Automatic Ultrasound Guidance Based on Deep Reinforcement Learning

Jarosik, Piotr; Lewandowski, Marcin

doi:10.1109/ultsym.2019.8926041

Cited by 10 publications

(10 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, in contrast with previous work [12,6,21] which uses simulations or phantoms, our proposed system is trained and evaluated on data from real-world routine ultrasound scanning. Moreover, instead of relying on the exact execution of the probe guidance as in previous work [12,6,21], our system reacts to the actual operator probe movements that are sensed with an IMU. This suggests that the system will perform well in future tests on volunteer subjects.…”

Section: Discussionmentioning

confidence: 99%

“…One study proposes an algorithm that learns to find a view of the adult heart in a grid of pre-acquired ultrasound images [12]. Moreover, learningbased systems have been proposed in which a robotic actuator finds predefined views of simple tissue phantoms [6] or a fetal US phantom [21]. However, a fetus in the mother's womb is a dynamic and highly variant object that can not be well-represented with static simulations or a phantom.…”

Section: Phantoms and Simulated Environmentsmentioning

confidence: 99%

See 1 more Smart Citation

Automatic Probe Movement Guidance for Freehand Obstetric Ultrasound

Droste

Drukker

Papageorghiou

et al. 2020

Lecture Notes in Computer Science

View full text Add to dashboard Cite

We present the first system that provides real-time probe movement guidance for acquiring standard planes in routine freehand obstetric ultrasound scanning. Such a system can contribute to the world-wide deployment of obstetric ultrasound scanning by lowering the required level of operator expertise. The system employs an artificial neural network that receives the ultrasound video signal and the motion signal of an inertial measurement unit (IMU) that is attached to the probe, and predicts a guidance signal. The network termed US-GuideNet predicts either the movement towards the standard plane position (goal prediction), or the next movement that an expert sonographer would perform (action prediction). While existing models for other ultrasound applications are trained with simulations or phantoms, we train our model with real-world ultrasound video and probe motion data from 464 routine clinical scans by 17 accredited sonographers. Evaluations for 3 standard plane types show that the model provides a useful guidance signal with an accuracy of 88.8 % for goal prediction and 90.9 % for action prediction.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Phantoms and Simulated Environmentsmentioning

confidence: 99%

Automatic Probe Movement Guidance for Freehand Obstetric Ultrasound

Droste

Drukker

Papageorghiou

et al. 2020

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…Although many recent approaches have focused on developing smart ultrasound equipment that adds interpretative capabilities to existing systems, Milletari et al [ 85 ] applied reinforcement learning to guide inexperienced users in POCUS to obtain clinically relevant images of the anatomy of interest. Jarosik and Lewandowski [ 86 ] developed a software agent that easily adapts to new conditions and informs the user on how to obtain the optimal settings of the imaging system during the scanning.…”

Section: Improving Workflow Efficiencymentioning

confidence: 99%

Technology trends and applications of deep learning in ultrasonography: image quality enhancement, diagnostic support, and improving workflow efficiency

Kang

Kwon

et al. 2021

Ultrasonography

View full text Add to dashboard Cite

Reviews of the most recent applications of deep learning on ultrasound imaging applications are presented. Architectures of deep learning networks are briefly explained for medical imaging application categories of classification, detection, segmentation, and generation. Ultrasonography applications are then reviewed and summarized for image processing and diagnosis along with some representative study cases of breast, thyroid, heart, kidney, liver, and fetal head. Efforts on workflow enhancement are also reviewed with emphasis on view recognition, scanning guide, image quality assessment, and quantification and measurement. Finally some future prospects are presented on image quality Enhancement, diagnostic support, and improving workflow efficiency, along with remarks on hurdles, benefits, and necessary collaborations.

show abstract

“…However, complete and accurate expert demonstrations can be intractable or expensive to obtain in the clinical US scans. Jarosik et al [19] customized an RL agent to move a virtual probe in a simple and static toy environment, but the real-world probe navigation task is much more complicated and challenging due to the highly variable anatomy among patients. In [20], the researchers used RL to learn cardiac US probe navigation in a simulation environment built with spatially tracked US frames acquired by a sonographer on a grid covering the chest of the patient.…”

Section: Related Workmentioning

confidence: 99%

“…3) State transition under restrictions: If there are no restrictions, the probe pose can be updated according to the selected action as in the previous work [19][20] [21]. Here, we instead consider two requirements for the probe pose in real-world US scans: i) the contact between the probe and the patient surface should be maintained to ensure sufficient acoustic coupling, and ii) the tilt angle of the probe should be limited to ensure the comfort and safety of the patient.…”

Section: A Reinforcement Learning For Probe Navigationmentioning

confidence: 99%

Autonomous Navigation of an Ultrasound Probe Towards Standard Scan Planes with Deep Reinforcement Learning

Li,

Wang,

et al. 2021

Preprint

View full text Add to dashboard Cite

Autonomous ultrasound (US) acquisition is an important yet challenging task, as it involves interpretation of the highly complex and variable images and their spatial relationships. In this work, we propose a deep reinforcement learning framework to autonomously control the 6-D pose of a virtual US probe based on real-time image feedback to navigate towards the standard scan planes under the restrictions in real-world US scans. Furthermore, we propose a confidencebased approach to encode the optimization of image quality in the learning process. We validate our method in a simulation environment built with real-world data collected in the US imaging of the spine. Experimental results demonstrate that our method can perform reproducible US probe navigation towards the standard scan plane with an accuracy of 4.91mm/4.65 • in the intra-patient setting, and accomplish the task in the intra-and inter-patient settings with a success rate of 92% and 46%, respectively. The results also show that the introduction of image quality optimization in our method can effectively improve the navigation performance.

show abstract

Automatic Ultrasound Guidance Based on Deep Reinforcement Learning

Cited by 10 publications

References 5 publications

Automatic Probe Movement Guidance for Freehand Obstetric Ultrasound

Automatic Probe Movement Guidance for Freehand Obstetric Ultrasound

Technology trends and applications of deep learning in ultrasonography: image quality enhancement, diagnostic support, and improving workflow efficiency

Autonomous Navigation of an Ultrasound Probe Towards Standard Scan Planes with Deep Reinforcement Learning

Contact Info

Product

Resources

About