Multi-Scale Deep Reinforcement Learning for Real-Time 3D-Landmark Detection in CT Scans

By Florin C. Ghesu, Bogdan Georgescu, Yefeng Zheng, Sasa Grbic, Andreas K. Maier, Joachim Hornegger, Dorin Comaniciu. 2019

299 citations recorded by the source corpus.

Robust and fast detection of anatomical structures is a prerequisite for both diagnostic and interventional medical image analysis. Current solutions for anatomy detection are typically based on machine learning techniques that exploit large annotated image databases in order to learn the appearance of the captured anatomy. These solutions are subject to several limitations, including the use of suboptimal feature engineering techniques and most importantly the use of computationally suboptimal search-schemes for anatomy detection. To address these issues, we propose a method that follows a new paradigm by reformulating the detection problem as a behavior learning task for an artificial agent. We couple the modeling of the anatomy appearance and the object search in a unified behavioral framework, using the capabilities of deep reinforcement learning and multi-scale image analysis. In other words, an artificial agent is trained not only to distinguish the target anatomical object from the rest of the body but also how to find the object by learning and following an optimal navigation path to the target object in the imaged volumetric space. We evaluated our approach on 1487 3D-CT volumes from 532 patients, totaling over 500,000 image slices and show that it significantly outperforms state-of-the-art solutions on detecting several anatomical structures with no failed cases from a clinical acceptance perspective, while also achieving a 20-30 percent higher detection accuracy. Most importantly, we improve the detection-speed of the reference methods by 2-3 orders of magnitude, achieving unmatched real-time performance on large 3D-CT scans.

Subfields: Biomedical Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Reinforcement Learning in Healthcare: A Survey (2019, neutral): “This approach was later extended to exploit multi-scale image representations for large 3D CT scans [65] and consider incomplete data [63] or a nonlinear multi-dimensional parametric space in MRI scans of the brain [51].”
Deep learning in medical image registration: a review (2019, methodology): “[50, 51] Cardiac, HN 2D, 3D MR, CT, US NA [78] Prostate 3D-3D MR Deformable”
Nonlinear Adaptively Learned Optimization for Object Localization in 3D Medical Images (2018, methodology): “The landmark-based method uses the proposed algorithm of [6] to detect 14 landmarks carefully chosen after the box definition.”
Hierarchical Fully Convolutional Network for Joint Atrophy Localization and Alzheimer’s Disease Diagnosis using Structural MRI (2018, background): “In recent years, deep convolutional neural networks (CNNs) are showing increasingly successful applications in various medical image computing tasks [24], [25], [26], [27], [28], [29], [30].”
Deep learning in medical imaging and radiation therapy. (2018, background): “144,145 Later on, Xu et al. 146 further extended this approach by turning the optimal action path searching problem into an image partitioning problem, in which a global action map across the whole image was constructed and learned by a DCNN network to guide the searching action.”
Empowering Things With Intelligence: A Survey of the Progress, Challenges, and Opportunities in Artificial Intelligence of Things (2020, methodology): “For examination, deep learning has been used for medical image understanding such as 3D-landmark detection in CT scans [212] and semantic segmentation [75].”
Deep learning in medical image registration: a survey (2019, background): “ange. They claimed that the learnt similarity metric outperformed MI and its variant MIND [63]. 6 Table 2 Overview of RL in medical image registration References ROI Dimension Modality Transformation [50,51] Cardiac, HN 2D, 3D MR, CT, US NA [78] Prostate 3D-3D MR Deformable [92] Spine, Cardiac 3D-3D”
Artificial Intelligence in Cardiovascular Imaging (2019, methodology): “One of the most recent methods for robust localization of anatomical landmarks relies on multiscale deep reinforcement learning (41,42).”
Automatic View Planning with Multi-scale Deep Reinforcement Learning Agents (2018, methodology): “However, in contrast to [3], we choose the terminating action with the lower Q-value.”
A Gentle Introduction to Deep Learning in Medical Image Processing (2018, background): “[71] and the X-ray transform-invariant landmark detection by Bier et al.”