Timothy L. Kline

Algorithm-Enabled Low-Dose Micro-CT Imaging

Micro-computed tomography (micro-CT) is an important tool in biomedical research and preclinical applications that can provide visual inspection of and quantitative information about imaged small animals and biological samples such as vasculature specimens. Currently, micro-CT imaging uses projection data acquired at a large number (300-1000) of views, which can limit system throughput and potentially degrade image quality due to radiation-induced deformation or damage to the small animal or specimen. In this work, we have investigated low-dose micro-CT and its application to specimen imaging from substantially reduced projection data by using a recently developed algorithm, referred to as the adaptive-steepest-descent-projection-onto-convex-sets (ASD-POCS) algorithm, which reconstructs an image through minimizing the image total-variation and enforcing data constraints. To validate and evaluate the performance of the ASD-POCS algorithm, we carried out quantitative evaluation studies in a number of tasks of practical interest in imaging of specimens of real animal organs. The results show that the ASD-POCS algorithm can yield images with quality comparable to that obtained with existing algorithms, while using one-sixth to one quarter of the 361-view data currently used in typical micro-CT specimen imaging.

Machine Learning for Medical Imaging

Machine learning is a technique for recognizing patterns that can be applied to medical images. Although it is a powerful tool that can help in rendering medical diagnoses, it can be misapplied. Machine learning typically begins with the machine learning algorithm system computing the image features that are believed to be of importance in making the prediction or diagnosis of interest. The machine learning algorithm system then identifies the best combination of these image features for classifying the image or computing some metric for the given image region. There are several methods that can be used, each with different strengths and weaknesses. There are open-source versions of most of these machine learning methods that make them easy to try and apply to images. Several metrics for measuring the performance of an algorithm exist; however, one must be aware of the possible associated pitfalls that can result in misleading metrics. More recently, deep learning has started to be used; this method has the benefit that it does not require image feature identification and calculation as a first step; rather, features are identified as part of the learning process. Machine learning has been used in medical imaging and will have a greater influence in the future. Those working in medical imaging must be aware of how machine learning works. ©RSNA, 2017.

Synchrotron‐Based Micro‐CT Imaging of the Human Lung Acinus

Structural data about the human lung fine structure are mainly based on stereological methods applied to serial sections. As these methods utilize 2D images, which are often not contiguous, they suffer from inaccuracies which are overcome by analysis of 3D micro‐CT images of the never‐sectioned specimen. The purpose of our study was to generate a complete data set of the intact three‐dimensional architecture of the human acinus using high‐resolution synchrotron‐based micro‐CT (synMCT). A human lung was inflation‐fixed by formaldehyde ventilation and then scanned in a 64‐slice CT over its apex to base extent. Lung samples (8‐mm diameter, 10‐mm height, N = 12) were punched out, stained with osmium tetroxide, and scanned using synMCT at (4 μm)3 voxel size. The lung functional unit (acinus, N = 8) was segmented from the 3D tomographic image using an automated tree‐analysis software program. Morphometric data of the lung were analyzed by ANOVA. Intra‐acinar airways branching occurred over 11 generations. The mean acinar volume was 131.3 ± 29.2 mm3 (range, 92.5–171.3 mm3) and the mean acinar surface was calculated with 1012 ± 26 cm2. The airway internal diameter (starting from the bronchiolus terminalis) decreases distally from 0.66 ± 0.04 mm to 0.34 ± 0.06 mm (P < 0.001) and remains constant after the seventh generation (P < 0.5). The length of each generation ranges between 0.52 and 0.93 mm and did not show significant differences between the second and eleventh generation. The branching angle between daughter branches varies between 113‐degree and 134‐degree without significant differences between the generations (P < 0.3). This study demonstrates the feasibility of quantitating the 3D structure of the human acinus at the spatial resolution readily achievable using synMCT. Anat Rec 293:1607–1614, 2010.

Automatic total kidney volume measurement on follow-up magnetic resonance images to facilitate monitoring of autosomal dominant polycystic kidney disease progression

BACKGROUND Renal imaging examinations provide high-resolution information about the anatomic structure of the kidneys and are used to measure total kidney volume (TKV) in autosomal dominant polycystic kidney disease (ADPKD) patients. TKV has become the gold-standard image biomarker for ADPKD progression at early stages of the disease and is used in clinical trials to characterize treatment efficacy. Automated methods to segment the kidneys and measure TKV are desirable because of the long time requirement for manual approaches such as stereology or planimetry tracings. However, ADPKD kidney segmentation is complicated by a number of factors, including irregular kidney shapes and variable tissue signal at the kidney borders. METHODS We describe an image processing approach that overcomes these problems by using a baseline segmentation initialization to provide automatic segmentation of follow-up scans obtained years apart. We validated our approach using 20 patients with complete baseline and follow-up T1-weighted magnetic resonance images. Both manual tracing and stereology were used to calculate TKV, with two observers performing manual tracings and one observer performing repeat tracings. Linear correlation and Bland-Altman analysis were performed to compare the different approaches. RESULTS Our automated approach measured TKV at a level of accuracy (mean difference ± standard error = 0.99 ± 0.79%) on par with both intraobserver (0.77 ± 0.46%) and interobserver variability (1.34 ± 0.70%) of manual tracings. All approaches had excellent agreement and compared favorably with ground-truth manual tracing with interobserver, stereological and automated approaches having 95% confidence intervals ∼ ± 100 mL. CONCLUSIONS Our method enables fast, cost-effective and reproducible quantification of ADPKD progression that will facilitate and lower the costs of clinical trials in ADPKD and other disorders requiring accurate, longitudinal kidney quantification. In addition, it will hasten the routine use of TKV as a prognostic biomarker in ADPKD.

Toolkits and Libraries for Deep Learning

Deep learning is an important new area of machine learning which encompasses a wide range of neural network architectures designed to complete various tasks. In the medical imaging domain, example tasks include organ segmentation, lesion detection, and tumor classification. The most popular network architecture for deep learning for images is the convolutional neural network (CNN). Whereas traditional machine learning requires determination and calculation of features from which the algorithm learns, deep learning approaches learn the important features as well as the proper weighting of those features to make predictions for new data. In this paper, we will describe some of the libraries and tools that are available to aid in the construction and efficient execution of deep learning as applied to medical images.

Noninvasive Assessment of Renal Fibrosis with Magnetization Transfer MR Imaging: Validation and Evaluation in Murine Renal Artery Stenosis.

Purpose To test the utility of magnetization transfer imaging in detecting and monitoring the progression of renal fibrosis in mice with unilateral renal artery stenosis. Materials and Methods This prospective study was approved by the Institutional Animal Care and Use Committee. Renal artery stenosis surgery (n = 10) or sham surgery (n = 5) was performed, and the stenotic and contralateral kidneys were studied longitudinally in vivo at baseline and 2, 4, and 6 weeks after surgery. After a 16.4-T magnetic resonance imaging examination, magnetization transfer ratio was measured as an index of fibrosis (guided by parameters selected in preliminary phantom studies). In addition, renal volume, perfusion, blood flow, and oxygenation were assessed. Fibrosis was subsequently measured ex vivo by means of histologic analysis and hydroxyproline assay. The Wilcoxon rank sum or signed rank test was used for statistical comparisons between or within groups, and Pearson and Spearman rank correlation was used to compare fibrosis measured in vivo and ex vivo. Results In the stenotic kidney, the median magnetization transfer ratio showed progressive increases from baseline to 6 weeks after surgery (increases of 13.7% [P = .0006] and 21.3% [P = .0005] in cortex and medulla, respectively), which were accompanied by a progressive loss in renal volume, perfusion, blood flow, and oxygenation. The 6-week magnetization transfer ratio map showed good correlation with fibrosis measured ex vivo (Pearson r = 0.9038 and Spearman ρ = 0.8107 [P = .0002 vs trichrome staining]; r = 0.9540 and ρ = 0.8821 [P < .0001 vs Sirius red staining]; and r = 0.8429 and ρ = 0.7607 [P = .001 vs hydroxyproline assay]). Conclusion Magnetization transfer imaging was used successfully to measure and longitudinally monitor the progression of renal fibrosis in mice with unilateral renal artery stenosis.

Relating function to branching geometry: A micro-CT study of the hepatic artery, portal vein, and biliary tree

Utilizing micro-computed tomography images, the hierarchical structure, interbranch segment lengths and diameters of a hepatic artery, a portal vein, and two biliary trees from intact rat liver lobes were characterized. The data were investigated by analyzing the geometric properties of the vascular structures, such as how interbranch segment diameters change at bifurcation points. In the case of the hepatic artery and portal vein trees (in which the flow rate is high by comparison with that in the biliary tree), the vascular geometry is consistent with a fluid transport system which aims to simultaneously minimize both the power loss of laminar flow, and a cost function proportional to the total volume of material needed to maintain the system (lumenal contents). In comparison, the biliary tree (which has a low flow rate and an opposite flow direction to that of the hepatic artery and portal vein) was found to have a geometry in which the lumen cross-sectional area is maintained at bifurcations. These findings imply that the histological makeup and therefore the pathophysiology of biliary tree vasculature are likely very different from that of the vasculature within the systemic arterial tree. The extent to which the characteristic variability/scatter in the data may have resulted from imaging and/or measurement errors was examined by simulating such errors in a theoretical tree model and comparing the results with the measured data.

Utilizing magnetization transfer imaging to investigate tissue remodeling in a murine model of autosomal dominant polycystic kidney disease.

Noninvasive imaging techniques that quantify renal tissue composition are needed to more accurately ascertain prognosis and monitor disease progression in polycystic kidney disease (PKD). Given the success of magnetization transfer (MT) imaging to characterize various tissue remodeling pathologies, it was tested on a murine model of autosomal dominant PKD.

IVUS Detection of Vasa Vasorum Blood Flow Distribution in Coronary Artery Vessel Wall

There is an increased body of evidence to suggest that the vasa vasorum play a major role in the progression and complications of vulnerable plaque leading to acute coronary syndrome. We propose that detecting changes in the flow in the vascular wall by intravascular ultrasound signals can quantify the presence of vasa vasorum. The results obtained in a porcine model of atherosclerosis suggest that intravascular ultrasound-based estimates of blood flow in the arterial wall can be used in vivo in a clinical research setting to establish the density of vasa vasorum as an indicator of plaque vulnerability.

Semi-automated segmentation of pre-operative low grade gliomas in magnetic resonance imaging.

BackgroundSegmentation of pre-operative low-grade gliomas (LGGs) from magnetic resonance imaging is a crucial step for studying imaging biomarkers. However, segmentation of LGGs is particularly challenging because they rarely enhance after gadolinium administration. Like other gliomas, they have irregular tumor shape, heterogeneous composition, ill-defined tumor boundaries, and limited number of image types. To overcome these challenges we propose a semi-automated segmentation method that relies only on T2-weighted (T2W) and optionally post-contrast T1-weighted (T1W) images.MethodsFirst, the user draws a region-of-interest (ROI) that completely encloses the tumor and some normal tissue. Second, a normal brain atlas and post-contrast T1W images are registered to T2W images. Third, the posterior probability of each pixel/voxel belonging to normal and abnormal tissues is calculated based on information derived from the atlas and ROI. Finally, geodesic active contours use the probability map of the tumor to shrink the ROI until optimal tumor boundaries are found. This method was validated against the true segmentation (TS) of 30 LGG patients for both 2D (1 slice) and 3D. The TS was obtained from manual segmentations of three experts using the Simultaneous Truth and Performance Level Estimation (STAPLE) software. Dice and Jaccard indices and other descriptive statistics were computed for the proposed method, as well as the experts’ segmentation versus the TS. We also tested the method with the BraTS datasets, which supply expert segmentations.Results and discussionFor 2D segmentation vs. TS, the mean Dice index was 0.90 ± 0.06 (standard deviation), sensitivity was 0.92, and specificity was 0.99. For 3D segmentation vs. TS, the mean Dice index was 0.89 ± 0.06, sensitivity was 0.91, and specificity was 0.99. The automated results are comparable with the experts’ manual segmentation results.ConclusionsWe present an accurate, robust, efficient, and reproducible segmentation method for pre-operative LGGs.