Glynn P. Robinson

Medical image collection indexing: Shape-based retrieval using KD-trees

The capacity to retrieve images containing objects with shapes similar to a query shape is desirable in medical image databases. We propose a similarity measure and an indexing mechanism for non-rigid comparison of shape which adds this capability to image databases. The (dis-)similarity measure is based on the observations that: (1) the geometry of the same organ in different subjects is not related by a strictly rigid transformation; and (2) the orientation of the organ plays a key role in comparing shape. We propose a similarity measure that computes a non-rigid mapping between curves and uses this mapping to compare oriented shape. We also show how KD-trees can index curves so that retrieval with our similarity measure is efficient. Experiments with real-world data from a database of magnetic resonance images are provided.

Accurate combination of ct and mr data of the head: Validation and applications in surgical and therapy planning

A method is presented for the accurate combination of magnetic resonance (MR) and computed tomographic (CT) images of the head. Our technique is based on user identified 3D landmarks followed by data combination and display as adjacent slices, a single fused slice representation, colour overlay and three-dimensional (3D) rendered scenes. Validation with a point phantom and computer simulation has established the relationship of registration accuracy with point location accuracy, the number of points used and their spatial distribution. The technique is in clinical use in the planning of skull base surgery, transferring MR images acquired without a stereotaxic frame to stereotaxic space, and staging and planning therapy of nasopharyngeal tumours.

Computer-aided interpretation of SPECT images of the brain using an MRI-derived 3D neuro-anatomical atlas

Nuclear medicine images have comparatively poor spatial resolution, making it difficult to relate the functional information which they contain to precise anatomical structures. A 3D neuro-anatomical atlas has been generated from the MRI data set of a normal, healthy volunteer to assist in the interpretation of nuclear medicine scans of the brain. Region growing and edge-detection techniques were used to semi-automatically segment the data set into the major tissue types within the brain. The atlas was then labelled interactively by marking points on each 2D slice. Anatomical structures useful in the interpretation of SPECT images were labelled. Additional, more detailed information corresponding to these structures is provided via an interactive index which allows access to images, diagrams and explanations. Registration of patient SPECT studies with the atlas is accomplished by using the position of the skull vertex and four external fiducial markers attached to the skin surface. The 3D coordinates determined from these points are used to calculate the transformation required to rotate, scale and translate the SPECT data, in 3D, to match the atlas. Corresponding 2D slices from the two 3D data sets are then displayed side-by-side on a computer screen. A cursor linking the two images allows the delineation of regions of interest (ROIs) in the SPECT scan based on anatomical structures identified from the atlas. Conversely regions of abnormal isotope distribution in the SPECT image can be localized by reference to corresponding structures in the atlas.

Preliminary Work on the Interpretation of SPECT Images with the Aid of Registered MR Images and an MR Derived 3D Neuro-anatomical Atlas

This paper describes two methods to aid interpretation and quantification of SPECT or PET images. In the first method 3D SPECT or PET data sets are aligned and scaled to a 3D MRI data set of the same patient using 4 skin markers visible on each modality. Three display schemes have been implemented for viewing the aligned slices. Examples of these displays are provided. The second method uses a labelled 3D MRI reference data set from a volunteer to identify major anatomical structures. The MR reference data set is aligned with the isotope image using the same 4 markers plus a marker on the vertex of the skull. The reference data set is segmented approximately into the major tissue types - cerebrospinal fluid (CSF) and grey and white matter. Major structures are identified via labels in the 3D data set. A linked cursor aids delineation of anatomical regions on the isotope image using the outline of structures on the reference data set as a template. Directions for future research in the generation of complete digital anatomical atlases, which include inter-individual variations, are outlined.

A model-based integrated approach to track myocardial deformation using displacement and velocity constraints

Accurate estimation of heart wall dense field motion and deformation could help to better understand the physiological processes associated with ischemic heart diseases, and to provide significant improvement in patient treatment. We present a new method of estimating left ventricular deformation which integrates instantaneous velocity information obtained within the mid-wall region with shape information found on the boundaries of the left ventricle. Velocity information is obtained from phase contrast magnetic resonance images, and boundary information is obtained from shape-based motion tracking of the endo- and cardial boundaries. The integration takes place within a continuum biomechanical heart model which is embedded in a finite element framework. We also employ a feedback mechanism to improve tracking accuracy. The integration of the two disparate but complementary sources overcomes some of the limitations of previous work in the field which concentrates on motion estimation from a single image-derived source.<<ETX>>

Myocardial motion and function assessment using 4D images

This paper describes efforts aimed at more objectively and accurately quantifying the local, regional and global function of the left ventricle (LV) of the heart from 4D image data. Using our shape-based image analysis methods, point-wise myocardial motion vector fields between successive image frames through the entire cardiac cycle will be computed. Quantitative LV motion, thickening, and strain measurements will then be established from the point correspondence maps. In the paper, we will also briefly describe an in vivo experimental model which uses implanted imaging-opaque markers to validate the results of our image analysis methods. Finally, initial experimental results using image sequences from two different modalities will be presented.

Model-based recognition of anatomical objects from medical images

Abstract We present both a high-level symbolic model of the human brain, and a method of using this model to aid in the recognition of objects from medical images. The model is stored as a frame-based semantic network consisting of three coexisting graphs (a spatial adjacency graph, a part hierarchy and an inheritance graph). We propose a method similar to assumption-based truth maintenance systems for the collating and reasoning processes required in the labelling of input images.

A Unified Framework to Assess Myocardial Function from 4D Images

This paper describes efforts aimed at developing a unified framework to more accurately quantify the local, regional and global function of the left ventricle (LV) of the heart, under both normal and ischemic conditions, using four—dimensional (4D) imaging data over the entire cardiac cycle. The approach incorporates motion information derived from the shape properties of the endocardial and epicardial surfaces of the LV, as well as mid—wall 3D instantaneous velocity information from phase contrast MR images, and/or mid—wall displacement information from tagged MR images. The integration of the disparate but complementary sources of information overcomes the limitations of previous work which concentrates on motion estimation from a single image—derived source. 1

Toward reliable, noninvasive measurement of myocardial function from 4D images

This paper describes efforts aimed at more accurately and objectively determining and quantifying the local, regional, and global function of the left ventricle (LV) of the heart under both normal and ischemic conditions. These measurements and evaluations are made using non-invasive, 3-D, cardiac diagnostic imaging sequences (i.e., 4-D data) and rely on an approach that follows the shape properties of the endocardial and epicardial surfaces of the LV over the entire cardiac cycle. Our efforts involve the development of an acute infarct animal model that permits us to establish the validity of our noninvasive image analysis algorithms, as well as permits us to study the efficacy of using in vivo, image-derived measures of function for predicting regional myocardial viability (immediately post mortem). We first describe the experimental setup for the animal model, including the use of implanted imaging-opaque markers that assist in setting up a gold standard against which image-derived measurements can be evaluated. Next, the imaging techniques are described, and finally the image analysis methods and their comparison to the validation technique are discussed.

Integrated Skeleton and Boundary Shape Representation for Medical Image Interpretation

We propose a method of extracting and describing the shape of features from medical images which provides both a skeleton and boundary representation. This method does not require complete closed boundaries nor regularly sampled edge points. Lines between edge points are connected into boundary sections using a measure of proximity. Alternatively, or in addition, known connectivity between points (such as that available from traditional edge detectors) can be incorporated if known. The resultant descriptions are objectcentred and hierarchical in nature with an unambiguous mapping between skeleton and boundary sections.