Join the next MICCAI Industry Talk: May 2, 2023

Thursday 27th April 2023

We look forward to offering our community another exciting and informative webinar. See details below:

Time/Date: May 2, 7am—8am (Pacific Time)/10am-11am (Eastern Time)

Talk Title: Deep learning-based image enhancement, image synthesis, and low-dose imaging.

Registration required: Register (link)

Speaker Bio:
Dr. Enhao Gong is a researcher in the field of AI in medical imaging. He is also the founder and CEO of Subtle Medical, a company that specializes in developing AI technologies to improve medical imaging acquisition, reconstruction, and quantification.

Dr. Gong graduated with Bachelor's degree in Biomedical Engineering from Tsinghua University and earned his Ph.D. in Electrical Engineering from Stanford University, where his research focused on deep learning applications in medical imaging, espeically the upstream challenges of image reconstruction, contrast synthesis, and low-dose imaging. He has received several industry recognition and awards, including Forbes 30-under-30 and RSNA research awards.

He has translated his research into commercial products at Subtle Medical. Subtle Medical has obtained 4 FDA clearances for its AI products and has helped ~500 hospitals and imaging centers worldwide to serve more patients better. The company has received numerous awards and recognition, including the NVIDIA Inception Award, CB Insight AI-100, and Top 150 digital health company awards.

In clinical practice, multi-contrast MRI is extensively utilized, but the availability of each imaging contrast may differ among patients. This presents challenges for radiologists and automated image analysis algorithms. To address this issue, missing data imputation is employed to synthesize absent contrasts from existing ones. This presentation will dive into the latest research on multi-contrast medical image data synthesis and completion using a multi-contrast-multiscale vision transformer. The approach frames data completion as a sequence-to-sequence learning task, designing the model to capture long-range dependencies while demonstrating advantages in performance and interpretability. The model outperforms state-of-the-art methods both quantitatively and qualitatively and is adaptable to accept any subset of input contrasts, synthesizing the missing ones.

Subtle Medical employs deep learning models, GANs, transformers, diffusion models, to enhance medical imaging quality and efficiency, with multiple FDA clearances, CE marks, China NMPA clearances, and more. The talk will also share how research is leveraged to transform innovations into Subtle's scalable product lines for image enhancement in clinical settings, which now power approximately 500 sites worldwide, including hospitals and imaging centers.