Extracting chemical-protein relations using attention-based neural networks.

Relation extraction is an important task in the field of natural language processing. In this paper, we describe our approach for the BioCreative VI Task 5: text mining chemical-protein interactions. We investigate multiple deep neural network (DNN) models, including convolutional neural networks, recurrent neural networks (RNNs) and attention-based (ATT-) RNNs (ATT-RNNs) to extract chemical-protein relations. Our experimental results indicate that ATT-RNN models outperform the same models without using attention and the ATT-gated recurrent unit (ATT-GRU) achieves the best performing micro average F1 score of 0.527 on the test set among the tested DNNs. In addition, the result of word-level attention weights also shows that attention mechanism is effective on selecting the most important trigger words when trained with semantic relation labels without the need of semantic parsing and feature engineering. The source code of this work is available at https://github.com/ohnlp/att-chemprot.

Intervertebral disc detection in X-ray images using faster R-CNN.

Automatic identification of specific osseous landmarks on the spinal radiograph can be used to automate calculations for correcting ligament instability and injury, which affect 75% of patients injured in motor vehicle accidents. In this work, we propose to use deep learning based object detection method as the first step towards identifying landmark points in lateral lumbar X-ray images. The significant breakthrough of deep learning technology has made it a prevailing choice for perception based applications, however, the lack of large annotated training dataset has brought challenges to utilizing the technology in medical image processing field. In this work, we propose to fine tune a deep network, Faster-RCNN, a state-of-the-art deep detection network in natural image domain, using small annotated clinical datasets. In the experiment we show that, by using only 81 lateral lumbar X-Ray training images, one can achieve much better performance compared to traditional sliding window detection method on hand crafted features. Furthermore, we fine-tuned the network using 974 training images and tested on 108 images, which achieved average precision of 0.905 with average computation time of 3 second per image, which greatly outperformed traditional methods in terms of accuracy and efficiency.

Correlating Lab Test Results in Clinical Notes with Structured Lab Data: A Case Study in HbA1c and Glucose.

It is widely acknowledged that information extraction of unstructured clinical notes using natural language processing (NLP) and text mining is essential for secondary use of clinical data for clinical research and practice. Lab test results are currently structured in most of the electronic health record (EHR) systems. However, for referral patients or lab tests that can be done in non-clinical setting, the results can be captured in unstructured clinical notes. In this study, we proposed a rule-based information extraction system to extract the lab test results with temporal information from clinical notes. The lab test results of glucose and HbA1c from 104 randomly sampled diabetes patients selected from 1996 to 2015 are extracted and further correlated with structured lab test information in the Mayo Clinic EHRs. The system has high F1-scores of 0.964, 0.967 and 0.966 in glucose, HbA1c and overall extraction, respectively.

An Infinite Mixture Model for Coreference Resolution in Clinical Notes.

It is widely acknowledged that natural language processing is indispensable to process electronic health records (EHRs). However, poor performance in relation detection tasks, such as coreference (linguistic expressions pertaining to the same entity/event) may affect the quality of EHR processing. Hence, there is a critical need to advance the research for relation detection from EHRs. Most of the clinical coreference resolution systems are based on either supervised machine learning or rule-based methods. The need for manually annotated corpus hampers the use of such system in large scale. In this paper, we present an infinite mixture model method using definite sampling to resolve coreferent relations among mentions in clinical notes. A similarity measure function is proposed to determine the coreferent relations. Our system achieved a 0.847 F-measure for i2b2 2011 coreference corpus. This promising results and the unsupervised nature make it possible to apply the system in big-data clinical setting.

A Topic-modeling Based Framework for Drug-drug Interaction Classification from Biomedical Text.

Classification of drug-drug interaction (DDI) from medical literatures is significant in preventing medication-related errors. Most of the existing machine learning approaches are based on supervised learning methods. However, the dynamic nature of drug knowledge, combined with the enormity and rapidly growing of the biomedical literatures make supervised DDI classification methods easily overfit the corpora and may not meet the needs of real-world applications. In this paper, we proposed a relation classification framework based on topic modeling (RelTM) augmented with distant supervision for the task of DDI from biomedical text. The uniqueness of RelTM lies in its two-level sampling from both DDI and drug entities. Through this design, RelTM take both relation features and drug mention features into considerations. An efficient inference algorithm for the model using Gibbs sampling is also proposed. Compared to the previous supervised models, our approach does not require human efforts such as annotation and labeling, which is its advantage in trending big data applications. Meanwhile, the distant supervision combination allows RelTM to incorporate rich existing knowledge resources provided by domain experts. The experimental results on the 2013 DDI challenge corpus reach 48% in F1 score, showing the effectiveness of RelTM.

Fast scale-invariant lateral lumbar vertebrae detection and segmentation in X-ray images.

Fully automatic localization of lumbar vertebrae from clinical X-ray images is very challenging due to the variation of X-ray quality, scale, contrast, number of visible vertebrae, etc. To overcome these challenges, we present a novel framework, where we accelerate a scale-invariant object detection method using Support Vector Machines (SVM) trained on Histogram of Oriented Gradients (HOG) features and segmenting a fine vertebra contour using Gradient Vector Flow (GVF) based snake model. Support Vector Machines trained on HOG features are now an object detection standard in many perception fields and have demonstrated good performance on medical images as well. However, the computational complexity and lack of robustness brought by rescaling the original images have prevented its applicability. The proposed multistage detection framework uses lower-level detection result to determine the rescaling regions to reduce the region of interest, thereby decreasing the execution time. We further refine the detection result by segmenting the contour of vertebra using GVF snake, where we use edge detection techniques to increase the robustness of the GVF snake. Finally, we experimentally demonstrate the effectiveness of this framework using a large set of clinical X-ray images.

Eighth to sixteenth virial coefficients of the Lennard-Jones model.

We calculated virial coefficients BN, 8 ≤ N ≤ 16, of the Lennard-Jones (LJ) model using both the Mayer-sampling Monte Carlo method and direct generation of configurations, with Wheatley's algorithm for summation of clusters. For N = 8, 24 values are reported, and for N = 9, 12 values are reported, both for temperatures T in the range 0.6 ≤ T ≤ 40.0 (in LJ units). For each N in 10 ≤ N ≤ 16, one to four values are reported for 0.6 ≤ T ≤ 0.9. An approximate functional form for the temperature dependence of BN was developed, and fits of LJ BN(T) based on this form are presented for each coefficient, 4 ≤ N ≤ 9, using new and previously reported data.

Automatic spinal canal detection in lumbar MR images in the sagittal view using dynamic programming.

As there is an increasing need for the computer-aided effective management of pathology in lumbar spine, we have developed a computer-aided diagnosis and characterization framework using lumbar spine MRI that provides radiologists a second opinion. In this paper, we propose a left spinal canal boundary extraction method, based on dynamic programming in lumbar spine MRI. Our method fuses the absolute intensity difference of T1-weighted and T2-weighted sagittal images and the inverted gradient of the difference image into a dynamic programming scheme and works in a fully automatic fashion. The boundaries generated by our method are compared against reference boundaries in terms of the Euclidean distance and the Chebyshev distance. The experimental results from 85 clinical data show that our methods find the boundary with a mean Euclidean distance of 3mm, achieving a speedup factor of 167 compared with manual landmark extraction. The proposed method successfully extracts landmarks automatically and fits well with our framework for computer-aided diagnosis in lumbar spine.

Supervised methods for detection and segmentation of tissues in clinical lumbar MRI.

Lower back pain (LBP) is widely prevalent all over the world and more than 80% of the people suffer from LBP at some point of their lives. Moreover, a shortage of radiologists is the most pressing cause for the need of CAD (computer-aided diagnosis) systems. Automatic localization and labeling of intervertebral discs from lumbar MRI is the first step towards computer-aided diagnosis of lower back ailments. Subsequently, for diagnosis and characterization (quantification and localization) of abnormalities like disc herniation and stenosis, a completely automatic segmentation of intervertebral discs and the dural sac is extremely important. Contribution of this paper towards clinical CAD systems is two-fold. First, we propose a method to automatically detect all visible intervertebral discs in clinical sagittal MRI using heuristics and machine learning techniques. We provide a novel end-to-end framework that outputs a tight bounding box for each disc, instead of simply marking the centroid of discs, as has been the trend in the recent past. Second, we propose a method to simultaneously segment all the tissues (vertebrae, intervertebral disc, dural sac and background) in a lumbar sagittal MRI, using an auto-context approach instead of any explicit shape features or models. Past work tackles the lumbar segmentation problem on a tissue/organ basis, and which tend to perform poorly in clinical scans due to high variability in appearance. We, on the other hand, train a series of robust classifiers (random forests) using image features and sparsely sampled context features, which implicitly represent the shape and configuration of the image. Both these methods have been tested on a huge clinical dataset comprising of 212 cases and show very promising results for both disc detection (98% disc localization accuracy and 2.08mm mean deviation) and sagittal MRI segmentation (dice similarity indices of 0.87 and 0.84 for the dural sac and the inter-vertebral disc, respectively).

Compression fracture diagnosis in lumbar: a clinical CAD system.

PURPOSE: Lower back pain affects 80-90 % of all people at some point during their life time, and it is considered as the second most neurological ailment after headache. It is caused by defects in the discs, vertebrae, or the soft tissues. Radiologists perform diagnosis mainly from X-ray radiographs, MRI, or CT depending on the target organ. Vertebra fracture is usually diagnosed from X-ray radiographs or CT depending on the available technology. In this paper, we propose a fully automated Computer-Aided Diagnosis System (CAD) for the diagnosis of vertebra wedge compression fracture from CT images that integrates within the clinical routine. METHODS: We perform vertebrae localization and labeling, segment the vertebrae, and then diagnose each vertebra. We perform labeling and segmentation via coordinated system that consists of an Active Shape Model and a Gradient Vector Flow Active Contours (GVF-Snake). We propose a set of clinically motivated features that distinguish the fractured vertebra. We provide two machine learning solutions that utilize our features including a supervised learner (Neural Networks (NN)) and an unsupervised learner (K-Means). RESULTS: We validate our method on a set of fifty (thirty abnormal) Computed Tomography (CT) cases obtained from our collaborating radiology center. Our diagnosis detection accuracy using NN is 93.2 % on average while we obtained 98 % diagnosis accuracy using K-Means. Our K-Means resulted in a specificity of 87.5 % and sensitivity over 99 %. CONCLUSIONS: We presented a fully automated CAD system that seamlessly integrates within the clinical work flow of the radiologist. Our clinically motivated features resulted in a great performance of both the supervised and unsupervised learners that we utilize to validate our CAD system. Our CAD system results are promising to serve in clinical applications after extensive validation.

Disc herniation diagnosis in MRI using a CAD framework and a two-level classifier.

PURPOSE: Disc herniation in the lumbar spine is a common condition, so an automated method for diagnosis could be helpful in clinical applications. A computer-aided framework for disk herniation diagnosis was developed for use in magnetic resonance imaging (MRI). MATERIALS AND METHOD: A computer-aided diagnosis framework for lumbar spine with a two-level classification scheme for disc herniation diagnosis was developed using heterogeneous classifiers: a perceptron classifier, a least mean square classifier, a support vector machine classifier, and a k-Means classifier. Each classifier makes a diagnosis based on a feature set generated from regions of interest that contain vertebrae, a disc, and the spinal cord. Then, an ensemble classifier makes a final decision using score values of each classifier. We used clinical MR image data from 70 subjects in T1-weighted sagittal view and T2-weighted sagittal view for evaluation of the system. RESULTS: MR images of 70 subjects were processed using the proposed framework resulting in successful detection of disc herniation with 99% accuracy, achieving a speedup factor of 30 in comparison with radiologist's diagnosis. CONCLUSION: The computer-aided framework works well to diagnose herniated discs in MRI scans. We expect the framework can be adapted to effectively diagnose a variety of abnormalities in the lumbar spine.

Automated boundary extraction of the spinal canal in MRI based on dynamic programming.

The spinal cord is the only communication link between the brain and the body. The abnormalities in it can lead to severe pain and sometimes to paralysis. Due to the growing gap between the number of available radiologists and the number of required radiologists, the need for computer-aided diagnosis and characterization is increasing. To ease this gap, we have developed a computer-aided diagnosis and characterization framework in lumbar spine that includes the spinal cord, vertebrae, and intervertebral discs. In this paper, we propose two spinal cord boundary extraction methods that fit into our framework based on dynamic programming in lumbar spine MRI. Our method incorporates the intensity of the image and the gradient of the image into a dynamic programming scheme and works in a fully-automatic fashion. The boundaries generated by our method is compared against reference boundaries in terms of Fréchet distance which is known to be a metric for shape analysis. The experimental results from 65 clinical data show that our method finds the spinal canal boundary correctly achieving a mean Fréchet distance of 13.5 pixels. For almost all data, the extracted boundary falls within the spinal cord. So, it can be used as a landmark when marking background regions and finding regions of interest.

Toward a clinical lumbar CAD: herniation diagnosis.

PURPOSE: A CAD system for lumbar disc degeneration and herniation based on clinical MR images can aid diagnostic decision-making provided the method is robust, efficient, and accurate. MATERIAL AND METHODS: A Bayesian-based classifier with a Gibbs distribution was designed and implemented for diagnosing lumbar disc herniation. Each disc is segmented with a gradient vector flow active contour model (GVF-snake) to extract shape features that feed a classifier. The GVF-snake is automatically initialized with an inner boundary of the disc initiated by a point inside the disc. This point is automatically generated by our previous work on lumbar disc labeling. The classifier operates on clinical T2-SPIR weighted sagittal MRI of the lumbar area. The classifier is applied slice-by-slice to tag herniated discs if they are classified as herniated in any of the 2D slices. This technique detects all visible herniated discs regardless of their location (lateral or central). The gold standard for the ground truth was obtained from collaborating radiologists by analyzing the clinical diagnosis report for each case. RESULTS: An average 92.5% herniation diagnosis accuracy was observed in a cross-validation experiment with 65 clinical cases. The random leave-out experiment runs ten rounds; in each round, 35 cases were used for testing and the remaining 30 cases were used for training. CONCLUSION: An automatic robust disk herniation diagnostic method for clinical lumbar MRI was developed and tested. The method is intended for clinical practice to support reliable decision-making.

Labeling of lumbar discs using both pixel- and object-level features with a two-level probabilistic model.

Backbone anatomical structure detection and labeling is a necessary step for various analysis tasks of the vertebral column. Appearance, shape and geometry measurements are necessary for abnormality detection locally at each disc and vertebrae (such as herniation) as well as globally for the whole spine (such as spinal scoliosis). We propose a two-level probabilistic model for the localization of discs from clinical magnetic resonance imaging (MRI) data that captures both pixel- and object-level features. Using a Gibbs distribution, we model appearance and spatial information at the pixel level, and at the object level, we model the spatial distribution of the discs and the relative distances between them. We use generalized expectation-maximization for optimization, which achieves efficient convergence of disc labels. Our two-level model allows the assumption of conditional independence at the pixel-level to enhance efficiency while maintaining robustness. We use a dataset that contains 105 MRI clinical normal and abnormal cases for the lumbar area. We thoroughly test our model and achieve encouraging results on normal and abnormal cases.

Composite features for automatic diagnosis of intervertebral disc herniation from lumbar MRI.

Lower back pain is widely prevalent in the world today, and the situation is aggravated due to a shortage of radiologists. Intervertebral disc disorders like desiccation, degeneration and herniation are some of the major causes of lower back pain. In this paper, we propose a robust computer-aided herniation diagnosis system for lumbar MRI by first extracting an approximate Region Of Interest (ROI) for each disc and then using a combination of viable features to produce a highly accurate classifier. We describe the extraction of raw, LBP (Local Binary Patterns), Gabor, GLCM (Gray-Level Co-occurrence Matrix), shape, and intensity features from lumbar SPIR T2-weighted MRI and also present a thorough performance comparison of individual and combined features. We perform 5-fold cross validation experiments on 35 cases and report a very high accuracy of 98.29% using a combination of features. Also, combining the desired features and reducing the dimensionality using LDA, we achieve a high sensitivity (true positive rate) of 98.11%.

Segmentation of lumbar vertebrae from clinical CT using active shape models and GVF-snake.

Lumbar area of the vertebral column bears the most load of the human body and thus it is responsible for the major portion of lower back pain from which 80% to 90% of people suffer from during their lifetime. Vertebra related diseases are mainly fracture and are usually diagnosed from X-ray radiographs or CT scans depending on the severity of the problem. In this paper, we propose a fully automated lumbar vertebra segmentation that accurately and robustly produces a smooth contour around each of the vertebrae. This segmentation is very useful in any subsequent CAD system for diagnosis and quantification of vertebrae fractures. It also serves the radiologist during the clinical routine. Our method shows an excellent level of vertebra boundary smootheness that was visually approved by our collaborating radiologist for each vertebra and each case from our fifty cases dataset that includes both normal and abnormal cases.

Automatic segmentation of the spinal cord and the dural sac in lumbar MR images using gradient vector flow field.

A Computer-aided diagnosis (CAD) system aims to facilitate characterization and quantification of abnormalities as well as minimize interpretation errors caused by tedious tasks of image screening and radiologic diagnosis. The system usually consists of segmentation, feature extraction and diagnosis, and segmentation significantly affects the diagnostic performance. In this paper, we propose an automatic segmentation method that extracts the spinal cord and the dural sac from T2-weighted sagittal magnetic resonance (MR) images of lumbar spine without the need of any human intervention. Our method utilizes a gradient vector flow (GVF) field to find the candidate blobs and performs a connected component analysis for the final segmentation. MR Images from fifty two subjects were employed for our experiments and the segmentation results were quantitatively compared against reference segmentation by two medical specialists in terms of a mutual overlap metric. The experimental results showed that, on average, our method achieved a similarity index of 0.7 with a standard deviation of 0.0571 that indicated a substantial agreement. We plan to apply this segmentation method to computer-aided diagnosis of many lumbar-related pathologies.

Computer-aided diagnosis of lumbar disc pathology from clinical lower spine MRI.

PURPOSE: Detection of abnormal discs from clinical T2-weighted MR Images. This aids the radiologist as well as subsequent CAD methods in focusing only on abnormal discs for further diagnosis. Furthermore, it gives a degree of confidence about the abnormality of the intervertebral discs that helps the radiologist in making his decision. MATERIALS AND METHODS: We propose a probabilistic classifier for the detection of abnormality of intervertebral discs. We use three features to label abnormal discs that include appearance, location, and context. We model the abnormal disc appearance with a Gaussian model, the location with a 2D Gaussian model, and the context with a Gaussian model for the distance between abnormal discs. We infer on the middle slice of the T2-weighted MRI volume for each case. These MRI scans are specific for the lumbar area. We obtain our gold standard for the ground truth from our collaborating radiologist group by having the clinical diagnosis report for each case. RESULTS: We achieve over 91% abnormality detection accuracy in a cross-validation experiment with 80 clinical cases. The experiment runs ten rounds; in each round, we randomly leave 30 cases out for testing and we use the other 50 cases for training. CONCLUSION: We achieve high accuracy for detection of abnormal discs using our proposed model that incorporates disc appearance, location, and context. We show the extendability of our proposed model to subsequent diagnosis tasks specific to each intervertebral disc abnormality such as desiccation and herniation.