Predicting anaplastic lymphoma kinase rearrangement status in patients with non-small cell lung cancer using a machine learning algorithm that combines clinical features and CT images.

Purpose: To develop an appropriate machine learning model for predicting anaplastic lymphoma kinase (ALK) rearrangement status in non-small cell lung cancer (NSCLC) patients using computed tomography (CT) images and clinical features. Method and materials: This study included 193 patients with NSCLC (154 in the training cohort, 39 in the validation cohort), 68 of whom tested positive for ALK rearrangements and 125 of whom tested negative. From the nonenhanced CT scans, 157 radiomic characteristics were extracted, and 8 clinical features were collected. Five machine learning (ML) models were assessed to find the best classification model for predicting ALK rearrangement status. A radiomic signature was developed using the least absolute shrinkage and selection operator (LASSO) algorithm. The predictive performance of the models based on radiomic features, clinical features, and their combination was assessed by receiver operating characteristic (ROC) curves. Results: The support vector machine (SVM) model had the highest AUC of 0.914 for classification. The clinical features model had an AUC=0.805 (95% CI 0.731-0.877) and an AUC=0.735 (95% CI 0.566-0.863) in the training and validation cohorts, respectively. The CT image-based ML model had an AUC=0.953 (95% CI 0.913-1.0) in the training cohort and an AUC=0.890 (95% CI 0.778-0.971) in the validation cohort. For predicting ALK rearrangement status, the ML model based on CT images and clinical features performed better than the model based on only clinical information or CT images, with an AUC of 0.965 (95% CI 0.826-0.882) in the primary cohort and an AUC of 0.914 (95% CI 0.804-0.893) in the validation cohort. Conclusion: Our findings revealed that ALK rearrangement status could be accurately predicted using an ML-based classification model based on CT images and clinical data.

Combined Diffusion Tensor Imaging and Arterial Spin Labeling as Markers of Early Parkinson's disease.

This study aimed to identify a PD-specific MRI pattern using combined diffusion tensor imaging (DTI) and arterial spin labeling (ASL) to discriminate patients with early PD from healthy subjects and evaluate disease status. Twenty-one early and 22 mid-late PD patients, and 22 healthy, age/gender-matched controls underwent 3-T MRI with apparent diffusion coefficient (ADC), fractional anisotropy (FA), fiber number (FN) and cerebral blood flow (CBF) measurements. We found that compared with healthy subjects, there was a profound reduction in FN passing through the SN in PD. FA in the SN and CBF in the caudate nucleus were inversely correlated with motor dysfunction. A negative correlation was observed between FA in the hippocampus (Hip) and the NMSS-Mood score, whereas CBF in the Hip and the prefrontal cortex(PFC) correlated with declined cognition. Stratified five-fold cross-validation identified FA in the SN(FA-SNAv), CBF in the PFC(CBF-PFCAv) and FA in the parietal white matter(FA-PWMAv), and the combination of these measurements offered relatively high accuracy (AUC 0.975, 90% sensitivity and 100% specificity) in distinguishing those with early PD from healthy subjects. We demonstrate that the decreased FNs through SN in combination with changes in FA-SNAv, CBF-PFCAv and FA-PWMAv values might serve as potential markers of early-stage PD.

[Multiple transmission electron microscopic image stitching based on sift features].

We proposed a new stitching method based on sift features to obtain an enlarged view of transmission electron microscopic (TEM) images with a high resolution. The sift features were extracted from the images, which were then combined with fitted polynomial correction field to correct the images, followed by image alignment based on the sift features. The image seams at the junction were finally removed by Poisson image editing to achieve seamless stitching, which was validated on 60 local glomerular TEM images with an image alignment error of 62.5 to 187.5 nm. Compared with 3 other stitching methods, the proposed method could effectively reduce image deformation and avoid artifacts to facilitate renal biopsy pathological diagnosis.

[Structural analysis based on adaptive window for pulmonary nodule detection].

Radiographic detection of pulmonary nodules based on three-dimensional Hessian matrix is highly sensitive but frequently produces false positive results in areas where blood vessels intersect. We propose a novel approach to pulmonary nodule detection using Hessian matrix-based adaptive window structure analysis, in which the structure coefficients is used to differentiate a voxel that belongs to a nodule or vascular structures, followed by construction of the 3D adaptive window to analyze the local structure characteristics; the nodules were then detected using the discrimination function. The experimental results on pulmonary CT images from 17 patients showed a 100% detection sensitivity for nodules of varying sizes and types, with also significantly reduced false positive results generated by the vessel junctions. This approach provides valuable assistance to follow-up positioning and segmentation of the pulmonary nodules.

DTIPrep: quality control of diffusion-weighted images.

In the last decade, diffusion MRI (dMRI) studies of the human and animal brain have been used to investigate a multitude of pathologies and drug-related effects in neuroscience research. Study after study identifies white matter (WM) degeneration as a crucial biomarker for all these diseases. The tool of choice for studying WM is dMRI. However, dMRI has inherently low signal-to-noise ratio and its acquisition requires a relatively long scan time; in fact, the high loads required occasionally stress scanner hardware past the point of physical failure. As a result, many types of artifacts implicate the quality of diffusion imagery. Using these complex scans containing artifacts without quality control (QC) can result in considerable error and bias in the subsequent analysis, negatively affecting the results of research studies using them. However, dMRI QC remains an under-recognized issue in the dMRI community as there are no user-friendly tools commonly available to comprehensively address the issue of dMRI QC. As a result, current dMRI studies often perform a poor job at dMRI QC. Thorough QC of dMRI will reduce measurement noise and improve reproducibility, and sensitivity in neuroimaging studies; this will allow researchers to more fully exploit the power of the dMRI technique and will ultimately advance neuroscience. Therefore, in this manuscript, we present our open-source software, DTIPrep, as a unified, user friendly platform for thorough QC of dMRI data. These include artifacts caused by eddy-currents, head motion, bed vibration and pulsation, venetian blind artifacts, as well as slice-wise and gradient-wise intensity inconsistencies. This paper summarizes a basic set of features of DTIPrep described earlier and focuses on newly added capabilities related to directional artifacts and bias analysis.

[Monte Carlo simulation and validation of the multi-leaf collimator of Varian 23EX accelerator].

OBJECTIVE: To simulate the multi-leaf collimator of Varian linear accelerator using Monte Carlo method. METHODS: The multi-leaf collimator model was established using the DYNVMLC module of BEAMnrc and validated by comparison of Monte Carlo simulation and actual measurement results. RESULTS: The simulation results were well consistent with the actual measurement results with a bias of less than 3%. CONCLUSION: The multi-leaf collimator of Varian linear accelerator can be successfully modeled using Monte Carlo method for analysis of the impact of the geometric properties of the multi-leaf collimator on the dose distribution.

[Design and realization of an algorithm for capsule endoscope image recognition system].

Discrimination of abnormal images from the numerous wireless capsule endoscope (WCE) video sequence images is laborious and time-consuming, so that a computer-based automatic image recognition system is desired for this task. We propose an algorithm to allow feature extraction from each image channel and decision fusion using multiple BP neural networks. The algorithm was tested and the results demonstrated its high efficiency and accuracy in identification of abnormalities in the WCE images.

DTI registration in atlas based fiber analysis of infantile Krabbe disease.

In recent years, diffusion tensor imaging (DTI) has become the modality of choice to investigate white matter pathology in the developing brain. To study neonate Krabbe disease with DTI, we evaluate the performance of linear and non-linear DTI registration algorithms for atlas based fiber tract analysis. The DTI scans of 10 age-matched neonates with infantile Krabbe disease are mapped into an atlas for the analysis of major fiber tracts - the genu and splenium of the corpus callosum, the internal capsules tracts and the uncinate fasciculi. The neonate atlas is based on 377 healthy control subjects, generated using an unbiased diffeomorphic atlas building method. To evaluate the performance of one linear and seven nonlinear commonly used registration algorithms for DTI we propose the use of two novel evaluation metrics: a regional matching quality criterion incorporating the local tensor orientation similarity, and a fiber property profile based metric using normative correlation. Our experimental results indicate that the whole tensor based registration method within the DTI-ToolKit (DTI-TK) shows the best performance for our application.

FRATS: Functional Regression Analysis of DTI Tract Statistics.

Diffusion tensor imaging (DTI) provides important information on the structure of white matter fiber bundles as well as detailed tissue properties along these fiber bundles in vivo. This paper presents a functional regression framework, called FRATS, for the analysis of multiple diffusion properties along fiber bundle as functions in an infinite dimensional space and their association with a set of covariates of interest, such as age, diagnostic status and gender, in real applications. The functional regression framework consists of four integrated components: the local polynomial kernel method for smoothing multiple diffusion properties along individual fiber bundles, a functional linear model for characterizing the association between fiber bundle diffusion properties and a set of covariates, a global test statistic for testing hypotheses of interest, and a resampling method for approximating the p-value of the global test statistic. The proposed methodology is applied to characterizing the development of five diffusion properties including fractional anisotropy, mean diffusivity, and the three eigenvalues of diffusion tensor along the splenium of the corpus callosum tract and the right internal capsule tract in a clinical study of neurodevelopment. Significant age and gestational age effects on the five diffusion properties were found in both tracts. The resulting analysis pipeline can be used for understanding normal brain development, the neural bases of neuropsychiatric disorders, and the joint effects of environmental and genetic factors on white matter fiber bundles.

Evaluation of DTI Property Maps as Basis of DTI Atlas Building.

Compared to region of interest based DTI analysis, voxel-based analysis gives higher degree of localization and avoids the procedure of manual delineation with the resulting intra and inter-rater variability. One of the major challenges in voxel-wise DTI analysis is to get high quality voxel-level correspondence. For that purpose, current DTI analysis tools are building on nonlinear registration algorithms that deform individual datasets into a template image that is either precomputed or computed as part of the analysis. A variety of matching criteria and deformation schemes have been proposed, but often comparative evaluation is missing. In our opinion, the use of consistent and unbiased measures to evaluate current DTI procedures is of great importance and our work presents two possible measures. Specifically, we propose the evaluation criteria generalization and specificity, originally introduced by the shape modeling community, to evaluate and compare different DTI nonlinear warping results. These measures are of indirect nature and have a population wise view. Both measures incorporate information of the variability of the registration results in the template space via a voxel-wise PCA model. Thus far, we have used these measures to evaluate our own DTI analysis procedure employing fluid-based registration on scalar DTI maps. Generalization and specificity from tensor images in the template space were computed for 8 scalar property maps. We found that for our procedure an intensity-normalized FA feature outperformed the other scalar measurements. Also, using the tensor images rather than the FA maps as a comparison frame seemed to produce more robust results.

Quality Control of Diffusion Weighted Images.

Diffusion Tensor Imaging (DTI) has become an important MRI procedure to investigate the integrity of white matter in brain in vivo. DTI is estimated from a series of acquired Diffusion Weighted Imaging (DWI) volumes. DWI data suffers from inherent low SNR, overall long scanning time of multiple directional encoding with correspondingly large risk to encounter several kinds of artifacts. These artifacts can be too severe for a correct and stable estimation of the diffusion tensor. Thus, a quality control (QC) procedure is absolutely necessary for DTI studies. Currently, routine DTI QC procedures are conducted manually by visually checking the DWI data set in a gradient by gradient and slice by slice way. The results often suffer from low consistence across different data sets, lack of agreement of different experts, and difficulty to judge motion artifacts by qualitative inspection. Additionally considerable manpower is needed for this step due to the large number of images to QC, which is common for group comparison and longitudinal studies, especially with increasing number of diffusion gradient directions. We present a framework for automatic DWI QC. We developed a tool called DTIPrep which pipelines the QC steps with a detailed protocoling and reporting facility. And it is fully open source. This framework/tool has been successfully applied to several DTI studies with several hundred DWIs in our lab as well as collaborating labs in Utah and Iowa. In our studies, the tool provides a crucial piece for robust DTI analysis in brain white matter study.

VOXEL-WISE GROUP ANALYSIS OF DTI.

Diffusion tensor MRI (DTI) is now a widely used modality to investigate the fiber tissues in vivo, especially the white matter in brain. An automatic pipeline is described in this paper to conduct a localized voxel-wise multiple-subject group comparison study of DTI. The pipeline consists of 3 steps: 1) Preprocessing, including image format converting, image quality check, eddy-current and motion artifact correction, skull stripping and tensor image estimation, 2) study-specific unbiased DTI atlas computation via affine followed by fluid nonlinear registration and warping of all individual DTI images into the common atlas space to achieve voxel-wise correspondence, 3) voxelwise statistical analysis via heterogeneous linear regression and wild bootstrap technique for correcting for multiple comparisons. This pipeline was applied to process data from a fitness and aging study and preliminary results are presented. The results show that this fully automatic pipeline is suitable for voxel-wise group DTI analysis.

[Study of 3D visualization methods on PACS workstation].

Based on a discussion on PACS and the way its image workstation obtains scanned sequential images, this paper presented a method of 3D surface construction and visualization on PACS workstation. Guest/Server structure was used between PACS application entities. Image storing and transmission were realized by service classes established by DICOM standards. Relation database was used to arrange the stored sequential images. Image workstation transformed the sequential images obtained from PACS net into volume data field. 3D reconstruction and rendering results were obtained by using surface-rendering and volume-rendering methods, which made the 3D construction results acquire vivid 3D structure details of high fidelity and strong sense of reality. 3 sets of application results were also presented in this paper.

[Medical image elastic registration smoothed by unconstrained optimized thin-plate spline].

Elastic registration of medical image is an important subject in medical image processing. Previous work has concentrated on selecting the corresponding landmarks manually and then using thin-plate spline interpolating to gain the elastic transformation. However, the landmarks extraction is always prone to error, which will influence the registration results. Localizing the landmarks manually is also difficult and time-consuming. We the optimization theory to improve the thin-plate spline interpolation, and based on it, used an automatic method to extract the landmarks. Combining these two steps, we have proposed an automatic, exact and robust registration method and have gained satisfactory registration results.

Elastic registration of medical images through multiquadric method.

OBJECTIVE: To improve the precision and reliability of elastic registration of the medical images and to simplify the registration process. METHODS: Previous study concerning elastic registration mostly focused on manual selection of the landmarks and then use of adequate interpolating for elastic transformation. The landmarks extraction, however, was prone to error that often showed impact on the registration results, besides the difficulty and time consumption of manual identification of the landmarks. On the basis of Multiquadric method that allowed smooth adjustment of the parameters, we utilized a semi-automatic method to extract the landmarks by combining these 2 steps, and proposed a novel registration method. RESULTS: Using this method for medical image elastic registration, rapid and accurate registration between standard and deformed images was achieved. CONCLUSION: The method proposed presently is accurate, convenient and reliable.

A method for three-dimensional surface reconstruction of volume data field.

OBJECTIVE: To explore a novel method of three-dimensional (3D) reconstruction based on vector field smoothing, for the purpose of 3D surface reconstruction of DICOM format volume data sets. METHODS: 3D external surface of three sets of volume data, namely craniocerebral volume data, pelvis volume data, and rat embryo volume data, were respectively extracted by Marching Cubes algorithm using small triangle flakes to approach the original 3D structure surfaces. Vector field smoothing was performed on the extracted 3D surfaces. The reconstructed 3D structures were rendered from different angles of view through arbitrary rotation. RESULTS: High-quality results of 3D surface reconstruction were obtained for each set of volume data, demonstrating fine 3D surface details and high fidelity. CONCLUSION: This method can improve 3D surface reconstruction from DICOM volume data sets, promising high quality, fidelity and reality.

[Registration and 3D rendering of serial tissue section images].

It is an important morphological research method to reconstruct the 3D imaging from serial section tissue images. Registration of serial images is a key step to 3D reconstruction. Firstly, an introduction to the segmentation-counting registration algorithm is presented, which is based on the joint histogram. After thresholding of the two images to be registered, the criterion function is defined as counting in a specific region of the joint histogram, which greatly speeds up the alignment process. Then, the method is used to conduct the serial tissue image matching task, and lies a solid foundation for 3D rendering. Finally, preliminary surface rendering results are presented.

Registration of serial tissue section images.

OBJECTIVE: This study aims to tackle the problem of image registration during computer-assisted three-dimensional (3D) reconstruction of serial tissue sections. METHODS: We proposed segmentation-counting algorithm for computerized image registration on the basis of joint histogram. This approach utilizes thresholding of the 2 images to be registered, and the criterion function is defined as the counting in a specific region of the joint histogram. The registration parameters can be obtained by optimizing the criterion function. RESULTS: In the trial application of this approach in image registration for the serial tissue sections of mouse wse embryos, a more efficient result was achieved. CONCLUSION: The method can rapidly accomplish the image registration task for serial tissue sections with simpler calculation processes.