Automatic segmentation of pathological lung using incremental nonnegative matrix factorization

Accurate lung segmentation from large-size 3-D chest-computed tomography images is crucial for computer-assisted cancer diagnostics. To efficiently segment a 3-D lung, we extract voxel-wise features of spatial image contexts by unsupervised learning with a proposed incremental constrained nonnegative matrix factorization (ICNMF). The method applies smoothness constraints to learn the features, which are more robust to lung tissue inhomogeneities, and thus, help to better segment internal lung pathologies than the known state-of-the-art techniques. Compared to the latter, the ICNMF depends less on the domain expert knowledge and is more easily tuned due to only a few control parameters. Also, the proposed slice-wise incremental learning with due regard for interslice signal dependencies decreases the computational complexity of the NMF-based segmentation and is scalable to very large 3-D lung images. The method is quantitatively validated on simulated realistic lung phantoms that mimic different lung pathologies (seven datasets), in vivo datasets for 17 subjects, and 55 datasets from the Lobe and Lung Analysis 2011 (LOLA11) study. For the in vivo data, the accuracy of our segmentation w.r.t. the ground truth is 0.96 by the Dice similarity coefficient, 9.0 mm by the modified Hausdorff distance, and 0.87% by the absolute lung volume difference, which is significantly better than for the NMF-based segmentation. In spite of not being designed for lungs with severe pathologies and of no agreement between radiologists on the ground truth in such cases, the ICNMF with its total accuracy of 0.965 was ranked fifth among all others in the LOLA11. After excluding the nine too pathological cases from the LOLA11 dataset, the ICNMF accuracy increased to 0.986.

In order to evaluate thoracic disease, lung segmentation is a basic and necessary step. In computed tomography, different methods have been developed for automatic lung segmentation. Deep learning models are considered as one of the most important approaches in this field because it can process a lot of images in a timely manner. However, in order for deep learning work efficiently, it needs significant large dataset with precise manually lung segmentation. In this article, we concentrate on approaches and algorithms for image processing to work with pixels value of images. We are able to remove lung mask with high precision from initial computed tomography images by managing the pixel value array. By evaluating the pixel value of its binary image output with 2-D ground truth images, automatic segmentation is assessed. The result shows that for most sections of thoracic pictures, this procedure worked well and in case if there is any fading in the image, such as bone or soft tissue, the algorithm needed to be enhanced.

Pathological lung segmentation (PLS) is an important, yet challenging, medical image application due to the wide variability of pathological lung appearance and shape. Because PLS is often a pre-requisite for other imaging analytics, methodological simplicity and generality are key factors in usability. Along those lines, we present a bottom-up deep-learning based approach that is expressive enough to handle variations in appearance, while remaining unaffected by any variations in shape. We incorporate the deeply supervised learning framework, but enhance it with a simple, yet effective, progressive multi-path scheme, which more reliably merges outputs from different network stages. The result is a deep model able to produce finer detailed masks, which we call progressive holistically-nested networks (P-HNNs). Using extensive cross-validation, our method is tested on multi-institutional datasets comprising 929 CT scans (848 publicly available), of pathological lungs, reporting mean dice scores of 0.985 and demonstrating significant qualitative and quantitative improvements over state-of-the art approaches.

Accurate lung segmentation in chest radiographs is a challenging problem due to the presence of strong edges at the rib cage and clavicle, the varying appearance in the upper clavicle bone region, too small costophrenic angle and the lack of a consistent anatomical shape among different individuals. In this paper, we propose a hybrid semi-automatic method called Hull-Closed Polygonal Line Method (Hull-CPLM) to detect the boundaries of the lung Region of Interest (ROI). To the best of our knowledge, this is the first attempt at lung segmentation using the Hull-CPLM in chest radiographs. The proposed method has two main steps: 1) an image preprocessing method is constructed to implement the coarse segmentation by using as low as 15% of the manually delineated points as the initial points, 2) a refinement step is used to fine-tune the segmentation results based on the improved principal curve model and the machine learning model at the refinement step. To prove the performance of the proposed method, both the private and public databases were used. The private database is used to select the optimal parameters for the proposed method, where the result showed a good performance with the Dice Similarity Coefficient (DSC) as high as 97.08%. While on the public databases, our proposed algorithm not only surpassed the performance of different hybrid algorithms but also reached superior segmentation results by comparing with state-of-the-art methods.

In this paper, we proposed an incremental kernel non-negative matrix factorization (IKNMF) to reduce the computing scale in hyperspectral unmixing. Kernel non-negative matrix factorization (KNMF) is an extended non-negative matrix factorization (NMF) able to capture nonlinear dependency features in data matrix through kernel functions. In KNMF algorithm, the size of kernel matrices is closely associated with the input data scale. To reduce calculation and storage of large matrices, we extend KNMF by introducing partition matrix theory. The decomposition results of data matrices are derived from smaller scale matrices incrementally. Experiments are conducted on synthetic hyperspectral images with multiple sizes, and the experimental results show that the proposed algorithm have effect in saving calculation and memory resource without degrading the unmixing performance.

Non-negative matrix factorization (NNMF) technique has been widely applicable for dimension reduction. NNMF has shown its usability to solve numerous challenges in areas such as signal processing, image classification, and text mining. Existing matrix factorization methods need a predefined value of rank of a base matrix. Our primary intention in this article is to automatically identify the optimum value of the rank to factorize the matrix. Here we propose an Incremental Cauchy NonNegative Matrix factorization (ICNNMF) which automatically finds the optimum rank and separates the overlapping sounds received from a single channel of a source. Classification is performed using an ensemble of one class fuzzy rule-based (FRB) classifiers. We evaluated the proposed method over the real data set, and the proposed model has shown better accuracy as compared to the traditional classifier. Moreover, the proposed matrix factorization method automatically determines the optimal value of the rank and adapt to new data.

Automated lung segmentation for thoracic CT: Impact on computer-aided diagnosis 1

Heterogeneous intensity distribution inside the prostate gland, significant variations in prostate shape, size, inter dataset contrast variations, and imaging artifacts like shadow regions and speckle in Trans Rectal Ultrasound (TRUS) images challenge computer aided automatic or semi-automatic segmentation of the prostate. In this paper, we propose a supervised learning schema based on random forest for automatic initialization and propagation of statistical shape and appearance model. Parametric representation of the statistical model of shape and appearance is derived from principal component analysis (PCA) of the probability distribution inside the prostate and PCA of the contour landmarks obtained from the training images. Unlike traditional statistical models of shape and intensity priors, the appearance model in this paper is derived from the posterior probabilities obtained from random forest classification. This probabilistic information is then used for the initialization and propagation of the statistical model. The proposed method achieves mean Dice Similarity Coefficient (DSC) value of 0.96±0.01, with a mean segmentation time of 0.67±0.02 seconds when validated with 24 images from 6 datasets with considerable shape, size, and intensity variations, in a leave-one-patient-out validation framework. The model achieves statistically significant t-test p-value<0.0001 in mean DSC and mean mean absolute distance (MAD) values compared to traditional statistical models of shape and intensity priors.

BackgroundMagnetic Resonance Imaging (MRI) has become part of clinical routine for diagnosis of neurodegenerative brain disorders. Acquisitions are performed at multiple centers, using multiple imaging systems and is therefore subjected to intra‐ and inter‐scanner variability. Detailed analysis of brain volumetry differences between MRI systems and scan‐rescan acquisitions can provide useful information to correct for different MRI scanner effects in future multi‐center longitudinal studies.MethodThe study population (n=10) consisted of five cognitively healthy controls and five patients belonging to different stages of the AD continuum. Subjects underwent brain MRI acquisitions on three different MRI systems with harmonized scan parameters: Philips Achieva dStream 1.5T, Philips Ingenia 3T and GE Discovery MR750w 3T. Sequences were harmonized by keeping acquisition parameters similar (Table 1). To test intra‐scanner variability, each participant underwent two subsequent MRI scans (between‐scan repositioning time 3‐5 minutes) per imaging system, repeated on three different MRI systems within two hours to test inter‐scanner variability. A total of 60 MRI scans were acquired for downstream comparative analysis. Brain volumes, including whole brain, (cortical) gray matter, hippocampal volumes, lateral ventricles, frontal, temporal and parietal cortices were computed with icobrain dm (v5.0). Coefficients of variation (CV), absolute volume differences (AVD), Dice similarity coefficients (DSC) and intraclass correlation coefficients (ICC) were computed to assess within‐ and between‐scanner agreement.ResultThe intra‐scanner CVs over all volumes were 0.948% for Discovery MR750w, 1.050% for Achieva and 1.152% for Ingenia. The inter‐scanner CVs were 3.016% for all‐scanner comparisons, 2.044% for Ingenia vs. Achieva, 3.191% for Achieva vs. GE and 3.326% for Ingenia vs. GE. The AVDs and ICCs were in line with the CV values. The intra‐scanner DSCs (mean±SD) over all volumes were 0.917±0.011 for Discovery MR750w, 0.917±0.010 for Achieva and 0.917±0.015 for Ingenia. The inter‐scanner DSCs were 0.898±0.017 for all‐scanner comparisons, 0.912±0.012 for Ingenia vs. Achieva, 0.888±0.013 for Achieva vs. Discovery MR750w and 0.893±0.015 for Ingenia vs. Discovery MR750w (Table 2, 3, Figure 1).ConclusionThe variability results show an excellent agreement (CV&lt;2%) for within‐scanner and a good agreement (CV&lt;5%) for between‐scanner comparisons in terms of clinical relevance, indicating that MRI acquisition harmonization is beneficial for automated MRI volumetry software.

Accurate lung segmentation from high resolution CT images is a challenging task due to various detail tracheal structures, missing boundary segments and complex lung anatomy. One popular method is based on gray-level threshold, however its results are usually rough. A united geometric active contours model based on level set is proposed for lung segmentation in this paper. Particularly, this method combines local boundary information and region statistical-based model synchronously: 1) Boundary term ensures the integrality of lung tissue.2) Region term makes the level set function evolve with global characteristic and independent on initial settings. A penalizing energy term is introduced into the model, which forces the level set function evolving without re-initialization. The method is found to be much more efficient in lung segmentation than other methods that are only based on boundary or region. Results are shown by 3D lung surface reconstruction, which indicates that the method will play an important role in the design of computer-aided diagnostic (CAD) system.

Intraoperative ultrasound (iUS) is increasingly used in pediatric brain tumor surgery. However, accurate tumor tissue identification and localization within the patient with iUS remains challenging. Automatic tumor segmentation may overcome these difficulties by providing quicker, more consistent, and reliable image analysis. This study evaluated the feasibility of automatic 3-dimensional segmentation of pediatric brain tumors using navigated iUS. A retrospective analysis was conducted on 109 navigated iUS acquisitions from 60 pediatric patients who underwent neurosurgical tumor resection. 3-dimensional nnU-Net was trained on 79 preresectional iUS acquisitions with corresponding tumor segmentations. Evaluation was performed on 30 iUS acquisitions against segmentations validated by 2 neurosurgeons. Performance benchmarks were set at a minimum Dice Similarity Coefficient (DSC) of 0.85 and a maximum mean surface distance and 95th percentile Hausdorff distance of 5 mm. The trained algorithm predicted tumor segmentations for all acquisitions in the test set, which met the predefined benchmark values with a median DSC of 0.89 (IQR: 0.13), a median mean surface distance of 1.1 mm (IQR: 1.0 mm), and a median 95th percentile Hausdorff distance of 5.6 mm (IQR: 3.5 mm). The lowest quartile of DSC values consisted of 5 patients, including craniopharyngiomas (n = 3) and high-grade gliomas (n = 2). Oversegmentation of the lateral ventricle as tumor was observed in 2 patients. In 1 high-grade glioma case, edema was misclassified as tumor because of unclear tumor boundaries. This study demonstrates that automatic segmentation of pediatric brain tumors using navigated iUS is feasible and achieved high performance in most cases, with few outliers that showed moderate performance in tumor prediction. Future efforts will focus on real-time segmentation of residual tumor tissue during tumor resection. Ultimately, real-time visualization of automatically segmented tumors in 3 dimensions could improve intraoperative navigation and support neurosurgical decision making.

Separate renal function assessment is important in clinical decision making. The single-photon emission computed tomography is commonly used for the assessment although radioactive, tedious and of high cost. This study aimed to automatically assess the separate renal function using plain CT images and artificial intelligence methods, including deep learning-based automatic segmentation and radiomics modeling. We performed a retrospective study on 281 patients with nephrarctia or hydronephrosis from two centers (Training set: 159 patients from Center I; Test set: 122 patients from Center II). The renal parenchyma and hydronephrosis regions in plain CT images were automatically segmented using deep learning-based U-Net transformers (UNETR). Radiomic features were extracted from the two regions and used to build radiomic signature using the ElasticNet, then further combined with clinical characteristics using multivariable logistic regression to obtain an integrated model. The automatic segmentation was evaluated using the dice similarity coefficient (DSC). The mean DSC of automatic kidney segmentation based on UNETR was 0.894 and 0.881 in the training and test sets. The average time of automatic and manual segmentation was 3.4s/case and 1477.9s/case. The AUC of radiomic signature was 0.778 in the training set and 0.801 in the test set. The AUC of the integrated model was 0.792 and 0.825 in the training and test sets. It is feasible to assess the renal function of each kidney separately using plain CT and AI methods. Our method can minimize the radiation risk, improve the diagnostic efficiency and reduce the costs.

Manual segmentation in lumbar paraspinal muscle studies is time-consuming, subject to variability, and limits comprehensive evaluation. To solve the problems, we aim to develop a deep learning algorithm for the automatic segmentation and multi-parameter quantification of eight paraspinal muscles in lumbar CT. We collected CT images covering the range from the superior endplate of the L1 vertebra to the inferior endplate of the S1 vertebra. The dataset was partitioned into training, validation, and test sets in a ratio of 7:1:2. Six convolutional neural networks were employed to automatically segment the bilateral psoas major, quadratus lumborum, erector spinae, and multifidus across all slices. Model performance was evaluated using the Dice Similarity Coefficient (DSC), Hausdorff Distance (HD), and mean Intersection over Union (mIoU). Muscle parameters including cross-sectional area, muscle volume, fat infiltration, CT density, and paraspinal muscle index were calculated. The Intraclass Correlation Coefficient [ICC(3,1)] was used to assess agreement between model's segmentations and manual ground truth. 100 lumbar CT scans (age, 55.02 ± 16.2; 62 female) were used for model development. The mean DSC across all six models ranged from 0.754 to 0.903. TransUNet demonstrated the best overall (DSC, 0.903; 95% CI: 0.902, 0.913) (HD, 9.934mm; 95% CI: 9.815, 11.902) (mIoU, 0.841; 95% CI: 0.839, 0.850). For individual muscles segmented by TransUNet, the highest DSC was 0.936 (95% CI: 0.934-0.938) for the right multifidus. The mean ICC for all parameters was 0.931. The highest ICC value was 0.999 (p < 0.001). The deep learning tool enables automatic and accurate segmentation of eight lumbar paraspinal muscles. Muscle parameter measurements derived from model's segmentations and manual ground truth exhibit high agreement. Accurate segmentation and quantification of paraspinal muscle parameters facilitates large-scale epidemiological studies of paraspinal muscles and spine-related diseases, overcoming the limitations in manual methods.

A fast and accurate automatic lung segmentation and volumetry method for MR data used in epidemiological studies

Lung field segmentation is a prerequisite for the development of automated computer aided diagnosis of interstitial lung diseases from chest high resolution computed tomography (HRCT) images. The traditional threshold-based or region growing methods tends to fail when dense pathology is present in the lung field. In such cases, it is found that prior shape knowledge increases the efficiency and accuracy of segmentation. The main contribution of this paper is the automated segmentation of lung field from HRCT images. The method is based on active shape model and it is able to segment the lung fields from HRCT images with various pathological regions like honeycomb, cavity, ground glass opacity (GGO), consolidation etc. Multiple atlases of the the lung fields are created using training data of size 100 for both left and right lungs. These atlases are used for the training of active shape model for estimation of lung shape fields. In this paper, after training, the seed selection part is automated in the matching step to ensure minimal human intervention. The result of segmentation is evaluated in terms of the Jaccard index, Dice Similarity Coefficient (DSC) and Modified Hausdorff Distance (MHD) for 80 HRCT slices from a publicly available database. The experimental results proves that the proposed method is reliable for segmentation of pathological lung fields in HRCT images with improved accuracy.