10] and our previous work [2] is carried out, to demonstrate the benefits in adopting LSTM layers and Attention Gate blocks in video segmentation. Section IV concludes the paper, summarising the contribution and discussing future developments.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A Comparative Study of Spatio-Temporal U-Nets for Tissue Segmentation in Surgical Robotics

Attanasio

Alberti

Scaglioni

et al. 2021

IEEE Trans. Med. Robot. Bionics

Self Cite

View full text Add to dashboard Cite

In surgical robotics, the ability to achieve high levels of autonomy is often limited by the complexity of the surgical scene. Autonomous interaction with soft tissues requires machines able to examine and understand the endoscopic video streams in real-time and identify the features of interest. In this work, we show the first example of spatio-temporal neural networks, based on the U-Net, aimed at segmenting soft tissues in endoscopic images. The networks, equipped with Long Short-Term Memory and Attention Gate cells, can extract the correlation between consecutive frames in an endoscopic video stream, thus enhancing the segmentation's accuracy with respect to the standard U-Net. Initially, three configurations of the spatiotemporal layers are compared to select the best architecture. Afterwards, the parameters of the network are optimised and finally the results are compared with the standard U-Net. An accuracy of 83.77% ± 2.18% and a precision of 78.42% ± 7.38% are achieved by implementing both Long Short Term Memory (LSTM) convolutional layers and Attention Gate blocks. The results, although originated in the context of surgical tissue retraction, could benefit many autonomous tasks such as ablation, suturing and debridement.

show abstract

Section: A Data Setupmentioning

confidence: 99%

Section: B Neural Network Developmentmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations