Registration of MR and CT images for skull base surgery using point-like anatomical features

Monomodal rigid-body registration and applications to the investigation of the effects of eicosapentaenoic acid intervention in neuropsychiatric disorders

The technique of monomodal rigid-body registration of serial magnetic resonance scans based on the sinc ((sin z)/z) interpolation function and its application to neuropsychiatric disorders, such as schizophrenia, depression and Huntington's disease, in relation to the assessment of the cerebral effects of intervention with the n-3 highly unsaturated fatty acid eicosapentaenoic acid are described. The evidence thus far indicates that researchers investigating the benefits of treatment with essential fatty acids in neuropsychiatric disorders should consider utilizing this technique.

Stereotactic radiosurgery planning with ictal SPECT images

This paper is motivated by a clinical requirement to utilise ictal SPECT images for target localisation in stereotactic radiosurgery treatment planning using the xknife system which only supports CT and MRI images. To achieve this, the SPECT images were converted from raw (pixel data only) format into a part 10 compliant DICOM CT fileset. The minimum requirements for the recasting of a raw format image as DICOM CT or MRI data set are described in detail. The method can be applied to the importation of raw format images into any radiotherapy treatment planning system that supports CT or MRI import. It is demonstrated that the combination of the low spatial resolution SPECT images, depicting functional information, with high spatial resolution MRI images, which show the structural information, is suitable for stereotactic radiosurgery treatment planning.

Serial imaging of the brain

Epilepsy is a heterogeneous condition and it is likely that susceptibility factors and genetic predisposition interact with acquired influences such as seizures, AEDs, cerebral insults, subclinical seizures,and ongoing neurodegenerative processes to render certain individuals selectively vulnerable to cerebral damage. There is currently in substantial evidence to suggest that neuroprotective treatments that rely entirely on their anticonvulsant properties are sufficient to prevent the development of cerebral atrophy. The development of postprocessing techniques in serial imaging studies have allowed the detection of subtle changes, and it is likely that development of more sensitive imaging techniques with higher-strength MR magnets and novel MR contrasts will expand our understanding of the factors that render an individual susceptible to hippocampal and extratemporal atrophy. This will allow a more informed assessment of the role required of neuroprotective agents in arresting the progression of cerebral damage and dysfunction.

Metallic copper as a fiducial marker for both CT and PET

There is great interest in augmenting computed tomography (CT) with information gained from other imaging modalities. Positron emission tomography (PET) provides valuable data related to patient physiology to aid in the delineation of tumor volumes. Combining the information provided by these imaging modalities requires accurate spatial registration of the two data sets. Fiducial based mapping provides straightforward registration based on corresponding landmark points or fiducials in the two image sets. When external fiducials are employed, consistent intermodality marker placement and centroid identification are essential to achieving an accurate and reliable registration. Similarity of marker design between modalities greatly aides in achieving this goal. Solid copper may serve as a fiducial marker in both CT and PET. Small spheres or wires of copper are readily visible in CT while neutron activation of these same markers produces positron emitting Copper-64 for detection by PET. The use of identical shaped markers in both imaging modalities greatly simplifies the task of intermodality centroid matching. Copper has excellent machining properties and, prior to activation, is easy and safe to handle. The feasibility of Cu as a marker for both CT and PET is demonstrated using imaging phantoms.

A technique to re-establish dose distributions for previously treated brain cancer patients in external beam radiotherapy

Tumor recurrences or new tumors may develop after irradiation of local lesion(s) in the brain, and additional radiotherapy treatments are often needed for previously treated patients. It is critical to re-establish the dose distributions delivered during the previous treatment in the current patient geometry, so that the previous dose distributions can be accurately taken into consideration in the design of the current treatment plan. The difficulty in re-establishing the previous treatment dose distributions in the current patient geometry arises from the fact that the patient position at the time of reirradiation is different from that at the previous treatment session. Simple re-entry of the previous isocenter coordinates, gantry, and couch and collimator angles into the new treatment plan would result in incorrect beam orientations relative to the new patient anatomy, and therefore incorrect display of the previous dose distributions on the current patient anatomy. To address this issue, a method has been developed so that the previous dose distributions can be accurately re-established in the framework of the current brain treatment. The method involves 3 matrix transformations: (1) transformation of beams from machine coordinate system to patient coordinate system in the previous treatment; (2) transformation of beams from patient coordinate system in the previous treatment to patient coordinate system in the current treatment; and (3) transformation of beams from patient coordinate system in the current treatment to machine coordinate system. The transformation matrices used in the second transformation are determined by registration using a mutual information-based algorithm with which the old and new computed tomography (CT) scan sets are registered automatically without human interpretation. A series of transformation matrices are derived to calculate the isocenter coordinates, the gantry, couch, and collimator angles of the beams for the previous treatment in the current patient geometry, and the previous dose distributions are re-established on the current CT images. The method has been proven to be successful and robust.

Clinical validation of the normalized mutual information method for registration of CT and MR images in radiotherapy of brain tumors

Image registration integrates information from different imaging modalities and has the potential to improve determination of target volume in radiotherapy planning. This paper describes the implementation and validation of a 3D fully automated registration procedure in the process of radiotherapy treatment planning of brain tumors. Fifteen patients with various brain tumors received computed tomography (CT) and magnetic resonance (MR) brain imaging before the start of radiotherapy. First, the normalized mutual information (NMI) method was used for image registration. Registration accuracy was estimated by performing statistical analysis of coordinate differences between CT and MR anatomical landmarks along the x‐, y‐ and z‐axes. Second, a visual validation protocol was developed to validate the quality of individual registration solutions, and this protocol was tested in a series of 36 CT‐MR registration procedures with intentionally applied registration errors. The mean coordinate differences between CT and MR landmarks along the x‐ and y‐axes were in general within 0.5 mm. The mean coordinate differences along the z‐axis were within 1.0 mm, which is of the same magnitude as the applied slice thickness in scanning. In addition, the detection of intentionally applied registration errors by employment of a standardized visual validation protocol resulted in low false‐negative and low false‐positive rates. Application of the NMI method for the brain results in excellent automatic registration accuracy, and the method has been incorporated into the daily routine at our institution. A standardized validation protocol ensures the quality of individual registrations by detecting registration errors with high sensitivity and specificity. This protocol is proposed for the validation of other linear registration methods.

PACS numbers: 87.53.Xd, 87.57.Gg

Functional–Anatomical Image Fusion in Neuroendocrine Tumors

Nuclear medicine provides physiologic and functional data for normal and pathologic organs but often the clear definition of the sites of radiotracers' uptake are difficult. Radiological methods are able to identify structural changes in a detailed way, but do not give precise information on function of organs or pathologic lesions. The registration and fusion of nuclear medicine studies with structural information obtained by radiological exams allows the precise correlation of functional and anatomical data. Software-based fusion of independently performed nuclear medicine and morphologic studies is uncertain of success and the alignment procedures are labor intensive. Recently, a new imaging device combining a dual-head, variable angle gamma camera with a low-dose x-ray tube has been introduced; the acquired single-photon emission computed tomography (SPECT) and x-ray computed tomography (CT) images are coregistered by means of the hardware in the same session. This new technology can be particularly useful when applied to scintigraphic procedures in neuroendocrine tumors. In-111 pentetreotide and radiolabeled MIBG play an important role in the study of patients with these tumors; the addition of anatomical maps provides a precise localization of SPECT findings and allows the exclusion of disease in sites of physiologic tracer uptake. SPECT/CT fused images are able to provide additional information that improves the accuracy of SPECT interpretation and leads to changes in therapeutic options, so enhancing the clinical role of nuclear medicine in evaluating patients with neuroendocrine tumors.

Fusion Imaging in Nuclear Medicine—Applications of Dual-Modality Systems in Oncology

Medical imaging has become of the utmost importance in evaluating patients with cancer. Single-photon emission computed tomography (SPECT) and positron emission tomography (PET) are accurate methods for detecting cancer and related metabolic abnormalities, but they often do not provide the anatomical landmarks needed to precisely localize lesions. Magnetic resonance imaging (MRI) and computed tomography (CT) scan, on the other hand, offer excellent anatomic detail but are less sensitive because they do not provide functional detail. Fusion imaging combines functional studies with morphological ones, so overcoming the drawbacks of both modalities. Software-based fusion of independently performed scintigraphic and radiological images has proven time consuming and impractical for routine use. Recently, dual-modality integrated imaging systems (SPECT/CT and PET/CT) have been developed: the acquired images are coregistered by means of the hardware in the same session. These new devices can be particularly useful for tumour imaging. The anatomical images provide precise localization and allow the exclusion of disease in sites of physiologic tracers' accumulation for SPECT and PET findings. Hybrid imaging in oncological applications has been very encouraging, indicating that these systems are suited for routine use in clinical practice. In fact, fused images provide additional information that improves diagnostic accuracy and impacts on patient management.

Alignment of CT images of skull dysmorphology using anatomy-based perpendicular axes

Rigid body registration of 3D CT scans, based on manual identification of homologous landmarks, is useful for the visual analysis of skull dysmorphology. In this paper, a robust and simple alignment method was proposed to allow for the comparison of skull morphologies, within and between individuals with craniofacial anomalies, based on 3D CT scans, and the minimum number of anatomical landmarks, under rigidity and uniqueness constraints. Three perpendicular axes, extracted from anatomical landmarks, define the absolute coordinate system, through a rigid body transformation, to align multiple CT images for different patients and acquisition times. The accuracy of the alignment method depends on the accuracy of the localized landmarks and target points. The numerical simulation generalizes the accuracy requirements of the alignment method. Experiments using a human dried skull specimen, and ten sets of skull CT images (the pre- and post-operative CT scans of four plagiocephaly, and one fibrous dysplasia patients), demonstrated the feasibility of the technique in clinical practice.

Display of fused images: methods, interpretation, and diagnostic improvements

The use of integrated visualization for medical images aims at assisting clinicians in the difficult task of mentally translating and integrating medical image data from multiple sources into a three-dimensional (3D) representation of the patient. This interpretation of the enormous amount and complexity of contemporary, multiparameter, and multimodal image data demands efficient methods for integrated presentation. This article reviews methods for fused display with the main focus on integration of functional with anatomical images. First, an overview of integrated two-dimensional (2D) and 3D medical image display techniques is presented, and topics related to the interpretation of the integrated images are discussed. Then we address the key issue for clinical acceptance, ie, whether these novel visualization techniques lead to diagnostic improvements. Methods for fused display appear to be powerful tools to assist the clinician in the retrieval of relevant information from multivariate medical image data. Evaluation of the different methods for fused display indicates that the diagnostic process improves, notably as concerns the anatomical localization (typically of functional processes), the registration procedure, enhancement of signal, and efficiency of information presentation (which increases speed of interpretation and comprehension). Consequently, fused display improves communication with referring specialists, increases confidence in the observations, and facilitates the intra- and intersubject comparison of a large part of the data from the different sources, thereby simplifying the extraction of additional, valuable information. In most diagnostic tasks the clinician is served best by providing several (interactive and flexible) 2D and 3D methods for fused display for a thorough assessment of the wealth of image information from multiple sources.

Software for image registration: algorithms, accuracy, efficacy

Image registration is finding increased clinical use both in aiding diagnosis and guiding therapy. There are numerous algorithms for registration, which all involve maximizing a measure of similarity between a transformed floating image and a fixed reference image. The choice of the similarity measure depends, to some extent, on the application. Methods based on the use of the joint intensity histogram have become popular because of their flexibility and robustness. A distinction is made between rigid-body and non-rigid transformations. The latter are needed for inter-subject registration or intra-subject registration in cases where the region of the body of interest is not considered rigid. Non-rigid transformation is normally achieved using a global model of the deformation but can also be defined by a set of locally rigid transformations, each constrained to a small block in the image. There is scope for further research on the incorporation of appropriate constraints, especially for the application of non-rigid transformations to nuclear medicine studies. Most of the initial practical concerns regarding image registration have been overcome and there is increasing availability of commercial software. There are several approaches to the validation of registration software, with validation of non-rigid algorithms being particularly difficult. Studies have demonstrated the accuracy on the order of half a pixel for both intra- and inter-modality registration (typically 2 to 3 mm). Although hardware-based registration has now become possible by using dual-modality instruments, software-based registration will continue to play an important role in nuclear medicine.

3-D US frame positioning using speckle decorrelation and image registration

In this paper, a new positioning system is proposed for the 3-D ultrasound (US). This system combines the image registration technique and speckle decorrelation algorithm to accurately position sequential ultrasonic images without any additional positioning hardware. The speckle decorrelation algorithm estimates the relative distance of two neighboring frames and the image registration technique gets the range of the whole 3-D ultrasonic data set and makes slight modification on each frame's position. The image registration technique is based on the reference image, which is perpendicular to the 3-D ultrasonic data set. This reference image intersects each frame of the 3-D ultrasonic data set in a line. For each frame, the intersectional line is first found and then the location in the reference image can be used to estimate the position of this frame. This system uses the data set of consecutive 2-D freehand-scanned US B-mode images to construct the 3-D US volume data, and it can be integrated into the 3-D US volume rendering system.

A comparison of planar scintigraphy and SPECT measurement of total lung deposition of inhaled aerosol

Planar gamma camera imaging of inhaled aerosol deposition is extensively used to assess the total deposition in the lung. However, validation of the measurements is not straightforward, as gold standard measurements of lung activity against which to compare are not readily available. Quantitative SPECT imaging provides an alternative method for comparison. Four different methods for planar image quantification are compared. Two attenuation correction techniques, thickness measurement and transmission measurement, have been combined with two scatter correction techniques, reduced attenuation coefficient and line-source scatter function convolution subtraction. Each technique has been applied to 10 studies of aerosol deposition of a fine aerosol (mass median aerodynamic diameter 1.8 microm) and 10 studies using a coarse aerosol (mass median aerodynamic diameter 6.5 microm). The total activity in the right lung for each measurement has been compared to the value determined from SPECT imaging on the same subjects. When the thickness measurement and transmission techniques were applied with scatter compensation using a reduced attenuation coefficient, activity was systematically overestimated by 5% in both cases. The corresponding random errors (coefficient of variation) were 8.6% and 6.6%. Separate scatter correction reduced these systemic errors significantly to -1.5% and 2.7%, respectively. The random errors were not affected. All techniques provided assessment of total lung activity with an accuracy and precision that differed by less than 10% compared to the SPECT values. Planar gamma camera imaging provides a good method of assessing total lung deposition of inhaled aerosol.

On the incorporation of multi-modality image registration into the radiotherapy treatment planning process

A technique is presented that allows the direct use of physiological image sets in the radiation therapy treatment planning process. When fused to the treatment planning CT, physiological image studies may allow one to define physiological tumor subvolumes consisting of areas of possible chronic hypoxia, areas of high perfusion, areas of high diffusion, and areas containing high choline concentrations. These physiological tumor subvolumes could be selectively boosted to increase local control of malignant brain tumors once one has determined which of these physiological tumor subvolumes predicts for local tumor recurrence after conventional radiotherapy. In this technique a user assisted automatic registration technique is used that is based on an analytical estimate for the transformation matrix needed to register two rigid bodies. The only user input needed is three non-collinear points selected based on landmarks in the primary image and the corresponding three points in the secondary image. Since this registration technique uses two sets of at least three user-defined landmark points each of which has some selection error associated with it, the final registration will have an error that depends only on the selection error associated with the point sets. Since physiological image studies are acquired at the same setting as the T1- w MRI their spatial orientation with respect to the T1- w MRI is known. Therefore, the registration of multiple physiological image studies to the treatment planning CT can be accomplished by first correlating them to the T1- w MRI, and in a second step the T1- w MRI is then registered to the treatment planning CT. The desired registration of the physiological image studies to the treatment planning CT is then accomplished by simply composing the appropriate transformation matrices.

Subvoxel image registration of multislice (2D) magnetic resonance images in patients with high-grade gliomas of the brain

AIMS

To implement a multislice two-dimensional (2D) T2-weighted sequence suitable for subvoxel image registration and to assess its usefulness in detecting change in high-grade intracranial gliomas.

MATERIALS AND METHODS

Twenty patients with high-grade gliomas were studied on two or more occasions. T2-weighted multislice pulse sequences with a Gaussian slice profile, 50% overlapping slices and nearly isotropic voxels were acquired. The images were registered and subtraction images were produced. The images were compared with three-dimensional (3D) T1-weighted registered images and conventional unregistered T2-weighted images. All images were scored for changes in the lesions and ventricular system.

RESULTS

The 2D and 3D registered subtraction images were the most sensitive for detecting changes in both the lesions and other regions in the brain. The mean rank scores were significantly higher for the lesions (chi2=86.742; df=5, n=38, P<0.0001) and for the ventricles (chi2=63.837; df=5, n=35, P<0.0001) compared with the unregistered and registered anatomical images. The subtraction images were also most sensitive for detecting signal intensity changes irrespective of the direction of change.

CONCLUSION

Rigid body subvoxel registration can be successfully performed with both multislice 2D and 3D imaging. In principle, virtually all forms of clinical MR images of the brain can be accurately registered and subtracted.

Image fusion for skull base neuronavigation. Technical note

An automatic image fusion module (BrainLab, Munich, Germany) is used for the fusion of the magnetic resonance (MR) imaging and computed tomography (CT) data sets. The procedure of image fusion takes 5 minutes prior to surgery. The image fusion of CT and MR imaging data visualizes the skull base and tumor margins clearly. Color display of the different data sets allows the tumor and the skull base to be distinguished easily. The fused CT data in bone window mode provides useful additional information on the osseous skull base.

A comparative study of surface- and volume-based techniques for the automatic registration between CT and SPECT brain images

Image registration of multimodality images is an essential task in numerous applications in three-dimensional medical image processing. Medical diagnosis can benefit from the complementary information in different modality images. Surface-based registration techniques, while still widely used, were succeeded by volume-based registration algorithms that appear to be theoretically advantageous in terms of reliability and accuracy. Several applications of such algorithms for the registration of CT-MRI, CT-PET, MRI-PET, and SPECT-MRI images have emerged in the literature, using local optimization techniques for the matching of images. Our purpose in this work is the development of automatic techniques for the registration of real CT and SPECT images, based on either surface- or volume-based algorithms. Optimization is achieved using genetic algorithms that are known for their robustness. The two techniques are compared against a well-established method, the Iterative Closest Point-ICP. The correlation coefficient was employed as an independent measure of spatial match, to produce unbiased results. The repeated measures ANOVA indicates the significant impact of the choice of registration method on the magnitude of the correlation (F = 4.968, p = 0.0396). The volume-based method achieves an average correlation coefficient value of 0.454 with a standard deviation of 0.0395, as opposed to an average of 0.380 with a standard deviation of 0.0603 achieved by the surface-based method and an average of 0.396 with a standard deviation equal to 0.0353 achieved by ICP. The volume-based technique performs significantly better compared to both ICP (p<0.05, Neuman Keuls test) and the surface-based technique (p<0.05, Neuman-Keuls test). Surface-based registration and ICP do not differ significantly in performance.

Automated CT marker segmentation for image registration in radionuclide therapy

In this paper a novel, automated CT marker segmentation technique for image registration is described. The technique, which is based on analysing each CT slice contour individually, treats the cross sections of the external markers as protrusions of the slice contour. Knowledge-based criteria, using the shape and dimensions of the markers, are defined to enable marker identification and segmentation. Following segmentation, the three-dimensional (3D) markers' centroids are localized using an intensity-weighted algorithm. Finally, image registration is performed using a least-squares fit algorithm. The technique was applied to both simulated and patient studies. The patients were undergoing 131I-mIBG radionuclide therapy with each study comprising several 99mTc single photon emission computed tomography (SPECT) scans and one CT marker scan. The mean residual 3D registration errors (+/- 1 SD) computed for the simulated and patient studies were 1.8 +/- 0.3 mm and 4.3 +/- 0.5 mm respectively.

A serial longitudinal quantitative MRI study of cerebral changes in first-episode schizophrenia using image segmentation and subvoxel registration

Lateral ventricular enlargement is the most consistently replicated brain abnormality found in schizophrenia. This article reports a first episode, longitudinal study of ventricular volume using high-resolution serial magnetic resonance imaging (MRI) and recently developed techniques for image registration and quantitation. Baseline and follow-up (on average 8 months later) MRI scans were carried out on 24 patients and 12 controls. Accurate subvoxel registration was performed and subtraction images were produced to reveal areas of regional brain change. Whereas there were no differences between patients and controls with respect to the mean change in ventricular volume, the patients were much more variable in this respect and showed larger increases and decreases. The percentage increase in ventricular size was greater than one standard deviation of control values for 14 patients and the percentage decrease exceeded one standard deviation in eight patients. Although the finding of progressive ventricular enlargement in a proportion of patients supports other studies indicating an ongoing neuropathological process in the early stages of schizophrenia, the reduction of ventricular size in the remaining patients is more difficult to explain. It is suggested that this may reflect improvement in nutrition and hydration following treatment.

Medical image registration

Radiological images are increasingly being used in healthcare and medical research. There is, consequently, widespread interest in accurately relating information in the different images for diagnosis, treatment and basic science. This article reviews registration techniques used to solve this problem, and describes the wide variety of applications to which these techniques are applied. Applications of image registration include combining images of the same subject from different modalities, aligning temporal sequences of images to compensate for motion of the subject between scans, image guidance during interventions and aligning images from multiple subjects in cohort studies. Current registration algorithms can, in many cases, automatically register images that are related by a rigid body transformation (i.e. where tissue deformation can be ignored). There has also been substantial progress in non-rigid registration algorithms that can compensate for tissue deformation, or align images from different subjects. Nevertheless many registration problems remain unsolved, and this is likely to continue to be an active field of research in the future.

Advanced image-guided skull base surgery

BACKGROUND

Tumors of the skull base frequently encase or extend into normal neural and vascular structures. Preoperative planning and intraoperative identification of anatomic landmarks is especially important in complex tumors since it helps avoid or minimize surgical morbidity.

METHODS

By creating a surgical plan the image guidance software offers help in the establishment of a surgical approach. During surgery, the neuronavigation system displays the location of anatomic landmarks of the skull base regardless of any erosion or displacement.

RESULTS

A series of 10 patients with complex tumors in various skull base locations is reported. Osseous structures are easily identified using the CT-based image guidance since these landmarks do not shift due to CSF loss. Image fusion of CT and MRI data gives additional information on the displacement of soft tissue structures. Image fusion in a substraction mode is helpful when a tumor has invaded bony structures or when the encasement of major vessels has to be visualized.

CONCLUSION

The preoperative data preparation (planning of the approach, image fusion) plays a vital role in modern neuronavigation and contributes useful information during surgery for complex skull base tumors. Such advanced neuronavigation increases the efficacy and safety of intraoperative maneuvers. Eroded and distorted anatomic landmarks are not subject to a significant amount of intraoperative shift throughout the surgical procedure.

Coregistration of brain single-positron emission computed tomography and magnetic resonance images using anatomical features

The present report describes a computer tool for the coregistration of single-positron emission computed tomography (SPECT) and magnetic resonance (MR) images to aid the diagnosis. Two types of images are used for some neurologic diseases: one of them anatomic (magnetic resonance) and the other metabolic (SPECT), with the specialist being required to make a mental integration of the examinations. This procedure can be improved by using a computer tool that might permit the presentation of this information in a single image. The coregistration is performed on the basis of pairs of points positioned by a specialist according to the structures present in the images and the least squares error is calculated between them using Euclidean distance. Coinciding planes and section thickness are selected for the two modalities and the SPECT image is processed so as to have the same spatial resolution as the resonance image.

Automatic retinal image registration scheme using global optimization techniques

Retinal image registration is commonly required in order to combine the complementary information in different retinal modalities. In this paper, a new automatic scheme to register retinal images is presented and is currently tested in a clinical environment. The scheme considers the suitability and efficiency of different image transformation models and function optimization techniques, following an initial preprocessing stage. Three different transformation models--affine, bilinear and projective--as well as three optimization techniques--downhill simplex method, simulated annealing and genetic algorithms--are investigated and compared in terms of accuracy and efficiency. The registration of 26 pairs of Fluoroscein Angiography and Indocyanine Green Chorioangiography images with the corresponding Red-Free retinal images, showed the superiority of combining genetic algorithms with the affine and bilinear transformation models. A comparative study of the proposed automatic registration scheme against the manual method, commonly used in the clinical practice, is finally presented showing the advantage of the proposed automatic scheme in terms of accuracy and consistency.

Algorithms for radiological image registration and their clinical application

This paper reviews recent work in radiological image registration and provides a classification of image registration by type of transformation and by methods employed to compute the transformation. The former includes transformation of 2D images to 2D images of the same individual, transformation of 3D images to 3D images of the same individual, transformation of images to an atlas or model, transformation of images acquired from a number of individuals, transformations for image guided interventions including 2D to 3D registration and finally tissue deformation in image guided interventions. Recent work on computing transformations for registration using corresponding landmark based registration, surface based registration and voxel similarity measures, including entropy based measures, are reviewed and compared. Recently fully automated algorithms based on voxel similarity measures and, in particular, mutual information have been shown to be accurate and robust at registering images of the head when the rigid body assumption is valid. Two approaches to modelling soft tissue deformation for applications in image guided interventions are described. Validation of complex processing tasks such as image registration is vital if these algorithms are to be used in clinical practice. Three alternative validation strategies are presented. These methods are finding application outside the original domain of radiological imaging.

Correcting scaling errors in tomographic images using a nine degree of freedom registration algorithm

PURPOSE

Clinical imaging systems, especially MR scanners, frequently have errors of a few percent in their voxel dimensions. We evaluate a nine degree of freedom registration algorithm that maximizes mutual information for determining scaling errors. We evaluate it by registering MR and CT images for each of five patients (patient scaling) and by registering MR images of a phantom to a computer model of the phantom (phantom scaling).

METHOD

Each scaling method was validated using bone-implanted markers localized in the patient images and also intraoperatively. The root mean square residual in the alignment of the fiducial markers [fiducial registration error (FRE)] was determined without scale correction, with patient scaling, and with phantom scaling.

RESULTS

Each scaling method significantly reduced the average FRE (p < 0.05) for MR to CT registration and for MR to physical space registration, indicating that voxel scaling errors were reduced. The greater reduction in scaling errors was achieved using the phantom scaling method.

CONCLUSION

We have demonstrated that a nine degree of freedom registration algorithm that maximizes mutual information can significantly reduce scaling errors in MR.

The use of positron emission tomography and computed tomography in the assessment of trismus associated with head and neck malignancy

The assessment of head and neck cancer has traditionally involved clinical examination and anatomical imaging by computed tomography (CT) or magnetic resonance imaging (MRI). We present a case where a problem of clinical confusion and inconclusive radiology was resolved by the use of positron emission tomography (PET) coregistered with CT.

A survey of medical image registration

The purpose of this paper is to present a survey of recent (published in 1993 or later) publications concerning medical image registration techniques. These publications will be classified according to a model based on nine salient criteria, the main dichotomy of which is extrinsic versus intrinsic methods. The statistics of the classification show definite trends in the evolving registration techniques, which will be discussed. At this moment, the bulk of interesting intrinsic methods is based on either segmented points or surfaces, or on techniques endeavouring to use the full information content of the images involved.

A new frame-based registration algorithm

This paper presents a new algorithm for frame registration. Our algorithm requires only that the frame be comprised of straight rods, as opposed to the N structures or an accurate frame model required by existing algorithms. The algorithm utilizes the full 3D information in the frame as well as a least squares weighting scheme to achieve highly accurate registration. We use simulated CT data to assess the accuracy of our algorithm. We compare the performance of the proposed algorithm to two commonly used algorithms. Simulation results show that the proposed algorithm is comparable to the best existing techniques with knowledge of the exact mathematical frame model. For CT data corrupted with an unknown in-plane rotation or translation, the proposed technique is also comparable to the best existing techniques. However, in situations where there is a discrepancy of more than 2 mm (0.7% of the frame dimension) between the frame and the mathematical model, the proposed technique is significantly better (p < or = 0.05) than the existing techniques. The proposed algorithm can be applied to any existing frame without modification. It provides better registration accuracy and is robust against model mis-match. It allows greater flexibility on the frame structure. Lastly, it reduces the frame construction cost as adherence to a concise model is not required.

Research-oriented image registry for multimodal image integration

To provide multimodal biomedical images automatically, we constructed the research-oriented image registry, Data Delivery System (DDS). DDS was constructed on the campus local area network. Machines which generate images (imagers: DSA, ultrasound, PET, MRI, SPECT and CT) were connected to the campus LAN. Once a patient is registered, all his images are automatically picked up by DDS as they are generated, transferred through the gateway server to the intermediate server, and copied into the directory of the user who registered the patient. DDS informs the user through e-mail that new data have been generated and transferred. Data format is automatically converted into one which is chosen by the user. Data inactive for a certain period in the intermediate server are automatically achieved into the final and permanent data server based on compact disk. As a soft link is automatically generated through this step, a user has access to all (old or new) image data of the patient of his interest. As DDS runs with minimal maintenance, cost and time for data transfer are significantly saved. By making the complex process of data transfer and conversion invisible, DDS has made it easy for naive-to-computer researchers to concentrate on their biomedical interest.

Integration of functional and anatomical brain images

This article concerns the integration of functional and anatomical volumetric brain images. Integration consists of two steps: matching or registration, where the images are brought into spatial agreement, and fusion or simultaneous display where the registered multimodal image information is presented in an integrated fashion. Approaches to register multiple images are divided into extrinsic methods based on artificial markers, and intrinsic matching methods based solely on the patient related image data. The various methods are compared by a number of characteristics, which leads to a clear preference for one class of intrinsic methods, viz. voxel-based matching. Furthermore, two- and three-dimensional techniques to display multimodality image information are outlined.

The registration of multiple medical images acquired from a single subject: why, how, what next?

This paper reviews some of the recent techniques which have been used to register multiple images of the same patient. Image registration is a problem which has been receiving significant attention from the medical image processing community in recent years. A successful image registration can aid in patient diagnosis, treatment assessment, image guided interventions, surgery planning and surgery. At present the majority of work has focused on rigid body transformations of images. We shall discuss some of the approaches used and outline a key automatic method in detail. In order to allow image registration of parts of the body which do not remain rigid, either due to patient movement or a change in pathology, nonlinear deformation techniques are being developed. We shall talk of the history of these methods before explaining deformations using landmarks and a recent extension to allow the definition of rigid structures in such warps in more detail. Validation of these methods is of great importance and we shall discuss work which has already been carried out on this topic for rigid body registrations as well as ideas for the validation of deformation algorithms.

Automated three-dimensional registration of magnetic resonance and positron emission tomography brain images by multiresolution optimization of voxel similarity measures

Approaches using measures of voxel intensity similarity are showing promise in fully automating magnetic resonance (MR) and positron emission tomography (PET) image registration in the head, without requiring extraction and identification of corresponding structures. In this paper a method of multiresolution optimization of these measures is described and five alternative measures are compared: cross correlation, minimization of corresponding PET intensity variation, moments of the distribution of values in the intensity feature space, entropy of the intensity feature space and mutual information. Their ability to recover registration is examined for ten clinically acquired image pairs with respect to the size of initial misregistration, the precision of the final result, and the accuracy assessed by visual inspection. The mutual information measure proved the most robust to initial starting estimate, successfully registering 98.8% of 900 trial misregistrations. Success is defined as providing a visually acceptable solution to a trained observer. A high resolution search (1/16 mm step size) of 30 trial misregistrations showed that optimization using the mutual information measure provided solutions with 0.13 mm, 0.11 mm and 0.17 mm standard deviations in the three Cartesian axes of the translation vector and 0.2 degree, 0.3 degree and 0.2 degree standard deviations for rotations about the three axes. The algorithm takes between 4 and 8 minutes to run on a typical workstation, including visual inspection of the result.

Validation and clinical application of computer-combined computed tomography and positron emission tomography with 2-[18F]fluoro-2-deoxy-D-glucose head and neck images

BACKGROUND

Positron emission tomography with 2-[18F]fluoro-2-deoxy-D-glucose (PET-FDG) improves the detection of head and neck squamous cell cancer (HNSCC), but lacks anatomical detail. The accuracy of registered computed tomography/magnetic resonance (CT/MR) and PET-FDG in delineation of HNSCC at the primary site and its clinical application was investigated.

METHOD

Preoperatively 30 patients were staged clinically and each had either CT (23), MR (5), or both CT and MR (2) scans, as well as CT/MR-PET-FDG registration. Tumor margins or infiltration of specific anatomical landmarks on the different scans were compared and judged against histology.

RESULTS

For primary tumors CT-PET-FDG (97%) and MR-PET-FDG (100%) delineated the tumor more accurately than CT (69%) or MR (40%) alone. Similarly, CT-PET-FDG (98%) and MR-PET-FDG (100%) were better than CT (70%) and MR alone (80%) in identifying tumor invasion of specific anatomical structures. Management was altered in 7 of 30 patients. The registered images were particularly useful in delineating tumor extension in the infratemporal fossa, maxilla and mandible, and identifying recurrences obscured by scar tissue.

CONCLUSIONS

It is possible to accurately register CT, MR, and PET-FDG data sets in the head and neck. The initial results show that registered CT/ MR-PET-FDG images provide additional clinically relevant information over that obtained from clinical evaluation or conventional CT/MR imaging.

Abdominal SPECT/MRI fusion applied to the study of splenic and hepatic uptake of radiolabeled thrombocytes and colloids

The importance of applying MRI (CT)/SPECT fusion in the abdominal and thoracic areas has been recognized in recent studies aiming at radionuclide therapy of cancer. According to our earlier results spleen and liver volume determination with different segmentation methods is inaccurate with SPECT alone. We therefore applied a SPECT/MRI registration procedure to the estimation of spleen and liver volumes and spleen/liver activity ratios in three male volunteers administered 111In-labeled thrombocytes and 99mTc-labeled colloids. The objectives of the study were to investigate if the uptake of thrombocytes in the spleen and liver can be measured more accurately when the anatomical borders of these organs are transferred from MRI to SPECT, and to test a SPECT/MRI registration method for improving three-dimensional dosimetry for radiotherapy treatment planning. A good correlation was found between spleen/liver activity ratios calculated from volumetric average activity per pixel values and from total volumetric counts derived from registered data but not from projection data. The average registration residual with this SPECT/MRI fusion method is approximately 1-2 cm in the abdominal area. Combining anatomical images with SPECT is therefore important for improving quantitative SPECT also in the abdomen.

Effect of geometrical distortion correction in MR on image registration accuracy

In this article we investigate the effect of geometrical distortion correction in MR images on the accuracy of the registration of X-ray CT and MR head images for both a fiducial marker (extrinsic point) method and a surface-matching technique. We use CT and T2-weighted MR image volumes acquired from seven patients who underwent craniotomies in a stereotactic neurosurgical clinical trial. Each patient had four external markers attached to transcutaneous posts screwed into the outer table of the skull. The MR images are corrected for static field inhomogeneity by using an image rectification technique and corrected for scale distortion (gradient magnitude uncertainty) by using an attached stereotactic frame as an object of known shape and size. We define target registration error (TRE) as the distance between corresponding marker positions after registration and transformation. The accuracy of the fiducial marker method is determined by using each combination of three markers to estimate the transformation and the remaining marker to calculate registration error. Surface-based registration is accomplished by fitting MR contours corresponding to the CSF-dura interface to CT contours derived from the inner surface of the skull. The mean point-based TRE using three noncollinear fiducials improved 34%-from 1.15 to 0.76 mm-after correcting for both static field inhomogeneity and scale distortion. The mean surface-based TRE improved 46%-from 2.20 to 1.19 mm. Correction of geometrical distortion in MR images can significantly improve the accuracy of point-based and surface-based registration of CT and MR head images. Distortion correction can be important in clinical situations such as stereotactic and functional neurosurgery where 1 to 2 mm accuracy is required.

An automatic technique for finding and localizing externally attached markers in CT and MR volume images of the head

An image processing technique is presented for finding and localizing the centroids of cylindrical markers externally attached to the human head in computed tomography (CT) and magnetic resonance (MR) image volumes. The centroids can be used as control points for image registration. The technique, which is fast, automatic, and knowledge-based, has two major steps. First, it searches the entire image volume to find one voxel inside each marker-like object. We call this voxel a "candidate" voxel, and we call the object a candidate marker. Second, it classifies the voxels in a region surrounding the candidate voxel as marker or nonmarker voxels using knowledge-based rules and calculates an intensity-weighted centroid for each true marker. We call this final centroid the "fiducial" point of the marker. The technique was developed on 42 scans of six patients-one CT and six MR scans per patient. There are four markers attached to each patient for a total of 168 marker images. For the CT images the false marker rate was zero. For MR the false marker rate was 1.4% (Two out of 144 markers). To evaluate the accuracy of the fiducial points, CT-MR registration was performed after correcting the MR images for geometrical distortion. The fiducial registration accuracy averaged 0.4 mm and was better than 0.6 mm for each of the eighteen image pairs.

Registration and display of brain SPECT and MRI using external markers

Accurate anatomical localisation of abnormalities observed in brain perfusion single-photon emission computed tomography (SPECT) is difficult, but can be improved by correlating data from SPECT and other tomographic imaging modalities. For this purpose we have developed software to register, analyse and display 99mTc-hexamethylpropyleneamine oxime SPECT and 1.0 T MRI of the brain. For registration of SPECT and MRI data external skin markers containing 99mTc (220 kBq) in 50 microliters of coconut butter were used. The software is coded in the C programming language, and the X Window system and the OSF/Motif standards are used for graphics and definition of the user interface. The registration algorithm follows a noniterative least-squares method using singular value decomposition of a 3 x 3 covariance matrix. After registration, the image slices of both data sets are shown at identical tomographic levels. The registration error in phantom studies was on average 4 mm. In the two-dimensional display mode the orthogonal cross-sections of the data sets are displayed side by side. In the three-dimensional mode MRI data are displayed as a surface-shaded 3 D reconstruction and SPECT data as cut planes. The usefulness of this method is demonstrated in patients with cerebral infarcts, brain tumour, herpes simplex encephalitis and epilepsy.

A New Technique for CT/MR Fusion For Skull Base Imaging

This paper presents our initial experience utilizing a new technique which allows CT and MR image fusion in patients with skull base lesions. Eleven patients with a variety of skull base lesions underwent CT and MR imaging prior to surgery. Both sets of images were coregistered using customized software. The CT and MR data sets were then combined and viewed in a single interactive image formar using a high-speed graphic computing system. Image fusion allowed simultaneous visualization of the bony skull base anatomy (CT) and detailed soft tissue anatomy (MR) using a single image format. Combining both modalities was felt to provide a better assessment of the extent of lesions and improve understanding of their relationship to adjacent bony and neurovascular anatomy. Specifically, image fusion enhanced awareness of location of skill base lesions with respect to the cavernous sinuses. Gasserian ganglia, carotid arteries, and jugular foramina. For tumors arising within the internal auditory canal (IAC), fused images allowed better delineation of the lateral aspect of the lesion with respect to the fundus of the IAC. Thus, fusion of CT and MR studies provides a unique image format which has advantages over single modality display. We believe image fusion is beneficial for surgical planning and for treatment planning of complex skull base malignancies treated with radiotherapy.