Accelerated data acquisition in Magnetic Resonance Imaging

Science and technology have provided the health-care professional of the 21st century with an enormous range of tools. Every facet of modern medical practice, be it diagnosing, treatment, preventive medicine, epidemiology, management or palliative care, has been empowered and enriched by the advent of tools that would have seemed to belong to the realm of science fiction even a few years ago.

This is, perhaps, truer in the case of medical imaging than any other aspect of medical science. There are imaging tools available today that can help a doctor to peer into the subtlest workings of a patient’s body without physically invading it by even so much as poking a needle in. These tools have greatly reduced the possibility of error in diagnosis. As a result, medical imaging has become an integral and indispensable part of modern medicine.

However, impressive as it appears to the layman, the situation is far from perfect; every wonder of technology available today represents a compromise. There are great benefits, but there are also pay-offs. For example, was the earliest form of medical imaging employed X-rays for radiography. Subsequently it was discovered that X-rays were actually hazardous, especially with continued and prolonged exposure which could cause cancer. However, X-ray radiography is still used because it is relatively inexpensive, in spite of more advanced and safer technology being available.

Magnetic Resonance Imaging or MRI is a relatively safer alternative for medical imaging. It is preferred because it has several advantages. For one, there is no harmful radiation. For another, it offers good contrasts resulting in clear images. Moreover, it can provide images at practically any depth within the human body. Perhaps more importantly, it provide images of an organ from any angle required and also enables “slicing”, that is, viewing it layer by layer.

However, MRI also has its drawbacks. Capturing images using MRI is a long and slow process. This results in discomfort for the patient. It also restricts the number of people who can be scanned per day, so that each patient has to pay more for a scan to make the facility viable. Moreover, when trying to create images of organs such as the heart and lungs which are constantly in motion, this forces a trade-off, so that image clarity is sacrificed for the sake of adequately accurate capture of movement.

There is a lot of research and development being carried out all over the world to find ways to accelerate the scanning process in MRI. The techniques employed for speeding up MRI scanning at the moment falls into two categories:

  • Parallel Imaging Techniques.

    Here, the image is captured by multiple sensors placed at different points around the patient. This is somewhat like the patient being observed by different people at the same time. The total volume of information to be acquired for creating an accurate image is divided up between these different sensors, so that it takes less time. The information from the different sensors is combined along with information about the location of each sensor to create a composite image of the required accuracy.

  • Compressive Sensing Techniques

    Image acquiring systems usually capture large volumes of information. Most of this information is then discarded for the purposes of transmission and storage. For example, a 10 megapixel camera captures 10 MB of data with every picture taken. 95% of this is usually discarded at the time of compressing. Compression involves discarding irrelevant data and retaining only the relevant. This begs the question: Why acquire data that is irrelevant, if it is only to be discarded? Therefore, to alleviate the problem of acquiring unnecessary data, the technique of compressive sensing creates a theoretical framework to acquire data in compressed form itself. In MRI compressive sensing theory can be used to acquire lower volumes of data and reconstruct the desired image by exploiting the inherent compressibility of medical MR images.

Since each of these techniques yield considerable gains in the speed of image acquisition, it is only logical to explore the possibility of combining them for even greater gains. At the IIT-B Monash Research Academy, research scholar Kamlesh Pawar is doing exactly that.

Working under the guidance of Prof. Arjun Arunachalam and Prof. Jingxin Zhang, Kamlesh has been working on designing novel data acquisition technique and reconstruction algorithms to address this problem of slow data acquisition in MRI.

Kamlesh’s work is aimed at developing methods to combine the technique of parallel imaging and compressed sensing to accelerate MRI scans. He has been able to develop a new method that acquires the data in the noiselet domain instead of the conventional Fourier domain. This new data acquisition technique has been combined with compressed sensing and parallel imaging technique to reduce the scan time by a factor of 8.

He has succeeded in implementing this new data acquisition technique on an MR scanner for imaging the human brain. This has resulted in a reduction in the scan time by the factor of 8.

Another technique that has been developed as a part of Kamlesh’s research is called k-space aliasing, which reduces the scan time for dynamic imaging. In this technique multiples lines of the raw data are overlapped and acquired simultaneously. The individual lines are resolved during the post processing to reconstruct the final image. This technique has also been implemented on an MR scanner and combined with parallel imaging to accelerate the dynamic MR scans by a factor of 9.

Says Kamlesh, “This research is quite challenging and exciting as it is a multidisciplinary field, requiring knowledge of multiple domains such as signal processing, radio frequency pulse design, medicine, physiology and psychology.” It gave him a chance to work on state of the art MR scanners at Monash University. Kamlesh even succeeded in programming his own RF pulse sequences on the Siemens scanner.

Kamlesh’s research is important because MRI has no known harmful radiation effects like computed Tomography. The only problem holding back MRI is its slow data acquisition speed. If this problem can be fully solved, it would create a fast, efficient imaging modality free from any harmful radiation.

Some of the benefits accruing as a result of Kamlesh’s work are:

  • Breath hold time for cardiac cine imaging can be reduced/eliminated, resulting in reduced patient discomfort.
  • The number of patients that can use the facility per day will be increase and as a result, cost per patient will come down.

Practical applications for the new techniques include:

  • If MRI can be fast, internal organs of human body can be seen in real time while the subject is performing some task such as exercise. This can provide valuable information on how human body really works and will provide valuable information in training sports persons.
  • Faster imaging scan in MRI will facilitate the creation a movie of a working human heart with high spatial and temporal resolution. These time images will provide valuable information for diagnosis of the abnormalities present in patients’ hearts.
  • This research will provide new fast imaging techniques that will equip the researcher in neuroscience with a new tool to further understand the complex human brain and also benefit the clinicians to diagnose abnormalities more accurately and quickly.

Research scholar: Kamlesh Pawar, IITB-Monash Research Academy

Project title: Accelerated data acquisition in Magnetic Resonance Imaging

Supervisors: Prof. Arjun Arunachalam, Prof. Jingxin Zhang

Contact details:

Contact research@ for more information on this, and other projects.