images by minimizing the penalty term for the weighted distance between filtered image patches and other patches in the search window. This weighted distance was first defined as a decreasing function of the Euclidean distance between gray level intensities in the patches, and then adapted to various high-order operators to enhance the performance [2–5]. This strategy has also been implemented effectively in statistical methods, such as expectation– maximization-based Bayesian estimation  and maximum likelihood estimation  to promote more effective denoising of 3DMRI. On the other hand, based on the assumption that latent images have a sparse representation in some transform domain, numerous trans- forms have been proposed to filter noisy images. These include the discrete cosine transform (DCT) , principal component analysis , over-complete dictionaries , and high-order singular value decomposition . The filtering process was implemented by applying a shrinkage function to the transform coefficients and recovering estimated patches with an inverse transform. For instance, the block matching in three dimensions (BM3D) method  collects a group of similar patches from the reference patches to construct a 3D array. After projecting the 3D array onto a 3D transform basis, the coefficients are truncated by a hard threshold and filtered patches are reconstructed by inversion of the transform.
Denoising can also be performed in transformed domains, where the separation between image and noise is expected to be easier. Of course, a proper transformation to the acquired data has to be applied. Among all, Wavelets are often adopted . In case of MRI image, such approach introduces a bias in the filtered image. In order to reduce this disadvantage, the squared values of the image can be considered as the initial noisy one. However, such approach hardly preserve fine details of the images, especially in case of low signal to noise ratio (SNR). By considering as the transformed domain the spatial frequency one, Wiener filter has also to be taken into account [7, 8].
There have been many researches about the process of building 3D images from 2D medical images in recent years. The 3D surface of the knee or a 3D spine image was constructed from 2D CT images [1, 2] using Marching Cube. This method allows to divide data blocks into cubes and each cube was made up of eight adjacent voxels. From these eight adjacent voxels, material surfaces were built using the triangular mesh. Therefore, the method has the advantage of fast calculation, simple construction operations and produces 3D images with the high resolution. However, the calculation will be slow, if one processes the large number of 2D image data. In addition, the images captured from sensors often have noise, because to improve the quality of constructing a 3D image, 2D images need to be pre-processed to reduce noise. In particular, a mean-unsharp filter may be applied to enhance high frequency components and filtered noise.
slice gaps, making it impossible to generate orthogonal or oblique reconstructions from dataset thus generated without significant loss of quality. Three dimensional (3D) MRI acquisitions with isotropic or near- isotropic resolution have the potential to generate complete planar thin reconstructions with overlapping sections can be generated in virtually any plane. Not only can this potentially improve the spatial resolution of quired images and avoid possible loss of information due to slice gaps in 2D sequences, it can also potentially improve the efficiency and increases patient comfort by reducing total time of study by eliminating the need to repeat 2D sequences in , 2009 and Kijowski, 2010; Ristow, , 2012). If established that multiplanar reconstructions acquired from 3DMRI protocols can provide at least a similar image quality (if not better) in lesser scan those provided by standard 2D MRI protocols, replacing 2D protocols by 3D protocols could have a significant impact on clinical practice and research OF CURRENT RESEARCH
This is a prospective study conducted over a period of two years in the Department of Radiodiagnosis and Imaging, Kasturba Medical College, Manipal. Institutional ethics committee approval was obtained. Forty five patients with shoulder pain and suspected glenoid labral tears are selected for the study. The sample size was decided by the statistician who was provided with the relevant data. Patients with previous shoulder surgery and with tumour in the joint are excluded from the study. Study was performed with 1.5T MR imaging system by Achieva, PHILIPS, Netherlands using a dedicated 8 channel shoulder coil in supine position with shoulder in neutral position. These patients were evaluated with conventional MRI (PDFS (coronal, axial), GRE axial, T2W sagittal, T2W FS Coronal and 3DMRI (PD VISTA) prior to MR arthrography (T1W TSE with fat saturation and T1WVISTA) and findings are compared with the final diagnosis obtained by arthroscopy (whenever performed). Procedure of MR Arthrography in brief: Patient lies in prone position and under USG guidance, intraarticular injection of 15-20ml diluted solution of gadolinium (12 drops of gadolinium with bupivacaine and normal saline) was done (Cicak et al., 1992). Later MRI sequences were acquired within fourty five minutes to one hour. Labrum is divided in to six segments based on clock positions for localization of tear. They are superior, antero-superior, antero-inferior, inferior, postero- inferior and postero-superior labral segments. Categorization of all labral segments was made into either normal or abnormal (tear absent/present-suspicious/definitive). Normal labrum shows normal contour and signal intensity without any irregularity. Suspicious tear is defined as ill-defined hyperintensity with irregular margins / contour abnormalities / attenuation of a part of labrum/no contrast extension into hyperintensity on MRA. Definitive tear is defined as sharp defect in the labrum with gap measuring >2 mm/ extension of contrast material into the defect on MRA/ non visualization of labrum.
Abstract — A novel algorithm for automatic head and neck 3D tumour segmentation from magnetic resonance imaging (MRI) is presented. The proposed algorithm pre-processes the MRI data slices to enhance quality and reduce artefacts. An intensity standardisation process is performed between slices, followed by cancer region segmentation of central slice, to get the correct intensity range and rough location of tumour regions. Fourier interpolation is applied to create isotropic 3DMRI volume. A new location-constrained 3D level set method segments the tumour from the interpolated MRI volume. The proposed algorithm is tested on real MRI data. The results show that the novel 3D tumour volume extraction algorithm has an improved dice score and F-measure when compared to the previous 2D and 3D segmentation method.
Switzerland). We have previously described the measure- ments of NVH and MNP in detail . Briefly, owing to the 3D nature of the MRI sequences the imaging planes could be adjusted in the multiplanar reconstruction (MPR) module of the DICOM viewer in a standardised fashion be- fore actual measurements of NVH and MNP (Fig. 2). Time consumption to perform all measurements of NVH and MNP was below 10 min per subject. One of the co-authors a 3rd year resident radiologist (PH) performed all radio- logical measurements reported in the present study. NVH and NMP were measured in SUP, ST and ST + W.
The apparent diffusion coefficient (ADC) calculated from hyperpolarized 3 He diffusion-weighted MRI (DW-MRI) has been shown to be sensitive to changes in lung micro- structure (1,2). The non-Gaussian diffusion behavior of the gas in the lungs results in a non-monoexponential signal attenuation with increasing b-value (3). The signal decay is determined by experimental and physiological factors including gas diffusivity, diffusion gradient strengths and timings, and the complexity of alveolar microstructure, which together influence the measure- ment of ADC (4,5). Theoretical diffusion models, such as the cylinder model (CM) (6,7), stretched exponential model (SEM) (8), and q-space analysis (9), have been proposed to model this non-Gaussian diffusion behavior and derive estimates of alveolar length scales (i.e., mor- phometry) from multiple b-value DW-MRI acquisitions. Compressed sensing (CS) has enabled multiple b-value
As an example, the sagittal 2 shape of the spine is thought to be associated with low-grade spondylolisthesis, in which one or more vertebrae are displaced relative to the vertebrae below [Roussouly et al., 2006]. Understanding the shape variation of individual lumbar vertebrae can allow for a better diagnosis of vertebral fractures and other pathologies of the spine [Whitmarsh et al., 2012]. It is hypothesised that the shape of individual vertebrae may contribute to the overall shape of the spine. In constructing a 3D statistical model of the lumbar spine, this could help to explain the source of spinal shape variation between subjects. A fully automated segmentation method would also enable further biomechanical analysis of the lumbar spine by providing a surface model of the lumbar vertebrae and intervertebral discs. A similar approach has been taken in previous studies of bone structures, such as in Bryan et al.  where a 3D statistical model of the whole femur bone was used to study fracture risk. This was carried out using finite element analysis, which enables numerical modelling of biomechanical properties such as stress [Kurtz and Edidin, 2006]. It has been recently noted in the literature that automatic segmentation is one of the major bottlenecks in constructing subject-specific finite element models of the spine [Jones and Wilcox, 2008].
One of preoperative planning and simulation limitations is the difficulty of accurately reproducing the planed and simulated gesture on the real patient. This limitation can be overcome by superimposing preoperative data on the real patient during intervention. However, this superimposition is complex to achieve in practice since it requires the accurate correspondence of reference landmarks between the virtual and the real patient. We have developed a set of tools so as to obtain a reliable result that can be used in clinical routine. Therefore, we propose to offer a view in transparency of the patient by superimposing the 3D virtual patient reconstructed from MRI or CT medical images, on the video image realized during the intervention. In order to retrieve the constraint linked to deformation and movement of organs of the abdominal area due to patient breathing, the medical image and the video image can be realized under general anesthesia with a constant air volume inside lungs. Such a procedure is observed in practice for needle insertion interventions, such as radiofrequency thermal ablation of hepatic tumors. Thanks to these restrictions, abdominal organs have the same position between both acquisitions with a movement observed in vivo of less than 1mm. Registration can thus be limited to a 2D (video) – 3D (modeling) rigid registration of images.
Although multislice multishot imaging of the brain offers flexible contrast, it is vulnerable to motion artifacts 1 . An integrated framework for retrospectively motion corrected reconstruction of multislice multishot MRI in the presence of 3D rigid motion is developed. The method is devised to fuse previous approaches which treat either within-plane 2,3 or through-plane 4 motion in isolation. The technique is applied to newborn brain imaging in natural sleep, where the subjects tend to move sporadically even when deeply settled.
prepared rapid gradient echo (MP-RAGE) sequence were acquired on a 1.5-T Vision scanner (Siemens, Erlangen, Germany) in a single imaging session. MRI acquisition details: repetition time TR=9.7 msec., echo time TE=4.0 msec., flip angle FA=10, inversion time TI=20 msec., delay time TD= 200 msec., 128 sagittal 1.25 mm slices without gaps and pixels resolution of 256x256 (1x1mm). An example of typical images is illustrated in figure 1.
Wilms´ tumors show a displacing growth behavior. The vena cava is frequently compressed by large tumor volumes. Diagnosis is based on morphological ﬁndings such as intra or extrarenal growth, calciﬁca- tions, possible pseudocapsula, structure of the tumor in imaging with or without cysts, relation to retroperitoneal vessels, size of the tumor and complications as tumorthrombus in the inferior vena cava or metastases in other organs. Additional clinical syndrome associations and age of the patients, position, volume are also very important. Every other crite- rion for a better diﬀerentiation of diﬀerent kind of tumors in childhood are of great importance to minimize the number of wrong therapies. Empirically, the tumor shape is visualized by the radiologist, but is not evaluated in a scientiﬁc way in nephroblastomas in comparison to other retroperitoneal tumors. Up till now, the shape of the tumor is only visu- alized in preoperative 3D-visualizations in the single patient for better surgical planning, as it is already shown in G¨ unther et al.(2004).
Prostate cancer diagnosis is fundamentally based on prostate-specific antigen (PSA) screening and transrectal ultrasound (TRUS)-guided prostate biopsy. However, some tumors in the anterior prostate region can be missed because the routine TRUS biopsy is non-targeted and directed toward the peripheral gland (Ahmed, et al. 2009). In this case, an another alternative test is needed to examine patients who have a negative initial biopsy while their PSA is growing. Since MRI is the most precise imaging modality for localization of prostate cancer, MRI-guided prostate biopsy gives the chance of more accurate targeting, and it is performed at the time of biopsy. (Bonekamp, et al., 2011). MRI-guided prostate biopsy contains either using MRI individually or the fusion technology between ultrasound and MRI. Furthermore, a combination of ultrasound- guided and MRI-guided prostate biopsy has been shown to be preferred to standard TRUS biopsy in prostate cancer detection (Pinto, at al., 2011).
The detailed view of the inside the body is acquired by MRI (Magnetic Resonance Imaging) which uses magnetic and radio waves. It is free from the radiations that form damaging (X-rays). The nuclei in the atom of the human are enforced to shift different positions from the rays of MRI equipment. These nuclei when moves back to places they emit the radio waves of themselves to the scanner that image in the computer. These images depend on the potency and location of signal which is incoming to the scanner. MRI scan is created by using the hydrogen atom nuclei as our body mainly consists of water. Water includes atoms of hydrogen. The part which has least hydrogen atoms will appear darker while the tissue that contains more number of hydrogen atoms will appear brighter. Thus MRI images are suitable to discover the tumor exist in the brain and to inspect the spinal cord. Two kinds of images fall MRI scan: T1 and T2. T1 weighted images present admirable detailed anatomic but contrast between ordinary normal tissues and the abnormal tissues are not shown good, while T2 weighted images even though has anatomical detail less but shows contrast between tissues of abnormal and normal excellently.
Samen met het 3DLab van het RadboudUMC heb ik vorm gegeven aan de ontwikkeling van software voor augmented reality visualisatie van 3d anatomische modellen. Ik heb de software van het 3DLab klinisch gebruikt in samenwerking met de chirurgen, feedback verwerkt, ideeën bedacht en dit teruggekoppeld aan het 3DLab. Zo heb ik vorm gegeven aan onze samenwerking. Inmiddels is de software in een dermate vergevorderd stadium dat ik verwacht dat we preopratieve planningen met behulp van augmented reality spoedig standaard in de kliniek kunnen gaan toepassen. Daarnaast heb ik ook de verantwoordelijkheid gekregen en genomen voor het gebruik van de 3D printer. Ik heb daar proactief nieuwe visualisatie technieken mee ontworpen. Deze 3D modellen zijn makkelijk mee te nemen naar de operatie kamer en helpen daardoor bij de intraoperatieve oriëntatie. Als laatste heb ik mijn toegevoegde waarde kunnen laten zien op de afdeling door in nauwe samenwerking met de chirurgie te laten zien wat de toegevoegde waarde van een TGer is. Door techniche ontwikkelingen toe te passen binnen het hele operatieve proces, van beeldvorming tot pathologische evaluatie, en niet alleen binnen het preoperative kader, heb ik mijzelf nuttig gemaakt bij ingewikkelde casuïstiek. Inmiddels word ik regelmatig bij patiënten betrokken en gevraagd om mee te denken en te helpen waar mogelijk.
Limitations exist for this study. Posterior tibial slope was measured on conventional 2D radiographs instead of using a 3D imaging technique, such as CT/MRI. Previous studies have reported that there is a good correlation between X-rays and CT/MRI, with an average error of about 3.4° [22, 27]. Because plain X-rays are cheap and easy to obtain, they are routinely adopted as a first line diagnostic modality in patients undergoing knee imaging for any reasons worldwide. Moreover, plain X-rays are the only imaging modality ordered in clinical practice in high tibial osteotomy or knee arthroplasty. Furthermore, it was deemed that there would be a smaller proportion of pathological X-rays in the databases, compared to MRI that are performed as a second line imaging modality to confirm knee disorders. For these reasons, it was supposed that a true lateral X-ray of the knee (with femoral condyles Table 1 Demographic data in
Nowadays, magnetic resonance cholangiopancreato- graphy (MRCP) is a well-established noninvasive and per se complication-free examination for many pan- creaticobiliary conditions including PD that makes use of heavily T2-weighted pulse sequences to depict the fluid-containing pancreaticobiliary duct system . Three-dimensional MRCP (3D-MRCP) using fast spin echo pulse sequences with (near) isotropic spatial resolution and synchronization to respiration repre- sents the most elaborate MRCP technique currently used in clinical routine . Despite synchronization to respiration, however, prolonged acquisition times of 3D-MRCP may lead to suboptimal image quality due to motion artifacts especially in non-cooperative patients, and limited anatomical coverage may hamper the diagnostic capability of the sequence [6–8].
Background: Hepatocyte-specific gadolinium based contrast agents (HSCA) provide substantial information for the classification of liver lesions in magnetic resonance imaging (MRI). However, breathing artifacts which reduce image quality and diagnostic confidence of hepatobiliary phase acquisitions are regularly observed in clinical routine. The aim of this study was to evaluate two approaches to reduce breathing artifacts for hepatobiliary phase imaging. Methods: Twenty minutes after administration of a HSCA (gadoxetic acid), a T1-weighted VIBE sequence with radial k-space sampling (radialVIBE, 180 s acquisition time in free breathing) and a highly accelerated Cartesian VIBE with Dixon fat separation (CD-VIBE, CAIPIRINHA acceleration with r = 2 × 2, breath-hold 8 – 10 s) were acquired in 35 patients (12 female, 57 ± 13 years), who showed breath-holding difficulties in early phases of the examinations. Image quality (image sharpness, noise, artifacts, homogeneity of fat saturation, bile duct delineation and overall image quality) as well as conspicuity and liver-to-lesion signal intensity (SI) ratios of focal liver lesions were assessed for both radial- and CD-VIBE.
Immobilisation for patients undergoing brain or head and neck radiotherapy is achieved using perspex or thermoplastic devices that require direct moulding to patient anatomy. The mould room visit can be distressing for patients and the shells do not always fit perfectly. In addition the mould room process can be time consuming. With recent developments in 3D printing technologies comes the potential to generate a treatment shell directly from a computer model of a patient. Typically, a patient requiring radiotherapy treatment will have had a CT scan and if a computer model of a shell could be obtained directly from the CT data it would reduce patient distress, reduce visits, obtain a close fitting shell and possibly enable the patient to start their radiotherapy treatment more quickly. This paper focusses on the first stage of generating the front part of the shell and investigates the dosimetric properties of the materials to show the feasibility of 3D printer materials for the production of a radiotherapy treatment shell. The majority of the possible candidate 3D printing materials tested result in very similar attenuation of a therapeutic RT beam as the Orfit soft-drape masks currently in use in many UK radiotherapy centres. The costs involved in 3d printing are reducing and the applications to medicine are becoming more widely adopted. In this paper we show that 3D printing of bespoke radiotherapy masks is feasible and warrants further investigation.