Skip to content
View denbonte's full-sized avatar

Highlights

  • Pro

Organizations

@BergmannLab @AIM-Harvard @MHubAI

Block or report denbonte

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this userโ€™s behavior. Learn more about reporting abuse.

Report abuse
denbonte/README.md

Hello There ๐Ÿ‘€

Hello There!

My name is Dennis. I am a passionate researcher, coder and Linux enthusiast working on AI, medical images, and genes ๐Ÿซ€๐Ÿงฌ

I obtained my BSc degree in Electronics and Telecommunication Engineering and my MSc degree (with a focus on Signal Processing) at the University of Brescia, Italy, where I was born and raised ๐Ÿ‡ฎ๐Ÿ‡น

After my Master Degree, I worked in a multidisciplinary group between Glasgow, Scotland ๐Ÿด๓ ง๓ ข๓ ณ๓ ฃ๓ ด๓ ฟ and my hometown to develop AI pipelines for Brain MRI segmentation. Throughout this wonderful experience I was supervised by my colleague (and friend!) @rockNroll87q.

I then pursued my Ph.D. at Maastricht University ๐Ÿ‡ณ๐Ÿ‡ฑ and was a research scholar at the Harvard's AIM Program ๐Ÿ‡บ๐Ÿ‡ธ

During this time, my research focused on AI applied to clinical and medical imaging data (such as CT, MRI, and X-Ray), especially in the fields of radiology, radiation oncology and cardiology. During my Ph.D., I developed a profound interest in open source, open science and reproducible research. This interest eventually led me and a few colleagues sharing the open source values to the development of MHub.ai - a platform that aims to revolutionize the dissemination of AI models in medical imaging by providing meticulously curated, optimized, and portable AI-based imaging pipelines, empowering researchers with reproducible science and cutting-edge advancements in the field.

In keeping with this theme, I was also part of the NCI's Imaging Data Commons team, where I worked on integrating AI-based medical image analysis pipelines into the platform and harnessing the power of the cloud for reproducible research.


๐ŸŒฑ What Do I Do

I am currently a Postdoctoral researcher at the Computational Biology Department of the University of Lausanne, where I work in a multidisciplinary team on projects that span genetics, computational science, and clinical practice. My research focuses on enhancing disease risk prediction by linking genetic profiles with vascular traits derived from retinal imaging, helping the diagnosis and management of vascular diseases.

On my GitHub you can also find random repositories forked or created typically while trying to learn something off-work.


๐Ÿ“ซ You Can Reach Me at

  • ๐Ÿ“ง dennis.bontempi [at] unil [dot] ch
  • ๐ŸŒ denbonte [dot] me

Pinned Loading

  1. CER3BRUM CER3BRUM Public

    A fast and fully-volumetric Convolutional Encoder-decodeR for weakly-supervised sEgmentation of BRain strUctures from out-of-the-scanner MRI

    Jupyter Notebook 5 5

  2. ImagingDataCommons/IDC-Tutorials ImagingDataCommons/IDC-Tutorials Public

    Self-guided notebook tutorials to help get started with using IDC

    Jupyter Notebook 28 14

  3. MHubAI/models MHubAI/models Public

    Stores the MHub models dockerfiles and scripts.

    Python 8 16

  4. AIM-Harvard/DeepCAC AIM-Harvard/DeepCAC Public

    Fully automatic coronary calcium risk assessment using Deep Learning.

    Python 37 28

  5. ImagingDataCommons/pyplastimatch ImagingDataCommons/pyplastimatch Public

    Python wrapper for Plastimatch.

    Jupyter Notebook 11 4

  6. AIM-Harvard/CXR-Lung-Risk AIM-Harvard/CXR-Lung-Risk Public

    Deep learning to estimate lung-related mortality from chest radiographs.

    Jupyter Notebook 8 4