Post-processed images of a breast tumor obtained with CADstreamTM software for breast MRI from ConfirmaTM, Kirkland, Wash. (Click the image for a larger version.)

Advances in hardware for MRI over the past several years have been frequent and far reaching. These include routine use of 3D imaging methods, diffusion and perfusion imaging, and cardiac MRI. The direct impact on the interpreting radiologist is equally dramatic, with an extremely large number of images, frequently more than 1,000, that must be interpreted. Integrating information from different pulse sequences as well as from 3D imaging is challenging and time-consuming.

To handle the increasing amounts of complex MRI data, an increasing number of software tools are available. Whereas previously these software tools were often in the hands of the specialist, there is a growing trend toward software tools that are more readily available. The best software innovations are those with simple user interfaces that can be mastered within minutes and that can be applied for diagnosis in an equally short period of time. This review focuses on several newer examples of software innovations for MRI interpretation.


Robson Macedo, MD

Since 1999, there has been a 40% increase per year in the number of breast MR studies performed in the United States. In addition, more than 1,200 sites in the United States have purchased surface coils for use in breast MR. This number is expected to grow to more than 2,000 sites by the end of 2007.1 It is well accepted that MR sensitivity for invasive breast cancers is near 100%, but as the use of breast MRI increases, radiologists interpreting breast MR are challenged to achieve high specificity while retaining high sensitivity. Reading the large number of acquired MR images in a reasonable amount of time also becomes more important as the number of studies increases. Breast magnetic resonance acquisition and image interpretation techniques have been refined through clinical optimization. The number of images to interpret, however, has increased to hundreds per case. Computer-aided detection (CAD) algorithms have allowed radiologists to regain efficiency while maintaining optimized acquisition techniques. The first CAD system for breast MR was launched in 2003. The CAD installed base has since grown to more than 150 systems in the United States.1 The primary reason for this quick adoption of CAD for breast MR is that the CAD software enables readers to increase their efficiency while potentially improving their overall accuracy. The full benefits of CAD for breast MR are realized when the interpreting radiologist has a thorough understanding of the algorithms used, and the limitations of CAD.

The latest software for breast MRI uses a registration algorithm to reduce motion artifacts and incorporates a cardiac artifact detection algorithm. Cardiac artifact detection ensures optimized registration by removing cardiac artifact from the registration process. Registration is automated and has been demonstrated to reduce artifact in subtraction images. These new software products also have analysis tools such as multiplanar reformatting, maximum intensity projections (MIPs), and 3D renderings with data calculation of ROIs. These tools automate the processing of breast MRI studies, eliminating technologist time and significantly reducing interpretation time. They standardize the processing and reporting via the BI-RADS® Atlas for breast MRI and are accessible throughout a network, providing access to studies anywhere.

The number of MR-guided breast interventions is rapidly growing as MR-compatible systems have become available and techniques have evolved.1 There are also products available which have, in addition, an interventional guidance tool that helps to calculate coordinates for MR-guided interventions at the point of procedure. These interventional guidance tools report needle position (insertion location, depth, and needle angle) for both grid and pillar methods in real time and display images and needle position in the patient’s orientation. Those products are compatible with interventional breast coils and vacuum-assisted biopsy systems, and are available on any networked PC at locations throughout the hospital or imaging center.


Cardiac MR images showing automatic contour detection for cardiac function analysis. (Click the image for a larger version.)

Magnetic resonance applied to the cardiovascular system has been termed cardiovascular magnetic resonance, or CMR, by the international scientific community. CMR has grown considerably in recent years and is now firmly established in clinical and research cardiovascular medicine in the larger centers. This growth and acceptance stem from a number of factors, including technical advancements (speed, reliability, ease of use, new applications), superb image quality and field of view, and the reporting of CMR-derived new insights into entrenched problem areas in cardiology. Cardiac MR is well known for its excellent diagnostic capabilities. CMR is also thought to be a complicated, time-consuming technique reserved for experts and research centers. Over the past few years, the improvements in MRI hardware and software have resulted in less complicated imaging protocols and reduction of the imaging time, thereby allowing the assessment of multiple aspects of cardiac anatomy and function. The examination time can possibly be further reduced by the use of automated scan planning methods utilizing 3D models2 or the use of real-time interactive scanning.3

Contour detection and quantification of myocardial infarct, both using QMASS® MR from Medis Medical Imaging, Leesburg, Va

Hardware and software developments, including 2D perfusion, high resolution, and full cardiac coverage even at high heart rates under pharmacological stress,4 cardiac function in 3D,5 and whole heart 3D coronary tree with free-breathing prospective acquisition correction techniques without contrast agent, allow new ways to perform CMR in a CT-like way: easy and fast.

There are also new techniques that permit more precise electrophysiology (EP) ablation planning, providing high-resolution data in 3D without the use of contrast agents for anatomical guidance and fusion with EP maps.6

Cardiac MR images obtained with QMASS® MR software from Medis Medical Imaging, Leesburg, Va, provides a 3D function analysis.

Software for post-processing cardiac studies is available for measuring regional function parameters such as wall motion, thickening, and perfusion at rest and stress. In order for MRI to become a valuable and routinely applicable imaging modality, the time required for quantification of the many images should be reduced considerably. New software is commercially available with analytical methods for left ventricular function, and vascular flow measurements based on automated contour detection approaches have been described.7 Validation studies of these quantification methods8 have confirmed their accuracy, precision, robustness, and applicability for clinical research studies. Those new available software programs also have features for analyzing stress function studies by visually comparing wall motion and perfusion studies, for analyzing myocardial time intensity during first pass perfusion,9 and for analyzing myocardial viability using algorithms for semiautomatic and automatic delineation of infarct transmurality.10 These developments represent significant steps toward the so-called ‘‘one-stop shop” approach. However, more research and developmental work needs to be carried out for other types of MR acquisitions.

The software tools required for post-processing of magnetic resonance angiograms include the following functions: data handling, image visualization, and vascular analysis. New software combines the most commonly used three-dimensional visualization techniques with image processing methods for analysis of vascular morphology on MR angiograms. The main contributions of this new software are (a) implementation of fast methods for stenosis quantification on 3D MR angiograms on a personal computer-based system; and (b) portability to the most widespread platforms. Quantification is performed in three steps: extraction of the vessel centerline, detection of vessel boundaries in planes locally orthogonal to the centerline, and calculation of stenosis parameters on the basis of the resulting contours. Qualitative results from application of the method to data from patients showed that the vessel centerline correctly tracked the vessel path and that contours were correctly estimated.

Other software developments are now making possible the use of automatic methods for segmentation and quantification of vessels from 3D contrast-enhanced (CE) MRA data with minimal user interaction, improving the accuracy of vessel visualization and quantification of stenosis.11,12

Velocity encoded cine-MR (VEC-MR) sequences have been available as MR-software packages over the last decade. VEC-MR has been demonstrated to be a viable tool in the assessment of velocities and flow volumes in a variety of vessels of the human cardiovascular system.13,14 The postacquisition evaluation of such measurements is done routinely either directly on the MR console or on a remote workstation by manual determination of vessel edges on magnitude images on each time frame. When transferring those regions of interest to the corresponding phase images, mean spatial velocities can be calculated. Using these values, one is able to calculate instantaneous flow per phase, flow per cardiac cycle, and thus flow volumes in mL/min. The routinely performed manual vessel edge detection is a rather time-consuming procedure taking between 10 and 20 minutes on average, depending on the software package. For that reason, automated edge detection softwares have been developed and experiences have been reported, demonstrating good correlation between manual and automated vessel edge detection for VEC-MR flow measurements of the ascending and descending aorta.


David Bluemke, MD, PhD

Recent advances in brain imaging with MRI are the underpinnings for the care of patients with stroke. In particular, the introduction of diffusion-weighted (DWI) and perfusion imaging (PI) have raised hopes of greater ability to discriminate necrotic from salvageable tissue and thus to target new treatments.

Diffusion-weighted imaging and perfusion-weighted neuro MR imaging demand accurate interpretation to determine the severity of the stroke and predict recovery.15 Apparent diffusion coefficient (ADC) maps permit improved estimation of the age of stroke lesions, enabling a better characterization of events. New software is now available with automated image processing in real-time data, including multiple parametric calculations, ADC maps, and neuro perfusion. These features provide the reviewing physician with the benefit of instantly viewing the calculated results directly while scanning the patient. This significantly enhances the diagnostic process with accelerated speed and more reliable diagnosis.

Diffusion tensor image of the brain shows a 3D analysis with color mapping

Another developing technique is susceptibility-weighted imaging (SWI), which uses both magnitude and phase images from a high-resolution, three-dimensional, fully velocity compensated gradient-echo sequence. New softwares for post-processing are applied to the magnitude image by means of a phase mask to increase the conspicuity of the veins and other sources of susceptibility effects.16 SWI seems to be extremely sensitive to deoxygenated blood present in blood products, or in the vascular system. This technique can also be a powerful tool for diagnosis of vascular malformations, stroke, traumatic brain injury, tumors, and neurodegenerative diseases.

Brain activity can be detected via the relatively indirect coupling of neural activity to cerebral blood flow. MRI assessment of cerebral flow is possible with software allowing continuous collection of successive images at a rapid rate. Statistical processing of these MRI time series information produces tomographic maps of brain activity in real time, with updates of 10 frames/s or better.17

A 3D surface of the brain

Another developing field in software is in the field of neurointerventional MRI. Precise resection of tumor and protection of healthy surrounding tissues are critical for long-term patient outcomes in neurosurgery. MRI, with its excellent soft tissue contrast, is the ideal imaging modality for guiding neurosurgical procedures.18 Dedicated real-time protocols and automatic image display are allowing “MR fluoroscopy.”19 Interactive real-time control of the scan plane and interactive graphical slice positioning are now available with a multimodality approach using workstations for image fusion.


In traditional 2D body MRI, the signal change of small lesions can remain undetected when the average slice thickness is between 3 and 6 mm. But 3D body MRI allows detection of much smaller lesions. 3D body MRI is time-consuming, but can be combined with parallel imaging to permit rapid, high-resolution 3D image acquisition.20 With isotropic resolution, these techniques can better depict the complex relationship between anatomical structures and pathological changes.

Diffusion imaging has been a powerful tool for brain MRI for quite some time. Diffusion imaging may also provide improved diagnoses throughout the body. New software using diffusion weighted imaging enables detection of lesions in a specific local area or even over the entire body.21

Diffusion weighted imaging tools based on echo planar imaging can be combined with the free-breathing prospective acquisition correction techniques. As the diffusion coefficient of metastases and primary tumors changes compared to normal tissue, the specificity of diagnosis is improved.22

A relatively new technique is MR colonography. Compared to barium enema studies and colonoscopy, MR colonography has a short history and is still being developed. MR colonography was described in 1997 by Luboldt et al.23 It is a technique that generates up to 700 images with relatively high spatial resolution in any desired plane. To efficiently read these images, post-processing on a workstation is necessary. Such workstations should be able to handle the data quickly and therefore should have adequate hardware and software to allow fast interaction with the data set.

Currently, post-processing for MR colonography images is commercially available. These include two-dimensional MPR, volume rendering techniques such as virtual colonoscopy and tissue transition projection, and older 3D rendering techniques such as shaded surface display and MIP. The difference between 3D rendering techniques and volume rendering is that in the latter the entire data set is used for the 3D image, while with the 3D rendering techniques only about 10% of the data set is used for rendering.24

Another promising field for MRI is in prostate cancer diagnosis. While trans-rectal ultrasound-guided biopsy is a useful and established method, guidance is limited by a low sensitivity of 60%, with only 25% positive predictive value.25 More than 20% of the cancers studied required more than one biopsy session to reach a diagnosis. New surgical navigation softwares for prostate MR are available to localize tumors and to place the needles into focal lesions with direct MR imaging guidance,26 decreasing the need of repeated biopsies.

Robson Macedo, MD, is a research fellow and David Bluemke, MD, PhD, is associate professor of radiology and medicine and clinical director of MRI at The Russell H. Morgan Department of Radiology and Radiological Sciences, The Johns Hopkins University School of Medicine, Baltimore.


  1. Wood C. Computer Aided Detection (CAD) for breast MRI. Technol Cancer Res Treat. 2005;4:49-53.
  2. Danilouchkine MG, van der Geest RJ, Westenberg JJ, Lelieveldt BP, Reiber JH. Influence of positional and angular variation of automatically planned short-axis stacks on quantification of left ventricular dimensions and function with cardiovascular magnetic resonance. J Magn Reson Imaging. 2005;22:754-64.
  3. Hardy CJ, Darrow RD, Pauly JM, et al. Interactive coronary MRI. Magn Reson Med. 1998;40:105-11.
  4. Takase B, Nagata M, Kihara T, et al. Whole-heart dipyridamole stress first-pass myocardial perfusion MRI for the detection of coronary artery disease. Jpn Heart J. 2004;45:475-86.
  5. Sorensen TS, Beerbaum P, Korperich H, Pedersen EM. Three-dimensional, isotropic MRI: a unified approach to quantification and visualization in congenital heart disease. Int J Cardiovasc Imaging. 2005;21:283-92.
  6. Shpun S, Gepstein L, Hayam G, Ben-Haim SA. Guidance of radiofrequency endocardial ablation with real-time three-dimensional magnetic navigation system. Circulation. 1997;96:2016-21.
  7. van der Geest RJ, Reiber JH. Quantification in cardiac MRI. J Magn Reson Imaging. 1999;10:602-8.
  8. van der Geest RJ, Lelieveldt BP, Reiber JH. Quantification of global and regional ventricular function in cardiac magnetic resonance imaging. Top Magn Reson Imaging. 2000;11:348-58.
  9. Nagel E, Klein C, Paetsch I, et al. Magnetic resonance perfusion measurements for the noninvasive detection of coronary artery disease. Circulation. 2003;108:432-7.
  10. Schuijf JD, Kaandorp TA, Lamb HJ, et al. Quantification of myocardial infarct size and transmurality by contrast-enhanced magnetic resonance imaging in men. Am J Cardiol. 2004;94:284-8.
  11. van Bemmel CM, Viergever MA, Niessen WJ. Semiautomatic segmentation and stenosis quantification of 3D contrast-enhanced MR angiograms of the internal carotid artery. Magn Reson Med. 2004;51:753-60.
  12. de Koning PJ, Schaap JA, Janssen JP, Westenberg JJ, van der Geest RJ, Reiber JH. Automated segmentation and analysis of vascular structures in magnetic resonance angiographic images. Magn Reson Med. 2003;50:1189-98.
  13. Kilner PJ, Firmin DN, Rees RS, et al. Valve and great vessel stenosis: assessment with MR jet velocity mapping. Radiology. 1991;178:229-35.
  14. Mostbeck GH, Caputo GR, Higgins CB. MR measurement of blood flow in the cardiovascular system. AJR Am J Roentgenol. 1992;159:453-61.
  15. Davis SM, Donnan GA, Butcher KS, Parsons M. Selection of thrombolytic therapy beyond 3 h using magnetic resonance imaging. Curr Opin Neurol. 2005; 18:47-52.
  16. Sehgal V, Delproposto Z, Haacke EM, et al. Clinical applications of neuroimaging with susceptibility-weighted imaging. J Magn Reson Imaging. 2005;22:439-50.
  17. Cohen MS. Real-time functional magnetic resonance imaging. Methods. 2001;25:201-20.
  18. Hall WA, Galicich W, Bergman T, Truwit CL. 3-Tesla intraoperative MR imaging for neurosurgery. J Neurooncol. November 29, 2005:1-7. Epub ahead of print.
  19. Derakhshan JJ, Duerk JL. Update to pulse sequences for interventional MR imaging. Magn Reson Imaging Clin N Am. 2005;13:415-29.
  20. Naganawa S, Kawai H, Fukatsu H, et al. High-speed imaging at 3 Tesla: a technical and clinical review with an emphasis on whole-brain 3D imaging. Magn Reson Med Sci. 2004;3:177-87.
  21. Ballon D, Watts R, Dyke JP, et al. Imaging therapeutic response in human bone marrow using rapid whole-body MRI. Magn Reson Med. 2004;52:1234-8.
  22. Spuentrup E, Buecker A, Adam G, van Vaals JJ, Guenther RW. Diffusion-weighted MR imaging for differentiation of benign fracture edema and tumor infiltration of the vertebral body. AJR Am J Roentgenol. 2001;176:351-8.
  23. Luboldt W, Bauerfeind P, Steiner P, Fried M, Krestin GP, Debatin JF. Preliminary assessment of three-dimensional magnetic resonance imaging for various colonic disorders. Lancet. 1997;349:1288-91.
  24. Calhoun PS, Kuszyk BS, Heath DG, Carley JC, Fishman EK. Three-dimensional volume rendering of spiral CT data: theory and method. Radiographics. 1999;19:745-64.
  25. Terris MK. Sensitivity and specificity of sextant biopsies in the detection of prostate cancer: preliminary report. Urology. 1999;54:486-9.
  26. Hata N, Jinzaki M, Kacher D, et al. MR imaging-guided prostate biopsy with surgical navigation software: device validation and feasibility. Radiology. 2001;220:263-8.