Medical Imaging AI in Brain Tumor

A brain tumor is an unfortunate life-changing diagnosis for a patient and everyone involved in the care of the patient. The journey with a brain tumor starts with an imaging study, more often an MRI study of the brain. At every stage in the treatment of the condition, an MRI study is the pivot for any decision. The recent stupendous progress of deep learning-based AI technologies is helping radiologists worldwide to analyze the images of patients with brain tumors more accurately and also measure several imaging markers of the tumor more precisely.

For any cancer, an ideal imaging study should help in the early detection of cancer, precise quantification of the tumor spread, the extent of tumor removal on a post-operative scan, differentiation of the effects of radiation therapy from a tumor recurrence, and early detection of tumor recurrence in follow up studies. Brain tumors are no different.

Several technological advancements in the past have helped in each stage of this journey.

Newer risk-free MR contrast agents, advanced MR imaging techniques like multidimensional diffusion (by Random Walk Imaging) sequences, DSC (dynamic susceptibility contrast), DCE (dynamic contrast-enhanced), and ASL (arterial spin labeling) perfusion techniques, and Compilation sequences for faster acquisitions (MAGIC by SyntheticMR) and several iterations of MR Spectroscopy have been helpful in either helping radiologists scan patients faster, see tumors earlier, or measure and monitor them better.

?The new-age AI solutions are helping radiologists to do all of them simultaneously -?scan, see, measure, and monitor – earlier, faster and better.

• Super-resolution algorithms that help scan faster (AIkenist QUICKSCAN, SubtleMR by subtle medical) can be executed to acquire high-quality scans in a much lesser time.

On these images, there are different types of AI solutions that can be executed

• Classification algorithms - that automatically detect cancerous lesions?(VinDr BrainMR)
• Texture analysis solutions - that quantify features related to the heterogeneity of the tissues and obtain texture parameters (Quibim Texture Analysis)
• Tumor auto contouring solutions – that can automatically segment, measure volume, and contour tumor margins for precise quantification (Sens.ai by graylight Imaging, Vbrain by Vysioneer) ?
• MR Perfusion auto quantification solutions – that automatically generate quantitative perfusion parameter maps (IB Neuro by Imaging Biometrics, Cercare Perfusion by Cercare Medical, Brainance MD by Advantis)

The possibility today is that you can orchestrate the execution of all such types of AI solutions together simultaneously in a concatenated fashion on the same study using CARPL.AI.

This helps radiologists to realize the full potential of a wide range of AI applications in MR studies for seeing, measuring, and monitoring brain tumors earlier, clearer, and more precisely.

?Reach out to Dr. Vasanth Venugopal at [email protected] or anyone in our team to know how we can do it at your hospital.

?* Some of the solutions mentioned in this post are yet to be onboarded on CARPL and will soon be, as they are available for clinical use.

-Article by Dr. Vasanth Venugopal (Chief Medical Officer, Radiologist & Clinical Product Lead at CARPL.ai)