Software Agent with Reinforcement Learning Approach for Medical Image Segmentation

Chitsaz, Mahsa; Seng, Woo Chaw

doi:10.1007/s11390-011-9431-8

Cited by 15 publications

(6 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Kalman filter estimated the boundary of the segmentation to guide action selection. A similar approach has been employed on cranial CT image segmentation, where the input image was divided into several sub-images, and each RL agent is trained to determine the local thresholding value by dividing the grayscale given the manually input object number Seng 2009, Chitsaz andSeng Woo 2011). In this case, different regions of interest were able to segment simultaneously.…”

Section: Medical Image Segmentationmentioning

confidence: 99%

Deep reinforcement learning and its applications in medical imaging and radiation therapy: a survey

Xu¹,

Zhu²

2022

Phys. Med. Biol.

View full text Add to dashboard Cite

Reinforcement learning takes sequential decision-making approaches by learning the policy through trial and error based on interaction with the environment. Combining deep learning and reinforcement learning can empower the agent to learn the interactions and the distribution of rewards from state-action pairs to achieve effective and efficient solutions in more complex and dynamic environments. Deep reinforcement learning (DRL) has demonstrated astonishing performance in surpassing the human-level performance in the game domain and many other simulated environments. This paper introduces the basics of reinforcement learning and reviews various categories of DRL algorithms and DRL models developed for medical image analysis and radiation treatment planning optimization. We will also discuss the current challenges of DRL and approaches proposed to make DRL more generalizable and robust in a real-world environment. DRL algorithms, by fostering the designs of reward function, agents interactions and environment models, can resolve the challenges from scarce and heterogeneous annotated medical image data, which has been a major obstacle to implementing deep learning models in the clinic. DRL is an active research area with enormous potential to improve deep learning applications in medical imaging and radiation therapy planning.

show abstract

Section: Medical Image Segmentationmentioning

confidence: 99%

Deep reinforcement learning and its applications in medical imaging and radiation therapy: a survey

Xu¹,

Zhu²

2022

Phys. Med. Biol.

View full text Add to dashboard Cite

show abstract

“…RL architectures have also been applied in analyzing medical images obtained from magnetic resonance imaging (MRI), computerized tomography (CT) scan, ultrasound (UlS), etc. However, to the best of our knowledge, the research on medical image segmentation is limited [ 48 , 49 , 50 , 51 , 52 , 53 , 54 ], especially in LV segmentation [ 55 , 56 , 57 ]. In [ 58 ], Mahmud et al reviewed various important applications of deep learning and reinforcement learning to biological data.…”

Section: Related Workmentioning

confidence: 99%

Edge-Sensitive Left Ventricle Segmentation Using Deep Reinforcement Learning

Xiong

Cheung

et al. 2021

Sensors

View full text Add to dashboard Cite

Deep reinforcement learning (DRL) has been utilized in numerous computer vision tasks, such as object detection, autonomous driving, etc. However, relatively few DRL methods have been proposed in the area of image segmentation, particularly in left ventricle segmentation. Reinforcement learning-based methods in earlier works often rely on learning proper thresholds to perform segmentation, and the segmentation results are inaccurate due to the sensitivity of the threshold. To tackle this problem, a novel DRL agent is designed to imitate the human process to perform LV segmentation. For this purpose, we formulate the segmentation problem as a Markov decision process and innovatively optimize it through DRL. The proposed DRL agent consists of two neural networks, i.e., First-P-Net and Next-P-Net. The First-P-Net locates the initial edge point, and the Next-P-Net locates the remaining edge points successively and ultimately obtains a closed segmentation result. The experimental results show that the proposed model has outperformed the previous reinforcement learning methods and achieved comparable performances compared with deep learning baselines on two widely used LV endocardium segmentation datasets, namely Automated Cardiac Diagnosis Challenge (ACDC) 2017 dataset, and Sunnybrook 2009 dataset. Moreover, the proposed model achieves higher F-measure accuracy compared with deep learning methods when training with a very limited number of samples.

show abstract

“…It is the problem faced by an agent that has to learn behavior through trial and error interactions with vigorous environments. And it was applied successfully in many agent systems [42][43][44]. The aim of reinforcement learning is to find out a useful behavior by evaluating the reward.…”

Section: Reinforcement Learning (Rl)mentioning

confidence: 99%

Dynamic Topology Model of Q-Learning LEACH Using Disposable Sensors in Autonomous Things Environment

Cho

Lee

2020

Applied Sciences

View full text Add to dashboard Cite

Low-Energy Adaptive Clustering Hierarchy (LEACH) is a typical routing protocol that effectively reduces transmission energy consumption by forming a hierarchical structure between nodes. LEACH on Wireless Sensor Network (WSN) has been widely studied in the recent decade as one key technique for the Internet of Things (IoT). The main aims of the autonomous things, and one of advanced of IoT, is that it creates a flexible environment that enables movement and communication between objects anytime, anywhere, by saving computing power and utilizing efficient wireless communication capability. However, the existing LEACH method is only based on the model with a static topology, but a case for a disposable sensor is included in an autonomous thing’s environment. With the increase of interest in disposable sensors which constantly change their locations during the operation, dynamic topology changes should be considered in LEACH. This study suggests the probing model for randomly moving nodes, implementing a change in the position of a node depending on the environment, such as strong winds. In addition, as a method to quickly adapt to the change in node location and construct a new topology, we propose Q-learning LEACH based on Q-table reinforcement learning and Fuzzy-LEACH based on Fuzzifier method. Then, we compared the results of the dynamic and static topology model with existing LEACH on the aspects of energy loss, number of alive nodes, and throughput. By comparison, all types of LEACH showed sensitivity results on the dynamic location of each node, while Q-LEACH shows best performance of all.

show abstract

Software Agent with Reinforcement Learning Approach for Medical Image Segmentation

Cited by 15 publications

References 15 publications

Deep reinforcement learning and its applications in medical imaging and radiation therapy: a survey

Deep reinforcement learning and its applications in medical imaging and radiation therapy: a survey

Edge-Sensitive Left Ventricle Segmentation Using Deep Reinforcement Learning

Dynamic Topology Model of Q-Learning LEACH Using Disposable Sensors in Autonomous Things Environment

Contact Info

Product

Resources

About