Centerline depth world for left atrial appendage orifice localization using reinforcement learning

Left atrial appendage (LAA) occlusion (LAAO) is a minimally invasive implant-based method to prevent cardiovascular stroke in patients with non-valvular atrial fibrillation. Assessing the LAA orifice in preoperative CT angiography plays a crucial role in choosing an appropriate LAAO implant size and a proper C-arm angulation. However, accurate orifice localization is hard because of the high anatomic variation of LAA, and unclear position and orientation of the orifice in available CT views. With the major research focus being on LAA segmentation, the only existing computational method for orifice localization utilized a rule-based decision. Nonetheless, using such a fixed rule may yield high localization error due to the varied anatomy of LAA. While deep learning-based models usually show improvements under such variation, learning an effective localization model is difficult because of the tiny orifice structure compared to the vast search space of CT volume. In this paper, we propose a centerline depth-based reinforcement learning (RL) world for effective orifice localization in a small search space. In our scheme, an RL agent observes the centerline-to-surface distance and navigates through the LAA centerline to localize the orifice. Thus, the search space is significantly reduced facilitating improved localization. The proposed formulation could result in high localization accuracy compared to the expert annotations. Moreover, the localization process takes about 7.3 s which is 18 times more efficient than the existing method. Therefore, this can be a useful aid to physicians during the preprocedural planning of LAAO.

Bone Suppression on Chest Radiographs for Pulmonary Nodule Detection: Comparison between a Generative Adversarial Network and Dual-Energy Subtraction

OBJECTIVE

To compare the effects of bone suppression imaging using deep learning (BSp-DL) based on a generative adversarial network (GAN) and bone subtraction imaging using a dual energy technique (BSt-DE) on radiologists' performance for pulmonary nodule detection on chest radiographs (CXRs).

MATERIALS AND METHODS

A total of 111 adults, including 49 patients with 83 pulmonary nodules, who underwent both CXR using the dual energy technique and chest CT, were enrolled. Using CT as a reference, two independent radiologists evaluated CXR images for the presence or absence of pulmonary nodules in three reading sessions (standard CXR, BSt-DE CXR, and BSp-DL CXR). Person-wise and nodule-wise performances were assessed using receiver-operating characteristic (ROC) and alternative free-response ROC (AFROC) curve analyses, respectively. Subgroup analyses based on nodule size, location, and the presence of overlapping bones were performed.

RESULTS

BSt-DE with an area under the AFROC curve (AUAFROC) of 0.996 and 0.976 for readers 1 and 2, respectively, and BSp-DL with AUAFROC of 0.981 and 0.958, respectively, showed better nodule-wise performance than standard CXR (AUAFROC of 0.907 and 0.808, respectively; p ≤ 0.005). In the person-wise analysis, BSp-DL with an area under the ROC curve (AUROC) of 0.984 and 0.931 for readers 1 and 2, respectively, showed better performance than standard CXR (AUROC of 0.915 and 0.798, respectively; p ≤ 0.011) and comparable performance to BSt-DE (AUROC of 0.988 and 0.974; p ≥ 0.064). BSt-DE and BSp-DL were superior to standard CXR for detecting nodules overlapping with bones (p < 0.017) or in the upper/middle lung zone (p < 0.017). BSt-DE was superior (p < 0.017) to BSp-DL in detecting peripheral and sub-centimeter nodules.

CONCLUSION

BSp-DL (GAN-based bone suppression) showed comparable performance to BSt-DE and can improve radiologists' performance in detecting pulmonary nodules on CXRs. Nevertheless, for better delineation of small and peripheral nodules, further technical improvements are required.

Partial Policy-Based Reinforcement Learning for Anatomical Landmark Localization in 3D Medical Images

Utilizing the idea of long-term cumulative return, reinforcement learning (RL) has shown remarkable performance in various fields. We follow the formulation of landmark localization in 3D medical images as an RL problem. Whereas value-based methods have been widely used to solve RL-based localization problems, we adopt an actor-critic based direct policy search method framed in a temporal difference learning approach. In RL problems with large state and/or action spaces, learning the optimal behavior is challenging and requires many trials. To improve the learning, we introduce a partial policy-based reinforcement learning to enable solving the large problem of localization by learning the optimal policy on smaller partial domains. Independent actors efficiently learn the corresponding partial policies, each utilizing their own independent critic. The proposed policy reconstruction from the partial policies ensures a robust and efficient localization, where the sub-agents uniformly contribute to the state-transitions based on their simple partial policies mapping to binary actions. Experiments with three different localization problems in 3D CT and MR images showed that the proposed reinforcement learning requires a significantly smaller number of trials to learn the optimal behavior compared to the original behavior learning scheme in RL. It also ensures a satisfactory performance when trained on fewer images.

Deep vessel segmentation by learning graphical connectivity

We propose a novel deep learning based system for vessel segmentation. Existing methods using CNNs have mostly relied on local appearances learned on the regular image grid, without consideration of the graphical structure of vessel shape. Effective use of the strong relationship that exists between vessel neighborhoods can help improve the vessel segmentation accuracy. To this end, we incorporate a graph neural network into a unified CNN architecture to jointly exploit both local appearances and global vessel structures. We extensively perform comparative evaluations on four retinal image datasets and a coronary artery X-ray angiography dataset, showing that the proposed method outperforms or is on par with current state-of-the-art methods in terms of the average precision and the area under the receiver operating characteristic curve. Statistical significance on the performance difference between the proposed method and each comparable method is suggested by conducting a paired t-test. In addition, ablation studies support the particular choices of algorithmic detail and hyperparameter values of the proposed method. The proposed architecture is widely applicable since it can be applied to expand any type of CNN-based vessel segmentation method to enhance the performance.

Joint Weakly and Semi-Supervised Deep Learning for Localization and Classification of Masses in Breast Ultrasound Images

We propose a framework for localization and classification of masses in breast ultrasound images. We have experimentally found that training convolutional neural network-based mass detectors with large, weakly annotated datasets presents a non-trivial problem, while overfitting may occur with those trained with small, strongly annotated datasets. To overcome these problems, we use a weakly annotated dataset together with a smaller strongly annotated dataset in a hybrid manner. We propose a systematic weakly and semi-supervised training scenario with appropriate training loss selection. Experimental results show that the proposed method can successfully localize and classify masses with less annotation effort. The results trained with only 10 strongly annotated images along with weakly annotated images were comparable to results trained from 800 strongly annotated images, with the 95% confidence interval (CI) of difference -3%-5%, in terms of the correct localization (CorLoc) measure, which is the ratio of images with intersection over union with ground truth higher than 0.5. With the same number of strongly annotated images, additional weakly annotated images can be incorporated to give a 4.5% point increase in CorLoc, from 80% to 84.50% (with 95% CIs 76%-83.75% and 81%-88%). The effects of different algorithmic details and varied amount of data are presented through ablative analysis.

Fast Adaptive RNN Encoder⁻Decoder for Anomaly Detection in SMD Assembly Machine

Surface Mounted Device (SMD) assembly machine manufactures various products on a flexible manufacturing line. An anomaly detection model that can adapt to the various manufacturing environments very fast is required. In this paper, we proposed a fast adaptive anomaly detection model based on a Recurrent Neural Network (RNN) Encoder–Decoder with operating machine sounds. RNN Encoder–Decoder has a structure very similar to Auto-Encoder (AE), but the former has significantly reduced parameters compared to the latter because of its rolled structure. Thus, the RNN Encoder–Decoder only requires a short training process for fast adaptation. The anomaly detection model decides abnormality based on Euclidean distance between generated sequences and observed sequence from machine sounds. Experimental evaluation was conducted on a set of dataset from the SMD assembly machine. Results showed cutting-edge performance with fast adaptation.

Automatic aortic valve landmark localization in coronary CT angiography using colonial walk

The minimally invasive transcatheter aortic valve implantation (TAVI) is the most prevalent method to treat aortic valve stenosis. For pre-operative surgical planning, contrast-enhanced coronary CT angiography (CCTA) is used as the imaging technique to acquire 3-D measurements of the valve. Accurate localization of the eight aortic valve landmarks in CT images plays a vital role in the TAVI workflow because a small error risks blocking the coronary circulation. In order to examine the valve and mark the landmarks, physicians prefer a view parallel to the hinge plane, instead of using the conventional axial, coronal or sagittal view. However, customizing the view is a difficult and time-consuming task because of unclear aorta pose and different artifacts of CCTA. Therefore, automatic localization of landmarks can serve as a useful guide to the physicians customizing the viewpoint. In this paper, we present an automatic method to localize the aortic valve landmarks using colonial walk, a regression tree-based machine-learning algorithm. For efficient learning from the training set, we propose a two-phase optimized search space learning model in which a representative point inside the valvular area is first learned from the whole CT volume. All eight landmarks are then learned from a smaller area around that point. Experiment with preprocedural CCTA images of TAVI undergoing patients showed that our method is robust under high stenotic variation and notably efficient, as it requires only 12 milliseconds to localize all eight landmarks, as tested on a 3.60 GHz single-core CPU.

Residual Error Based Anomaly Detection Using Auto-Encoder in SMD Machine Sound

Detecting an anomaly or an abnormal situation from given noise is highly useful in an environment where constantly verifying and monitoring a machine is required. As deep learning algorithms are further developed, current studies have focused on this problem. However, there are too many variables to define anomalies, and the human annotation for a large collection of abnormal data labeled at the class-level is very labor-intensive. In this paper, we propose to detect abnormal operation sounds or outliers in a very complex machine along with reducing the data-driven annotation cost. The architecture of the proposed model is based on an auto-encoder, and it uses the residual error, which stands for its reconstruction quality, to identify the anomaly. We assess our model using Surface-Mounted Device (SMD) machine sound, which is very complex, as experimental data, and state-of-the-art performance is successfully achieved for anomaly detection.

A Novel Cascade Classifier for Automatic Microcalcification Detection

In this paper, we present a novel cascaded classification framework for automatic detection of individual and clusters of microcalcifications (μC). Our framework comprises three classification stages: i) a random forest (RF) classifier for simple features capturing the second order local structure of individual μCs, where non-μC pixels in the target mammogram are efficiently eliminated; ii) a more complex discriminative restricted Boltzmann machine (DRBM) classifier for μC candidates determined in the RF stage, which automatically learns the detailed morphology of μC appearances for improved discriminative power; and iii) a detector to detect clusters of μCs from the individual μC detection results, using two different criteria. From the two-stage RF-DRBM classifier, we are able to distinguish μCs using explicitly computed features, as well as learn implicit features that are able to further discriminate between confusing cases. Experimental evaluation is conducted on the original Mammographic Image Analysis Society (MIAS) and mini-MIAS databases, as well as our own Seoul National University Bundang Hospital digital mammographic database. It is shown that the proposed method outperforms comparable methods in terms of receiver operating characteristic (ROC) and precision-recall curves for detection of individual μCs and free-response receiver operating characteristic (FROC) curve for detection of clustered μCs.

Forest Walk Methods for Localizing Body Joints from Single Depth Image

We present multiple random forest methods for human pose estimation from single depth images that can operate in very high frame rate. We introduce four algorithms: random forest walk, greedy forest walk, random forest jumps, and greedy forest jumps. The proposed approaches can accurately infer the 3D positions of body joints without additional information such as temporal prior. A regression forest is trained to estimate the probability distribution to the direction or offset toward the particular joint, relative to the adjacent position. During pose estimation, the new position is chosen from a set of representative directions or offsets. The distribution for next position is found from traversing the regression tree from new position. The continual position sampling through 3D space will eventually produce an expectation of sample positions, which we estimate as the joint position. The experiments show that the accuracy is higher than current state-of-the-art pose estimation methods with additional advantage in computation time.

Structured patch model for a unified automatic and interactive segmentation framework

We present a novel interactive segmentation framework incorporating a priori knowledge learned from training data. The knowledge is learned as a structured patch model (StPM) comprising sets of corresponding local patch priors and their pairwise spatial distribution statistics which represent the local shape and appearance along its boundary and the global shape structure, respectively. When successive user annotations are given, the StPM is appropriately adjusted in the target image and used together with the annotations to guide the segmentation. The StPM reduces the dependency on the placement and quantity of user annotations with little increase in complexity since the time-consuming StPM construction is performed offline. Furthermore, a seamless learning system can be established by directly adding the patch priors and the pairwise statistics of segmentation results to the StPM. The proposed method was evaluated on three datasets, respectively, of 2D chest CT, 3D knee MR, and 3D brain MR. The experimental results demonstrate that within an equal amount of time, the proposed interactive segmentation framework outperforms recent state-of-the-art methods in terms of accuracy, while it requires significantly less computing and editing time to obtain results with comparable accuracy.

NONRIGID VOLUME REGISTRATION USING SECOND-ORDER MRF MODEL

In this paper, we introduce a nonrigid registration method using a Markov Random Field (MRF) energy model with second-order smoothness priors. The registration determines an optimal labeling of the MRF energy model where the label corresponds to a 3D displacement vector. In the MRF energy model, spatial relationships are constructed between nodes using second-order smoothness priors. This model improves limitations of first-order spatial priors which cannot fully incorporate the natural smoothness of deformations. Specifically, the second-order smoothness priors can generate desired smoother displacement vector fields and do not suffer from fronto-parallel effects commonly occurred in first-order priors. The usage of second-order priors in the energy model enables this method to produce more accurate registration results. In the experiments, we will show comparative results using uni- and multi-modal Brain MRI volumes.

Rolled fingerprint construction using MRF-based nonrigid image registration

This paper proposes a new rolled fingerprint construction approach incorporating a state-of-the-art nonrigid image registration method based upon a Markov random field (MRF) energy model. The proposed method finds dense correspondences between images from a rolled fingerprint sequence and warps the entire fingerprint area to synthesize a rolled fingerprint. This method can generate conceptually more accurate rolled fingerprints by preserving the geometric properties of the finger surface as opposed to ink-based rolled impressions and other existing rolled fingerprint construction methods. To verify the accuracy of the proposed method, various comparative experiments were designed to reveal differences among the rolled construction methods. The results show that the proposed method is significantly superior in various aspects compared to previous approaches.

Partition-based extraction of cerebral arteries from CT angiography with emphasis on adaptive tracking

In this paper a method to extract cerebral arteries from computed tomographic angiography (CTA) is proposed. Since CTA shows both bone and vessels, the examination of vessels is a difficult task. In the upper part of the brain, the arteries of main interest are not close to bone and can be well segmented out by thresholding and simple connected-component analysis. However in the lower part the separation is challenging due to the spatial closeness of bone and vessels and their overlapping intensity distributions. In this paper a CTA volume is partitioned into two sub-volumes according to the spatial relationship between bone and vessels. In the lower sub-volume, the concerning arteries are extracted by tracking the center line and detecting the border on each cross-section. The proposed tracking method can be characterized by the adaptive properties to the case of cerebral arteries in CTA. These properties improve the tracking continuity with less user-interaction.

Robust segmentation of cerebral arterial segments by a sequential Monte Carlo method: particle filtering

In this paper a method to extract cerebral arterial segments from CT angiography (CTA) is proposed. The segmentation of cerebral arteries in CTA is a challenging task mainly due to bone contact and vein contamination. The proposed method considers a vessel segment as an ellipse travelling in three-dimensional (3D) space and segments it out by tracking the ellipse in spatial sequence. A particle filter is employed as the main framework for tracking and is equipped with adaptive properties to both bone contact and vein contamination. The proposed tracking method is evaluated by the experiments on both synthetic and actual data. A variety of vessels were synthesized to assess the sensitivity to the axis curvature change, obscure boundaries, and noise. The experimental results showed that the proposed method is also insensitive to parameter settings and requires less user intervention than the conventional vessel tracking methods, which proves its improved robustness.