Agency for Science, Technology and Research

Kent Ridge Digital Labs

Other

Johns Hopkins University

Agency for Science, Technology and Research (a*star)

Singapore Agency for Science, Technology and Research Act

Wieslaw Lucjan Nowinski

Ihar Volkau

Qingmao Hu

Varsha Gupta

Chee-Kong Chui

Yaoping Wang

Ting Ting Ng

Bhanu Prakash K N

Bhanu K N Prakash

Timothy Poston

James H Anderson

Jimin Liu

Yiyu Cai

Su Huang

Choie Cheio Tchoyoson Lim

Guo Liang Yang

Hua Wei

William R Brody

Aamer Aziz

Dzmitry Bialou

Arumugam Thirunavuukarasuu

Yan Xia

Anand Anthanasubramaniam

Bhanu Prakash

Bhanu Prakash Kirgaval Nagaraja Rao

Norman J Beauchamp

Bhanu Prakash Kn

Guoyu Qian

Anand Ananthasubramaniam

Bhanu KN Prakash

Bhanu Prakash KN

K N Bhanu Parkash

A method is proposed for segmenting a brain image into a CSF region, a WM region and a GM region. An upper limit for the intensity values of a CSF region in the image is estimated such that the points of the image having an intensity less than this upper limit include a subset of the points which form a spatially connected group and which have a peaked intensity distribution. In other words, the invention exploits both the expected spatial distribution and expected intensity distribution of the CSF region. This makes it possible for the method to provide reliable discrimination of the CSF region even in CT images with poor image quality. Various methods are proposed for using the upper limit, and for improving the segmentation accuracy.

Method and system for segmenting a brain image

A method () and an apparatus () are disclosed for registering an input image and an atlas. The atlas is warped () using non-rigid registration. In particular, boundary displacements of structures in the atlas are calculated () which register those structures with equivalent structures in the input image. The boundaries of structures in the atlas are then warped () according to their respective boundary displacements using topology preserving propagation of boundary points of the structures.

Method and apparatus for registration of an atlas to an image

A method for identifying a pathological region of a scan (such as a stroke region within a MRI DWI volume scan) is proposed. A region of the scan which is likely to contain pathological tissue (e.g. infracted tissue) is identified by obtaining a parameter which, for a given slice, or portion of a slice, characterizes the distribution of the intensity of pixels, e.g. the relative proportion of high intensity pixels. In a first case, such a parameter is used to identify those slices of a volume scan which are likely to include infarction. In a second case, such a parameter (hemisphere parameter) is obtained for each of the left- and right-hemispheres of a brain, to estimate which hemisphere contains the stroke. In either case, the parameter may be calculated based on ranges, percentiles and functions of the percentiles of the intensity distribution. These ranges, percentiles and functions of the percentiles are not pre-defined but are selected to maximize sensitivity.

Method for identifying a pathological region of a scan, such as an ischemic stroke region of an MRI scan

In a method of registering three-dimensional brain images, a reference slice for a midsagittal plane of each image is constructed. The reference slice comprises image points forming a cortical edge. Edge points are selected from these image points such that an ellipse fit to the edge points approximates the cortical edge. The reference ellipse in each image that fits the edge points is determined. The images are registered in a same coordinate system such that the reference ellipses in the images are aligned with one another.

Registering brain images by aligning reference ellipses

Disclosed herein are methods, computer systems and computer program products for labeling components. One method includes the step of labeling () with a current label all voxels that are internal to a predetermined sub-volume oriented with respect to an unlabeled voxel, and directly connected to the unmarked voxel. The labeling step is repeated for all voxels that are not internal to the predetermined sub-volume, but which are labeled with a current label. The method includes the step of incrementing the current label and may include the step of increasing a window size to a predetermined maximum. The preceding steps are repeated for remaining unlabeled object voxels.

Component labeling

Methods (), apparatuses (), and computer program products for segmenting an infarct in a diffusion-weighted imaging (DWI) volume are disclosed. A Region of Interest in at least one slice of the DWI volume is selected (). The ROI comprises at least a portion of the slice. A threshold for a minimum size of an infarct region is selected (). An energy mask is convolved () with that slice, and the resulting energy image is normalized (). The ROI in the convolved energy image is selected (). An initial threshold is determined () using a histogram of the ROI of the slice without a background region, and an initial segmentation of the slice is performed (). Individual components of the initial segmentation of the slice are labeled (). A final threshold is determined () using histograms of labeled components if the initial segmentation, and a final segmentation if the slice is performed () using that threshold.

Segmenting infarct in diffusion-weighted imaging volume

A method of estimating the location of the anterior and posterior commissures in a brain scan image is proposed. Firstly, a geometrical object is constructed using points on a brain scan image of an individual which are on the surface of the brain, such as an ellipse fitting the cerebral surface of a sagittal image of the mid-sagittal plane, or an adjacent sagittal plane. The locations on the MSP of the AC and PC landmarks (and optionally other landmarks) are estimated using the five parameters which define the ellipse, plus numerical values obtained in advance from statistical analysis of other individuals.

Localization of brain landmarks such as the anterior and posterior commissures based on geometrical fitting

Brain images are processed and analyzed with the aid of a computer for stroke diagnosis or therapeutic decision making, where multiple stroke-related images are superimposed. The superimposed images include brain images that have infarct and penumbra regions, and patient-specific brain atlas images. The infarct and penumbra regions are determined and delineated on the superimposed images. Each patient-specific brain atlas image may be formed by mapping a pre-existing brain atlas to a co-ordinate system in which the brain images are co-registered. A brain atlas may depict brain structures such as anatomy structures, blood supply territories (BST), or cerebral vasculature. The superimposed images may be used to determine any overlap between a particular brain structure and the infarct and penumbra regions.

Superimposing brain atlas images and brain images with delineation of infarct and penumbra for stroke diagnosis

A method and apparatus for identifying pathology in a brain image comprises the steps of firstly determining the location of the midsagittal plane (MSP) of the brain illustrated in the image under examination by identifying the symmetry of the two hemispheres based on the determination of up to 16 approximated fissure line segments (AFLSs). Those AFLSs with a larger angular deviation from the MSP than a predefined threshold are considered as outlier AFLSs while the rest are taken as inlier AFLSs. The ratio of the number of the outlier AFLSs to the number of inlier AFLSs is then calculated. A comparison of the ratio with a further predetermined threshold value is made and if the ratio exceeds the further predetermined threshold value, pathology is present in the brain image.

Method and apparatus for identifying pathology in brain images

Volumes of interest may be defined, within a three-dimensional brain image, for each of three orthogonal directions. Measures, which may, for example, be energy or entropy measures, are determined for slices of the volumes of interest in the three directions. The volume of interest corresponding to the sagittal direction is then identified. The slice, among the slices in the volume of interest corresponding to the identified sagittal direction, having the optical measure is used to define a first estimate of the mid-sagittal plane. The first estimate of the mid-sagittal plane may then be used to build an input to an optimization technique, which operates until a convergence criterion is satisfied, at which point a final estimate of the mid-sagittal plane may be produced. 

Locating a mid-sagittal plane

A method for determining asymmetry in an image such as an MR image of a brain comprises determining a symmetry plane to divide the image into a first part and a second part representative of, for example, the hemispheres of the brain. The probability distributions of voxels against intensities are determined for the first and second parts and histograms of intensities representative of the parts are generated. Compensation is made for any relative shift along a predetermined axis between the histograms. A divergence value based on a distance between the first and second histograms is then calculated and it is determined if the calculated divergence value is greater than a predetermined threshold. A divergence of greater than the predetermined threshold is indicative of asymmetry in the image that may be considered as suspicious for abnormality. There is also disclosed an apparatus for determining asymmetry in an image.

Method and apparatus for determining asymmetry in an image

The AC and/or PC landmarks are identified in a midsagittal MRI image by firstly identifying structures in the brain (specifically the fornix and/or brainstem) as groups of pixels in a radiological image which have an intensity in ranges defined by one or more thresholds and which obey predefined geometrical criteria. The thresholds are varied until the predefined geometrical criteria are met. Initial estimates of the position of the AC and/or PC are derived from the identified structures. These estimates can be improved in various ways, especially by making use of axial and/or corona/radiological images in planes including, and/or proximate to, the initial estimated position of the AC and/or PC. 

Automatic identification of the anterior and posterior commissure landmarks

The present invention discloses a method for registering a measured MRI volume image with appropriate anatomical and blood supply territory Atlases to enable Atlas information to be mapped onto the measured MRI volume image. The disclosed arrangements provide an efficient method for mapping brain Atlas information (including gross anatomy and blood supply territories) into magnetic resonance perfusion and diffusion images.

Method and apparatus for atlas-assisted interpretation of magnetic resonance diffusion and perfusion images

A method and apparatus for extracting a cerebral ventricular system from images of one or more cerebral ventricular regions comprises defining multiple regions of interest (ROI) in the images, defining seed points within each ROI, growing images of ventricular regions while correcting for leakages into extraventricular space and connecting the ventricular regions grown.

Method and apparatus for extracting cerebral ventricular system from images

A method is disclosed for obtaining the location of a landmark in an MR image of a brain. In a first step, a region of interest in a plane within the MR image containing the landmark is defined. In a second step, the ROI is binarised into foreground and background voxels based on at least one threshold selected using anatomical knowledge. In a third step a set of object voxels is identified from the foreground voxels, excluding voxels which were only classified as object due to proximity of cortical and non-cortical structures. This can be done by morphological processing which reclassifies voxels which may have been incorrectly classified as object, followed by restoring voxels due to the partial volume effect and/or morphological erosion/opening. In a fourth step, an automatic process is then carried out to identify one or more landmarks in the modified binarised image.

Growing community of inventors

Singapore, Singapore

Wieslaw Lucjan Nowinski