Securely Procuring and Annotating Large Volumes of 3D DICOM Medical Data

Disclaimer: Aya Data holds to the confidentiality agreements made with our clients. We shall not disclose any confidential information regarding clients and/or projects, pursuant to the individual agreements made. Consequently, client and project names and any identifiable information may be kept anonymous in our Case Studies.

Overview

Computer vision has already demonstrated its effectiveness in diagnosing medical conditions from a range of medical images, including X-rays, MRIs, ultrasounds, nuclear medicine imaging (PET), optical coherence tomography (OCT), and CT scans. Google DeepMind is already capable of outperforming experts in diagnosing complex eye diseases from OCT scans, for example.

Harnessing AI and ML in medical imaging enables experts to direct their time and skills toward higher-level tasks such as treatment and research, rather than time-consuming diagnostics. Speeding up diagnostic processes helps streamline the entire healthcare system, ensuring patients get timely treatment.

The Client is a European MedTech company who wanted to collect and annotate DICOM format 3D scans of vascular systems. Their data science team had been assigned the task of developing models to understand how vascular systems may be disrupted by invasive surgery.

Data Acquisition

Image segmentation

Industry

Medical/Healthcare

Headquarters

London, UK

Company Size

250+

Challenge

Creating or procuring medical and healthcare datasets is difficult because of patient privacy, anonymity, data protection laws, and internal protocol. Companies must gain full consent for all personal data, and then ensure that it is sufficiently anonymized to be shared.

Since this task also required complex 3D annotations that would require medically trained specialists, the client was also concerned about the cost of finding specialists who could meet the required levels of quality and precision.

Contact us

Solutions

In order to help ethically source the medical data set that the client needed, Aya Data worked with the Department of Radiology at the University of Ghana Medical Centre (UGMC). The resulting anonymized 3D vascular scans went on to augment the Client’s data set and cover key features that had been missing.

Once the data had been sourced, Aya Data’s teams of medical specialists worked on annotating the data using the Client’s proprietary platform.

Results

After several iterations of the workflow, the client felt that the annotations done by the Aya medical team were of the same standard or better than those done by their physicians in Europe, and at a fraction of the cost.

Aya has an ongoing relationship with the Client and continues to perform medical annotation and collection.