Custom-Made Healthcare AI

What we do:

Theta Tech AI is a firm of biomedical engineer coders, consultants, machine learning experts, software developers and data scientists specializing in bringing cutting edge artificial intelligence to our healthcare clients.

We train custom neural network models, design data-driven algorithms, create stunning dashboards, architect scalable cloud infrastructure, build intuitive webapps, and more for our healthcare clients.

Our philosophy:

In healthcare AI, deeply understanding both the clinical problem and the biomedical data is at the heart of any project we undertake.

The future of medicine is enhanced by AI, with doctors having access to a suite of AI systems to aid in their decision-making and provide better patient care.

It is therefore more important than ever to make sure the "healthcare" side gets as much attention as the "AI" side.

We at Theta Tech AI believe that both the analyzing the available medical data and understanding the clinical problem are just as important as choosing the right neural network architecture.

How are we different?

We are not AI generalists but rather AI biomedical engineers. We build and train custom healthcare algorithms and intuitive dashboards for our clients, from university research labs to healthcare startups to large medical device corporations.

We pride ourselves in always staying on the cutting edge - from 3D medical image segmentation transformer neural networks to biomedical signal analysis, we have hands-on experience with many types of healthcare data.


The Future of Medicine is AI-Enhanced

Modern Deep Learning

We have extensive experience training cutting edge biomedical models on our cloud GPU cluster. Whether using Transformers or Convolution Neural Network architectures, whether 1D, 2D, 3D, 4D, multimodal, or tabular datasets, we have you covered.

See our Case Study example for Artificial Intelligence and Deep Learning

Clinically Driven Algorithm Design

We always start any project with the available data and the clinical problem at hand. Every step of the way, we keep the clinical goal in mind and consider how to best enhance downstream patient care with AI. A proper cross validation setup, generalizability testing, and reproducibility are important factors in good clinically-driven algorithm design.

See our Case Study example for Biostatistics

Data Analysis & Visualization

A crucial (and oft-overlooked) component to proper data analysis is having the right visualizations. Whether that entails viewing violin plots representing biostatistics, encoding millions of healthcare database records into large latent spaces for easy navigation and cluster analysis, or 3D viewing of volumetric datasets, we believe that the more clearly you can see and understand both the data and the AI results, the better the downstream AI.

Experts in Medical Datatypes

From 4D radiology DICOM datasets to heart beat signals, from multi-resolution H&E stained histopathology data to messy clinical trial data, we've seen it all. Our team of biomedical engineers has extensive experience in data pipelines. As the old coding adage goes: "garbage in, garbage out" and we understand that data preprocessing pipelines are often as important, if not more important, as the neural network architecture chosen.

Case Studies

Previous projects delivered to clients

About the Founder

Dr. Robert Toth, founded Theta Tech in 2013. His Ph.D. was in Biomedical Engineering, with a focus on radiology AI. He has published over 25 peer-reviewed papers on artificial intelligence for medical image analysis and has co-invented several medical AI patents, since 2007.

Notable Patents

  • Image Similarity-Based Finite Element Model Registration” by Robert Toth, Anant Madabhushi, United States US 14/225,850, Mar, 2014. Link
  • Computer Assisted Diagnosis (CAD) of Cancer Using Multi-Functional in vivo Magnetic Resonance Spectroscopy (MRS) and Imaging (MRI)” by Anant Madabhushi, Satish Viswanath, Pallavi Tiwari, Robert Toth, Mark Rosen, John Tomaszeweski, Michael Feldman, PCT/US08/81656, Oct. 2008. Link

Notable Publications

  • Toth, R., et al. “A Multi-Attribute, Non-Initializing, Texture Reconstruction Based Active Shape Model (MANTRA),” MICCAI 5241, pp. 653-661, 2008. doi: 10.1007/978-3-540-85988-8_78
  • Toth, R., Kurhanewicz, J., Madabhushi, A. “Registration of Pre and Post Intensity Modulated Radiation Therapy Prostate MRI for Quantification of MR Imaging Marker Changes and Precise Local Prostate Deformations” ISMRM 2013.

See a complete list of Dr. Toth's patents, theses, and publications here.