Unlock entirely novel insights from your radiology images

Generate new insights from your existing radiology workflows through AI-powered radiomics analytics with expert-driven visualization, segmentation, and extraction functionalities in one user-friendly interface.

Explore new horizons in your radiomics research projects

The SOPHiA DDM™ Platform is compatible with any 3D imaging modality, equipment, and cancer type.  Gain novel insights through image visualization and automated processing in a collaborative, multicenter research environment.

Streamlined 3D segmentation

Automatically or semi-automatically segment anatomical areas of interest in a matter of seconds.

Reliable automatic feature extraction

Leverage validated methods for automated extraction of radiomic features.

Collaborative environment

Participate in multicenter studies across our global network, utilizing our multimodal research database.

Easily navigate the path from images to insights

1. Data upload

Import 3D images (e.g., CT, MRI, and PET scans) to a project-specific database and research environment.

2. Data visualization

Visualize images in 3D, from multiple angles at the same time with the user-friendly interface.

3. 3D segmentation

Automatically or semi-automatically segment anatomical areas of interest using our proprietary algorithms.

4. Feature extraction

Confidently extract radiomic features from segments and store in the research environment for further analysis.

5. Analysis

Analyze radiomic features and add data to multimodal cohorts and longitudinal analyses.

Reaching new heights in precision medicine

Metastatic non-small cell lung cancer

A machine learning model was developed for the TRIDENT post hoc analysis of AstraZeneca’s POSEIDON trial to identify multimodal signatures predicting which mNSCLC patients would benefit most from adding tremelimumab to durvalumab + chemotherapy. Radiomics was a key contributor to the model’s predictive power.

Triple-negative breast cancer

A machine learning algorithm applied to baseline multimodal data of a retrospective cohort was developed to predict complete pathological response to neoadjuvant chemotherapy in TNBC. Radiomics was one of the key contributors to the predictive power of the model.

Solid renal tumors

In the UroCCR-75 study, a machine-learning model was developed to differentiate benign from malignant solid renal tumors. The model was based on clinical, radiologic, and radiomics features from pre-operative multiphasic contrast-enhanced CT scans from a national kidney cancer database.

Meningioma

Spatial mechanistic modeling was used to predict growth of benign asymptomatic meningiomas aiding decisions on whether to extend follow-up or treat the tumor. This approach was able to predict changes in tumor volume and shape after only two MRI examinations, outperforming naïve linear regression.

Colorectal liver metastases

A post hoc evaluation of the TRIBE2 study on liver-limited metastatic colorectal cancer found that radiomics may be a valid tool for predicting prognosis in patients receiving first-line treatment, and for stratification according to risk of relapse after curative resection.