Shreyas Vasanawala, MD, PhD

Shreyas Vasanawala, MD, PhD

William R. Brody Professor of Radiology
Division Chief, Pediatric Radiology
Associate Chair, Department of Radiology 
Radiologist-in-Chief, Stanford Childrens

    Clinical Interests

     Pediatric abdominal, cardiovascular, and musculoskeletal imaging.

     I particularly enjoy sports MRI, transplant imaging, and inflammatory bowel disease imaging.

Abdominal and pelvic MRI

Application of fast volumetric imaging to adult and pediatric abdominal tumors, solid organ transplants, renal function, and inflammatory bowel disease

Transplant kidney with multiple arteriovenous fistulae
Pediatric Cardiovascular MRI

MRI of congenital heart disease

3D printing from a cardiac MRI in a child with partial anomalous pulmonary venous return (blue arrows) 3D printing from a cardiac MRI in a child with partial anomalous pulmonary venous return (blue arrows)

 

Pediatric Musculoskeletal MRI

Applications of fast high-resolution volumetric MRI to sports injuries, tumors, and inflammatory arthropathies

Patellar tendon lateral femoral condyle impingement

 

 

 

Research Interests

Novel MRI hardware:

Many applications of body MRI may be improved or enabled by optimization of dedicated radiofrequency receiver arrays. Thus, we are actively developing new approaches to design and construction of miniaturized radiofrequency receiver arrays for optimized and personalized MRI exams.  Work is focused on enhancing flexibility of coils and developing wireless coils.

Researchers: John Pauly, Dwight Nishimura, Michael Lustig, Brian Hargreaves, Dan Ennis, Ali Syed, Ana Aria

Publication: A semiflexible 64-channel receive-only phased array for pediatric body MRI at 3T

Publication: Evaluation of a Flexible 12-Channel Screen-printed Pediatric MRI Coil

 

Fast MRI techniques Fast MRI techniques Coil with elements printed with conductive inks. This approach has enabled highly flexible and lightweight coil fabrication
Prev Next

 

 

Fast and motion-robust MRI techniques

We have focused on developing new ways of accelerating pediatric MRI exams.  These methods include compressive sensing and other model-based image reconstruction techniques. Additionally, we have developed methods of enabling rapid reconstruction of images from highly under sampled, and thus accelerated exams.  Some approaches include novel parallel computing and deep learning methods.

Publication: Improved pediatric MR imaging with compressed sensing

Publication: Practical parallel imaging compressed sensing MRI: Summary of two years of experience in accelerating body MRI of pediatric patients 

Publication: Improving high frequency image features of deep learning reconstructions via k-space refinement with null-space kernel 

Publication: Free-breathing Accelerated Cardiac MRI Using Deep Learning: Validation in Children and Young Adults 

Publication:  Compressed Sensing: From Research to Clinical Practice with Deep Neural Networks 

These methods are naturally suited to high dimensional acquisitions, such as dynamic contrast enhancement and volumetric time resolved phase contrast.

Low rank reconstruction of 4D flow enables a rapid high spatiotemporal resolution acquisition.  This allows a comprehensive scan without anesthesia.

Publication:  Fast pediatric 3D free-breathing abdominal dynamic contrast enhanced MRI with high spatiotemporal resolution 

Model based low rank imaging for dynamic contrast enhancement with high frame rates

Another example of high dimensional imaging is volumetric joint scans in which an additional dimension is the echo time.  We have leveraged this approach to enable rapid walk-in orthopedic MRI. 

Publication: Fast comprehensive single-sequence four-dimensional pediatric knee MRI with T 2 shuffling

 

Model based low rank imaging for dynamic contrast enhancement with high frame rates.

 

A single 5-minute scan can be reformatted into any plane and also generate contrast with various echo times

Much can be gained from non-cartesian k-space acquisitions, which are highly motion robust.  However, the image reconstruction for these exams is quite complex.

Publication: Free-breathing Rāˆ—2R2āˆ— mapping of hepatic iron overload in children using 3D multi-echo UTE cones MRI

A picture containing close Description automatically generated

Volumetric high spatial resolution DCE at high frame raes with stochastic optimization

To manage the complexity of image reconstruction and speed the computation for practical clinical workflow, we have been exploring a range of approaches to deep learning image reconstructions.

Publication: Variable-Density Single-Shot Fast Spin-Echo MRI with Deep Learning Reconstruction by Using Variational Networks

Publication: Data-driven self-calibration and reconstruction for non-cartesian wave-encoded single-shot fast spin echo using deep learning

Publication: Compressed Sensing: From Research to Clinical Practice with Deep Neural Networks

Publication: Analysis of deep complex-valued convolutional neural networks for MRI reconstruction and phase-focused applications

Deep learning image reconstruction enables higher accelerations and faster image reconstructions.  This permits practical free-breathing or single-breathhold short axis stack cardiac imaging for improved cardiac function quantification.

We also have a range of projects on automated image analysis.

Researchers:

Quantitative MRI methods:

Development and validation of accurate and precise cardiovascular flow and function measurements, noninvasive renal function assessment, and tumor therapy response.

 

Researchers: