Maximum Intensity Projection Algorithm Research Articles

Tecniche di angio-RM: Stato dell'arte

MRI is very sensitive to flow effects and they may be exploited to develop non-invasive techniques for imaging of the vascular anatomy. These techniques are referred to as Magnetic Resonance Angiography (MRA). MRA techniques can be classified in two major categories: time-of-flight (TOF) and phase contrast (PC). Both techniques rely on different physical effects; inflow-enhancement and flow-induced phase shift respectively. In both techniques definitive angiographic images are obtained by means of post-processing of data with a Maximum Intensity Projection (MIP) algorithm. TOF MRA can be classified in three major categories: 2D-sequential, 3D-volumetric and multi-slab-3D (MOTSA). To improve vessel-background contrast in TOF MRA, new techniques, such as ramped radiofrequency (rf), pulses across the slab to partially compensate for the progressive saturation of spins (TONE) and magnetization transfer contrast (MTC) have been introduced. In PC MRA complex subtraction of two data sets with a different amount of flow sensitivity produces an image with signal intensities depending on flow velocity. Quantitative information about flow velocity and flux can also be obtained in PC MRA. More recently the use of contrast agents with ultrafast imaging (CE-MRA) has shown significant improvements in the delineation of the vessel lumen.

Rapid Lumbar Spine MR Myelography: Imaging Findings Using a Single-Shot Technique

MR myelography has been performed by several authors. Most authors have used techniques based on three-dimensional gradient-echo pulse sequences or fast spin-echo pulse sequences. The examination time varied between 5 and 13 minutes and postprocessing with a maximum-intensity projection algorithm was necessary for three-dimensional visualization. The rapid acquisition with relaxation enhancement (RARE) was initially described by Hennig et Al. In this technical note we present our experience with a single-shot turbo spin-echo sequence, derived from RARE, of approximately 2 seconds to obtain one view. MR myelography can replace conventional myelography in all different types of pathology. In addition MR myelography provides additional information compared to MR imaging in a minority of the cases. Taking into account the ultra-rapid acquisition and the absence of postprocessing procedures, we recommended this technique in the appropriate clinical setting. The MR myelogram increases the level of conficence of the neuroradiologist.

MR angiography of intracranial aneurysms: A comparison of 0.5 T and 1.5 T

The objective of this paper is to compare the diagnostic efficacy of 3 DFT time-of-flight MR angiography (MRA) at middle-field-strength and high-field-strength in diagnosis of intracranial aneurysms. Thirty-one patients, including 26 patients with angiographically confirmed intracranial aneurysms ( n = 28), underwent MRA at 0.5 and 1.5 T. Images were interpreted by six trained observers who were blinded to diagnosis. Twelve projections of MRA of the circle of Willis, by maximum-intensity projection algorithm, were reviewed using continuous confidence-judgment scales. It was found that MRA at 1.5 T was more sensitive than that at 0.5 T in detection of aneurysms. The sensitivity of six observers ranged between 50 and 64% (mean 56%) at 0.5 T, and between 61 and 86% (mean 75%) at 1.5 T. In the smaller aneurysms less than 5 mm, the differences between the units became greater. The sensitivity for small aneurysms ranged from 23 to 54% (mean 32%) at 0.5 T, and from 31 to 69% (mean 58%) at 1.5 T. Our conclusion was that high field strength confers higher accuracy in the detection of intracranial aneurysms with MRA with current-generation MR imagers.

MR hysterography using axial long TR imaging with three-dimensional projections of the uterus

A rapid screening technique for the presentation of uterine cavity anatomy obtained with MR is described. MR hysterographic (MRHG) images were generated and evaluated for uterine cavity abnormalities. Six patients referred for MRI evaluation of the uterus for infertility had an MRI using a pelvic phased array coil with axial 3 mm interleaved fast spin echo images (TR 9000 ms, TE 288 ms with fat saturation) processed using maximum intensity projection algorithms to construct MRHG images. Using the conventional MR sequence as the gold standard the MRHG images were evaluated by two reviewers and accuracy calculated. MRHG took less than 3 min, demonstrated the correct diagnosis in all with a quality score of diagnostic to optimal, and with excellent interobserver agreement (kappa 0.9)

The importance of ray pathlengths when measuring objects in maximum intensity projection images

It is important to understand any process that affects medical data. Once the data have changed from the original form, one must consider the possibility that the information contained in the data has also changed. In general, false negative and false positive diagnoses caused by this post-processing must be minimized. Medical imaging is one area in which post-processing is commonly performed, but there is often little or no discussion of how these algorithms affect the data. This study uncovers some interesting properties of maximum intensity projection (MIP) algorithms which are commonly used in the post-processing of magnetic resonance (MR) and computed tomography (CT) angiographic data. The appearance of the width of vessels and the extent of malformations such as aneurysms is of interest to clinicians. This study will show how MIP algorithms interact with the shape of the object being projected. MIP's can make objects appear thinner in the projection than in the original data set and also alter the shape of the profile of the object seen in the original data. These effects have consequences for width-measuring algorithms which will be discussed. Each projected intensity is dependent upon the pathlength of the ray from which the projected pixel arises. The morphology (shape and intensity profile) of an object will change the pathlength that each ray experiences. This is termed the pathlength effect. In order to demonstrate the pathlength effect, simple computer models of an imaged vessel were created. Additionally, a static MR phantom verified that the derived equation for the projection-plane probability density function (pdf) predicts the projection-plane intensities well (R(2)=0.96). Finally, examples of projections through in vivo MR angiography and CT angiography data are presented.

Optimization of parameters for the detection of cerebral aneurysms: CT angiography of a model.

To optimize parameters with computed tomographic angiography for the detection of cerebral aneurysms. Model aneurysms were placed randomly at various branch points and scanned multiple times with spiral technique. The final analysis included 63 branch points and 22 aneurysms. Each spiral scan used a different parameter combination. Collimation ranged from 1.5 to 4.0 mm and pitch ranged from 1:1 to 1.5:1. Images were constructed with shaded surface display (SSD) and maximum intensity projection (MIP) algorithms and were interpreted by three readers for the presence or absence of aneurysm. The receiver operating characteristic (ROC) curve area for 1.5-mm collimation was greater than those of 3- or 4-mm collimation (P < .01 and P < .001, respectively). There was no statistically significant difference in the ROC curve areas between 3- and 4-mm collimation (P = .37). There was no statistically significant decrease in ROC curve area when increasing pitch from 1:1 to 1.5:1 for any value of collimation (P = .96). For all parameter combinations the ROC curve areas for SSD images was greater than that of MIP images (P < .0001). For cerebral aneurysm detection, narrow collimation is superior to wider collimation. Mild increases in pitch do not substantially degrade diagnostic accuracy. SSD offers improved diagnostic accuracy over MIP display in this model.

Carotid magnetic resonance angiography: improved image quality with dual 3-inch surface coils.

Magnetic resonance angiography (MRA) hs inherent artifacts due to variation in velocity and direction of flowing blood in the carotid bulb and regions of stenosis. We examined the efficiency of dual 3-inch surface coils to delineate carotid artery flow better. Carotid MRA was performed on ten healthy volunteers and six patients, on a 1.5 T system. A special adapter was constructed to use with 3-inch (receive-only) coils, which were placed over the carotid bifurcations. Routine anterior neck coils were also used. Contiguous axial two-dimensional (45/8.7, 1.5 mm, flip angle 60 degrees) time-of-flight sequences were used. Image matrix was 256 x 256 with two signals averaged and acquisition time 6-10 min. These images were postprocessed and reformatted into angiographic views using a maximum intensity projection algorithm. Computer simulation of carotid artery blood flow throughout the cardiac cycle based on vessel contours derived from digital subtraction angiography was carried out by finite element analysis. Improved definition of vessel margin, particularly at the carotid bifurcation, and substantially increased signal-to-background ratio of flowing blood were obtained with 3-in-chcoils. Apparent loss of signal in the carotid bulb was diminished. In one patient, contiguous flow throughout a high-grade stenosis was well defined, with the surface coil method, while drop-off of signal was observed with routine neck coil imaging.

Improved target volume definition in radiosurgery of arteriovenous malformations by stereotactic correlation of MRA, MRI, blood bolus tagging, and functional MRI.

In this methodological paper the authors report the stereotactic correlation of different magnetic resonance imaging (MRI) techniques [MR angiography (MRA), MRI, blood bolus tagging (STAR), and functional MRI] in 10 patients with cerebral arteriovenous malformations (AVM) and its application in precision radiotherapy planning. The patient's head was fixed in a stereotactic localization system that is usable at the MR and the linear accelerator installations. By phantom measurements different materials (steel, aluminium, titanium, plastic, wood, ceramics) used for the stereotactic system were tested for mechanical stability and geometrical MR image distortion. All metallic stereotactic rings (closed rings made of massive metal) led to a more or less dramatic geometrical distortion and signal cancellation in the MR images. The best properties-nearly no distortion and high mechanical stability-are provided by a ceramic ring. If necessary, the remaining geometrical MR image distortion can be "corrected" (reducing displacements to the size of a pixel) by calculations based on modeling the distortion as a fourth-order two-dimensional polynomial. Using this method multimodality matching can be performed automatically as long as all images are acquired in the same examination and the patient is sufficiently immobilized. Precise definition of the target volume could be performed by the radiotherapist either directly in MR images or in calculated projection MR angiograms obtained by a maximum-intensity projection algorithm. As a result, information about the hemodynamics of the AVM was provided by a three-dimensional (3D) phase-contrast flow measurement and a dynamic MRA with the STAR technique leading to an improved definition of the size of the nidus, the origin of the feeding arteries, and the pattern of the venous drainage. In addition, functional MRI was performed in patients with lesions close to the primary motor cortex area leading to an improved definition of structures at risk for high-dose application in radiosurgery. The different imaging techniques of MR provide a sensitive, noninvasive, 3D method for defining target volume, critical structures, and for calculating dose distributions for radiosurgery of cerebral arteriovenous malformations, because dose calculation of radiosurgery at sufficient accuracy can be based on 3D MR data of the geometrical conformation of the patient's head.

MR myelography of the cervical spine.

In previous studies, magnetic resonance (MR) myelography was not effective in the cervical region. However, effective cervical MR myelography is possible with a modified fast spin-echo sequence (8,000/360 [repetition time msec/effective echo time msec], four signals averaged, echo train length of 24). In a clinical study of this protocol, MR myelography was performed as an additional sequence following conventional MR imaging; composite images were obtained with a maximum-intensity projection algorithm. The MR myelographic sequence added only 8.5 minutes to the total imaging time and yielded information not provided by MR imaging in 13% of patients. MR myelography was especially valuable in demonstrating abnormal intraspinal vascularity and in guiding patient treatment by providing detailed preoperative information about intradural masses and posttraumatic and postoperative diverticulum and meningocele. This MR myelographic technique is a useful adjunct to MR imaging, especially in cases in which the more detailed information provided will be helpful in the analysis of complex intraspinal disease.

Three-dimensional spiral CT during arterial portography: comparison of three rendering techniques.

The three most common techniques for three-dimensional reconstruction are surface rendering, maximum-intensity projection (MIP), and volume rendering. Surface-rendering algorithms model objects as collections of geometric primitives that are displayed with surface shading. The MIP algorithm renders an image by selecting the voxel with the maximum intensity signal along a line extended from the viewer's eye through the data volume. Volume-rendering algorithms sum the weighted contributions of all voxels along the line. Each technique has advantages and shortcomings that must be considered during selection of one for a specific clinical problem and during interpretation of the resulting images. With surface rendering, sharp-edged, clear three-dimensional reconstruction can be completed on modest computer systems; however, overlapping structures cannot be visualized and artifacts are a problem. MIP is computationally a fast technique, but it does not allow depiction of overlapping structures, and its images are three-dimensionally ambiguous unless depth cues are provided. Both surface rendering and MIP use less than 10% of the image data. In contrast, volume rendering uses nearly all of the data, allows demonstration of overlapping structures, and engenders few artifacts, but it requires substantially more computer power than the other techniques.

Magnetic resonance angiography of intravascular endoprostheses: investigation of three devices.

To assess the value of magnetic resonance angiography (MRA) in the evaluation of vascular patency after intravascular endoprosthesis placement. Three different metallic stents (Wallstent, Strecker, Palmaz) were studied in vitro, and in vivo in six patients with spin-echo (SE) and gradient-echo (GRE) MR imaging. Time-of-flight, two-dimensional (2D) gadolinium-enhanced MRA was performed with GRE and flow-compensation technique, and reconstructed with a maximum-intensity projection (MIP) algorithm. MRA was compared to digital angiograms. In vitro studies demonstrated that the signal intensity (SI) within the stent differed according to the device employed, the lowest SI being observed within the Palmaz stent (p = .001). There was no difference in SI or apparent diameter of the stent according to the sequence (SE vs GRE) or length of echo time (TE). In patients, the endoprostheses recorded as a well-defined area of signal void or drop-out (p = 0.004), whereas vessels above and below the stent displayed high signal intensities. MRA does not seem as yet to be well suited for evaluating vascular patency after endoprosthesis placement, even if the Strecker and Wallstent endoprostheses provide fewer artifacts than the Palmaz stent.

CT angiography of the common carotid artery bifurcation: comparison between two techniques and conventional angiography

We prospectively compared CT angiography (CTA) of the common carotid artery bifurcation using two different techniques with conventional angiography in patients with suspected stenoses of the internal carotid arteries in 20 symptomatic patients. Ten patients (Group 1) received 60 cc of contrast (medium 2 cc/sec) and CTA was acquired using 5 mm slices, reconstructed at 3 mm slice thickness. Ten patients (Group 2) received 90 cc of contrast (medium 3 cc/sec) and CTA was acquired using 3 mm slices reconstructed at 1 mm slice thickness. All CTA studies were postprocessed using maximum intensity projection algorithm. Stenoses were graded prospectively from CT angiograms and compared with selective conventional catheter angiograms. In Group 1, CTA overestimated the degree of narrowing in 9 of 10 stenoses proven by conventional angiograms. We interpreted 2 nearly occluded internal carotid arteries, 2 with moderate and marked stenoses, and 2 with no narrowings, but fibromuscular dysplasia on conventional angiograms as occluded on CTA, and 3 vessels as showing marked stenoses, not confirmed by angiography. CTA clearly depicted 1 mild stenosis, 4 normal bifurcations, and 6 occluded internal carotid arteries. In Group 2, CTA overestimated two stenoses; a correct diagnosis was made in 7 normal bifurcations, 3 mild, 2 moderate and 2 severe stenoses, 2 near occlusions, and 2 occlusions. Ulcerations were missed by CTA regardless of the technique utilized.

Intracranial vascular stenosis and occlusion: diagnostic accuracy of three-dimensional, Fourier transform, time-of-flight MR angiography.

To assess the accuracy of three-dimensional, Fourier transform, time-of-flight magnetic resonance (MR) angiography in the detection of intracranial steno-occlusive diseases. One hundred thirty-one patients (62 male and 69 female patients, aged 6-77 years [mean, 53 years 8 months]) underwent MR and conventional angiography for evaluation of possible intracranial vascular disease. A total of 502 arteries were assessed. Eight projections and a collapsed image postprocessed by means of a maximum-intensity projection algorithm were reviewed by five observers in a blinded manner, with conventional angiography as the standard. A total of 32 steno-occlusive lesions were available for review. Receiver operating characteristic analysis from the pooled data revealed overall sensitivities of 85% and 88% and specificities of 96% and 97% for the internal carotid artery and the middle cerebral artery, respectively. MR angiography is useful as the primary diagnostic tool for evaluating suspected intracranial steno-occlusive disease.

Visualization of vasculature from volume data

In biomedical computing the need for visualization methods of the human vascular system has been triggered by recent advances in image acquisition technology. In this paper we describe a number of approaches for the three-dimensional display of vessels from volumetric datasets. Our approaches are based on the analysis of the deficiencies of the Maximum Intensity Projection algorithm, which today is the state-of-the-art technique for vascular display, and takes into account different diagnostic and therapeutic situations. For the qualitative as well as quantitative evaluation of the major vessels, e.g., the carotids, a model-driven Computer Vision method to segment, reconstruct and render the vascular tree surface including branches is presented. For the assessment of smaller vessels a Pattern Recognition technique for contrast enhancement of line-like structures is introduced. It serves as a preprocessing step prior to the application of a volume-rendering algorithm that consists of an advanced maximum projection scheme with depth-cueing. The integrated 3D display of soft-tissue surfaces with adjacent vasculature, as required, e.g., for neurosurgery planning, is enabled by a raycaster, simultaneously rendering and merging two volume datasets. We emphasize the clinical relevance of techniques for explorative volume data analysis by a walkthrough example. The paper demonstrates the necessity of incorporating Computer Vision and Pattern Recognition methodologies into the Scientific Visualization pipeline for biomedical imaging.

A fast maximum-intensity projection algorithm for generating magnetic resonance angiograms

Maximum-intensity projection (MIP) algorithms, a class of algorithms for construction of magnetic resonance (MR) angiograms, are reviewed. The blood flow in a volume of interest is represented by bright intensities in the MR data volume. MIP algorithms search for the maximum intensity along parallel rays cast through the MR image volume. A projection image is formed from these maximum intensity values. The flow within the vasculature shows up in the projection plane. An approach to calculating projections in which each image slice is presorted into bins of intensities is discussed. By relating the intensities of the pixels to their location in the slice, the total number of pixels considered for the projection plane is reduced, saving calculation time. Only the brighter intensities that relate to flow are used in the projection. Additional time savings result from precalculating projection templates and filling multiple projection planes at the same time. The algorithm was written in C on a 80386-based system. Results indicate a sixfold increase in projection calculation speed over a benchmark algorithm.

Rapid dynamic contrast-enhanced magnetic resonance imaging of the liver and portal vein.

The purpose of this study was firstly to show the diagnostic value of a rapid acquisition multislice sequence (TurboFLASH) during bolus injection of gadolinium-DTPA by comparing it with the pulse sequences currently used for abdominal studies and secondly to develop improved scanning protocols for the liver. Patients were referred for upper abdominal studies including portal vein assessment. 40 patients were imaged in the coronal plane using a multislice TurboFLASH (TF) sequence (TR = 100; TE = 4) acquired during a breath-hold period of 19 s. The short echo time allows up to 11 slices of 5 mm thickness to be acquired simultaneously. Images were obtained before, during and after bolus administration of Gd-DTPA. The slices from each acquisition were combined using a maximum intensity projection algorithm to include all the vessels on a single image. Initially each patient was scanned using a conventional axial T2 weighted spin-echo sequence (T2W-SE) (TR2000; TE = 45/90) and a coronal T1 weighted spin-echo sequence (TR450; TE = 15). The clarity of the demonstration of vascular anatomy was compared and scored for all sequences by two radiologists. Vessel patency, the conspicuity of mass lesions and the spread of tumour to adjacent structures were also scored. The anatomy of the main portal vein was significantly better shown by coronal TF images after bolus injection than on T2W-SE images or TF before Gd-DTPA. The right and left portal veins were equally well shown by coronal TF with Gd-DTPA and T2W-SE images. There was no significant difference between contrast enhanced TF and T2W-SE imaging in visualization of the hepatic veins. More lesions were demonstrated by post-contrast TF than by T2W-SE imaging. Portal venous occlusion was better appreciated by post-contrast TF. Our results demonstrate that Gd-DTPA TF imaging improves visualization of the main portal vein compared with SE sequences and provides a more accurate assessment of vessel patency. The segmental anatomy of the liver is better appreciated and the demonstration of focal liver lesions compared with T2W-SE images is improved during the non-equilibrium phase of enhancement. TF acquisitions before and after Gd-DTPA are obtained in approximately 5 min; thus a marked reduction in examination time can be achieved.

Clinical potentialities and limitations of two- and three-dimensional time-of-flight MR angiography in the diagnosis of carotid stenosis

For diagnosing a carotid artery stenosis, 26 patients underwent MR angiography. A spin echo (SE) sequence with presaturation on which the flowing blood appears with low signal intensity was applied, as well as a two- and three-dimensional gradient echo (GRE) sequence with flow compensation, showing blood flow with high signal intensity. Subsequently, projection angiograms were made from the MR images with a maximum intensity projection algorithm. Degree, length and localisation of a carotid artery stenosis were reviewed. To find out the clinical usability, the results of the MR angiograms were compared subsequently with the findings of intravenous digital subtraction angiography (i.v. DSA). In comparison with DSA a correlation in the degree of stenosis was noticed in 42 out of 48 SE-, in 39 out of 52 2-D GRE and in 41 out of 48 3-D GRE sequence angiograms. The length of 34 moderate and severe stenoses, demonstrated by DSA, was overestimated 5 times with the SE-sequence, 9 times with the 3-D and 23 times with the 2-D GRE sequence. MR angiography with a 3-D GRE sequence is suitable for screening for carotid artery stenosis. In cases of severe stenosis an SE sequence should be performed for more precise delineation of the stenotic lesion.

MR myelography using heavily T2-weighted fast spin-echo pulse sequences with fat presaturation.

A new method for generating myelogramlike images of the thecal sac by MR imaging is presented. The method is based on suppressing background signal by using heavily T2-weighted fast spin-echo pulse sequences and obliterating fat signal by presaturation. The resulting slices are then projected into a composite image using a standard maximum intensity projection (MIP) algorithm. The technique is implemented with commercially available hardware and software and yields reproducible high-quality images of the lumbar thecal sac, which show excellent definition of the thecal margins, nerve roots, and nerve root sheaths. This method could replace conventional lumbar myelography and postmyelographic CT studies.

Lumbar myelography with three-dimensional MR imaging.

A strongly T2*-weighted, three-dimensional (3D) PSIF (time-reversed FISP [fast imaging with steady-state precession]) gradient-echo magnetic resonance (MR) sequence, with postprocessing of the 3D data set with a maximum-intensity projection (MIP) algorithm, produced x-ray myelography-equivalent images. The method was tested in 10 healthy volunteers to optimize sequence parameters and was evaluated in 30 patients with proven lumbar disk disease. MIP myelograms, unlike two-dimensional MR images, could not show the pathologic disks themselves but clearly demonstrated the effect on the thecal sac, giving a clear overall view of its geometry and dimensions, especially when displayed in a cine loop. All 28 medial and mediolateral disk herniations could be visualized, whereas only three of eight intraforminal disk herniations were seen on MR myelograms.

MRI of intracranial neurovascular compression.

Twelve patients with clinical indications of intracranial neurovascular compression (6, hemifacial spasm; 4, trigeminal neuralgia; 2, vertigo and tinnitus) were studied by MRI. Axial 1 mm thick slices were obtained in a 1.5 T unit using a three-dimensional (3D) fast low angle shot data set of 32 contiguous slices. The data were post-processed with multiplanar reconstruction algorithms to obtain oblique sagittal (along the long axis of a nerve) images or coronal images with a slice thickness of 0.8 mm. The MR angiographic images were also obtained from the same 3D data set with maximum intensity projection algorithms. The MR studies showed that 9 of 12 patients had neurovascular compression caused by one or two arteries. These findings were verified at surgery. In two of the remaining three patients, MRI failed to delineate the affected nerves compressed by a vein in one, and a previously applied prosthesis in the other. In the last patient no neurovascular compression was found by MRI or at surgery.