Bio

Publications and Patents

Summary: Technology creation in Medical Imaging

Bio: Karthik Krishnan received the B.E.(Hons) degree in Electrical and Electronics engineering from BITS Pilani, India in 2002. As an undergrad, he pursued his internship at Motorola working on microcontrollers. He obtained and an M.S. degree in Electrical and Computer engineering from University of Arizona, Tucson, USA on a scholarship in 2004. His thesis investigated the use of image compression methods, in particular JPEG2000 for remote volume visualization. During the course of his Masters degree, he spent both summers as an intern at Siemens Corporate Research, Princeton, NJ working in the Imaging and Visualization lab on remote volume visualization.

    From 2004 to 2010, he worked at Kitware, New York and continued to consult with them from Bangalore until 2013. During this time, he was a  developer of the ITK and VTK libraries, widely used in research circles for Medical Image analysis and Visualization. His work involved the development of algorithms for Medical Image Analysis, GPU accelerated rendering collaborating with several academic institutions on numerous research projects and imaging platforms as part of the National Alliance for Medical Image Computing (NA-MIC). He was the principal author of several libraries and modules in both libraries and also of other related libraries, notably: principal author of VTKEdge, a library of high performance GPU accelerated rendering algorithms and editing tools, now subsumed into VTK. He was also the sole author of the Insight LesionSizingToolkit, an open source library for accurate segmentation of segmenting lung nodules, now subsumed into ITK. He authored several segmentation and registration algorithms for medical imaging, algorithms for computational geometry, including a library for Fast Marching on meshes used for geodesic compuations.  His research collaborators, co-authors included the Brigham and Womens Hospital, Stanford University Medical Center, Georgetown University Medical Center, Duke, Univ. Washington, Weill Cornell Medical Center, NYU Medical center, MD Anderson Medical Center and several mid-sized companies in Medical Imaging.

    From 2011 to 2013, he worked at Philips Research in Bangalore developing algorithms for Interventional oncology for a new corporate venture in Prostate Brachytherapy. He developed modules for Dose planning and Visualization for HDR and LDR brachytherapy. He was also the principal author of a digital dentistry Visualization Software written from inception using open source libraries such as VTK and now sold under the brand name Digimodel by OrthoProofUSA.

    From 2013 to  2018, he was employed with Samsung R&D in Bangalore, India, in the Medical Imaging group. His work was in the area of Medical Image Analysis (segmentation, registration, Artificial Intelligence) and Visualization applied to diagnostic imaging (CT, MR, US). He  led the development of clinical applications and algorithms (both for Image Analysis and Advanced Visualization) for CT Colonography, Perfusion, Digital Subtraction Angiography and CT liver segmentation, Interventional Prostate biopsy. He wrote the Advanced Visualization platform from the ground up using open source technologies, MITK, VTK, ITK, Qt and DCMTK. He has contributed to clinical applications and algorithms for musculoskeletal MRI, Cardiac CT. He has developed high quality GPU accelerated real time visualization techniques for endo-luminal navigation in Virtual Colonoscopy, 4D rendering for perfusion and Monte-Carlo cinematic rendering. He has led the development of image analysis algorithms for segmentation and multi-modal fusion of MR onto TRUS for prostate biopsy. He developed AI models for segmentation of several anatomies, including Liver and Prostate, using AI with Decision trees. Some of this work was published in Grand Challenges at MICCAI, with a spot on the leaderboard for the CPU based run-time (a practical consideration for methods deployed on Ultrasound carts). He co-developed anatomical segmentation for MRI using Caffe. Some of this work has been incorporated into FDA approved products and he has been the recipient of 5 technology/innovation awards within Samsung Bangalore, including an Outstanding innovator. 

    In 2018, he was the recipient of the prestigious Zinnov Award, given to 10 professionals once a year from Bangalore, selected from nominations from various Fortune 500 companies in the city. He won the award in the Technical Role Model (Mid Level) category for sustained outstanding contributions from Samsung.

    From 2018 to 2020, he was a Principal Scientist leading AI in Ultrasound at Philips Research, Bangalore. He drove the development of AI algorithms both as an individual and as the head of the team, worked with clinicians to create value propositions. His work on AI in Fetal Ultrasound included AI based real time pose detection, real time anatomical identification, real time object localization. This work covered a range of Real Time and near-RealTime AI networks for Classificaiton, detection, key-point regression, real time pose identification, near real time semantic segmentation from medical images with PyTorch, best plane and view identification, auto bookmarking for improved Ultrasound scanning workflows. He is an inventor on seven global patents with Philips.

    In 2020, he joined Terarecon, as part of their India Management team. As part of a two member Management team and the first two hires, he played an active role in growing the Data science team and led a cross-geographic cross-functional team and played an active role in the growth of the India center from 2 to 40. As Senior Director of AI, he led the development of AI based solutions for Cardiac CT Structural heart.

    He is currently employed as Director of Advanced Visualization at Elucid Bioimaging, Boston, a company developing a novel, non-invasive medical software to comprehensively characterize cardiovascular disease, closing a $27m Series B in Jun 22 and a $80m Series C in Nov 23 financing round. Elucid is a Boston-based medical technology company using uniquely interpretable and validated AI to provide physicians with a cost-effective means to optimize treatment decisions for patients with cardiovascular disease. Advanced clinical insight from Elucid equips physicians with critical information designed to enable precision medicine. Elucid is the first FDA-Cleared and CE-marked non-invasive software to quantify atherosclerotic plaque characteristics compared to histopathology, including Lipid Rich Necrotic Core.

    He is a author of 2 book chapters, 9 journal articles and 30 conference publications/abstracts in Medical Imaging. He has represented multiple organizations at trade shows including RSNA, served as the session chair at IEEE conferences on Medical Imaging and given several invited talks.

    He has a background in AI, Medical Image Analysis, Computational geometry, Visualization, HPC, DICOM. He is adept at several open source libraries such as ITK, VTK, MITK, Slicer, CGAL, OpenCV, Boost, OpenInventor, Caffe, PyTorch. His programming languages of choice are C++ and Python.