OncoDesk is a startup that develops products and services to provide higher quality care for cancer patients while minimizing total treatment cost at the same time. It enables the right diagnosis and treatment for the right patient given at the right time through evidence-based medicine, big 'omics, quantitative imaging and deep learning.
Our solutions rely on the latest advances in oncology, medical image analysis and machine learning. Unlike other similar companies, we speed up the computing experience by developing high-performance computing software for NVIDIA GPUs. Furthermore, our cloud-based platform allows efficiently to capture and integrate daily data generated from hospitals.
Image segmentation using k-means clustering, expectation maximization and normalized cuts.
Multi-atlas segmentation propagation using local appearance-specific atlases and patch-based voxel weighting.
Rigid and deformable multi-level registration of mono- and multi-modality imaging.
Integration of Linear-Quadratic-Linear (LQL) model for predicting radiobiological response and secondary effects.
CUDA implementation of algorithms for super fast performance.
OncoDesk products combines clear design of interfaces with enterprise functionality.
It allows the physicians to create organ templates from treated patients and new patients can then be delineated by automatic local-weighted multi-atlas-based segmentation using the best-matching atlases.
Complete interactive 3D gross human anatomy atlas. Cadaver slices paired with high quality 3D models. Surgery simulator for the purpose of training medical professionals. 3D moving models of muscles and bones.
Computer-aided diagnosis and risk stratification application that characterizes tumor phenotypes by extracting texture, edge and geometry quantitative metrics
Suite of quantitative mathematical models for professionals linked to the specialty of Oncology. It offers advanced interactive modeling techniques.
RadioFeatures characterizes the intra- and inter-tumor phenotypic heterogeneity in radiological images. Millions of texture, edge, and shape statistical image features are extracted. Subsequently, only the most independent informative metrics are selected to train Deep Learning classifiers.
Radiomic analysis has been retrospectively performed on images from a wide range of subtypes of cancers. RadioFeatures may potentially support clinicians in assessing cancer diagnosis, better predict treatment response, and improve assessment of clinical outcome.
Software that generates automatically accurate delineations of organs at risk and target volumes using sophisticated local-weighted multi-atlas-based segmentation and structure learning methods.
Mobile and desktop software that provides more than 13 treatment decision support models (including radiobiology, risk of radiocarcinogenesis, and survival models) based on clinical multiparametric databases for achieving precision oncology.
This software is publicly available for free.