Processing millions of medical images in the cloud: DoiT helps QMENTA keep costs under control
Meet QMENTA
When running clinical trials, pharmaceutical companies frequently scan medical images of patients to assess their response to treatment. But with potentially thousands of patients images taken by numerous scanners and hospitals, one trial can easily rack up tens of thousands of images, all of which need to be systematically collated, organized, and analyzed. Workflows become highly complex, leading to inaccuracies and soaring costs.
Enter QMENTA. With its cloud-based system to transfer and store medical images, QMENTA uses AI to anonymize, organize, and standardize the images before running algorithms to extract clinical data from each image to ensure clinicians have a single source of truth to assess images accurately. In this way, QMENTA not only improves the time and cost efficiency of clinical trials, but their accuracy too, with objective quantitative data.
The Challenge
QMENTAโs systems run on Google Cloud. However, focusing on its medical imaging platform makes it hard to keep up with GCP releases and tailor them to QMENTAโs needs. QMENTA needed a cloud partner for expert, actionable guidance on maximizing Google Cloud.
Specifically, QMENTA wanted to ensure cost-efficient use of Google Cloud. Between the cost of storing so many images, and the price of running such compute-intensive AI technology, QMENTA needed to find efficiencies it could pass on to its customers in the form of reduced costs.
โWe are currently in a growth phase, our goal being to scale our number of customers and users exponentially,โ explains Tim Peeters, Director of Engineering at QMENTA. โWe needed a partner that could help ensure that our cloud costs didnโt grow in parallel.โ
Working in a highly regulated healthcare context, QMENTA also needed to strengthen security and compliance posture. QMENTA wanted support to implement industry security standards cost-effectively.
The Solution
Google Cloud recommended DoiTโs services to QMENTA in 2022, and after an initial architecture review, DoiT began working with the QMENTA team to get the most from Google Cloud. Through regular consultations, the QMENTA team shared its specific challenges, with DoiT offering tailored, technical expertise to help improve efficiencies across the company.
โIf we need expertise in a particular area, we could spend a week learning everything ourselves, or we could ask DoiT,โ says Pauo Rodrigues, CTO and Co-Founder of QMENTA. โDoiT has an engineer on hand at all times who in one day can tell us everything we need to know. This allows us to accelerate our business growth, as it means we can focus on our imaging software, which is our core offering.โ
Controlling costs while making the most of the cloud
A key part of DoiTโs consultation focused on FinOps-driven cloud cost optimization. Through a series of FinOps consultations with DoiT Cloud Reliability Engineers (CREs), reviewing its infrastructure, QMENTA received actionable recommendations and guardrails to optimize its cloud usage. Among the advice was a recommendation to use Google Cloud Committed Use Discounts (CUDs), signing up to a set amount of compute and storage at a heavily discounted rate. If inappropriately made, these commitments can lead to companies paying for more usage than they need, which is why DoiTโs careful analysis of QMENTA specific cloud usage was important to ensure QMENTA made an appropriate commitment.
DoiT then worked with QMENTA to develop a commitment automation strategy to continuously secure best-available pricing without manual tracking, thereby saving QMENTA from having to continually search for the best discounts.
QMENTA also benefits from the Cloud Analytics dashboards in the DoiT console, which give the team more visibility and control of its cloud expenditure. โUsing the DoiT dashboards, with its billing reports and aggregated data, allows us to see exactly where we are spending more than we need to, enabling us to dig into the details and make small changes for big savings,โ says Paulo Rodrigues.
Plotting the right path to efficient computing and data storage
With the high demand its AI models place on computing power, QMENTA was keen to ensure it was making the most of Google Kubernetes Engine (GKE), and used the DoiT platform to request a number of Kubernetes training sessions with DoiT engineers. As a result, QMENTA learned how to scale its usage up and down according to need, saving costs by reducing the number of GKE nodes used for tasks that require less memory, while scaling up the number of nodes to add memory to tasks that required higher performance.
As for the large amount of storage required for the millions of medical images in its system, QMENTA worked with DoiT to strategize the best approach to storing images in Google Cloud, balancing accessibility with costs. With so many images, paying for premium storage quickly adds up, so DoiT worked through QMENTAโs options for placing different images in different levels of storage, according to how often they needed to be used. Following this consultation, QMENTA decided to use Google Cloudโs new Autoclass feature in Cloud Storage, which will enable QMENTA to automatically classify every image, helping to reduce costs by ensuring that it pays for the right kind of storage for each image.
QMENTA also knew it wanted to start using an SQL database, but with so many options in the Google Cloud product suite, the QMENTA team didnโt have the time to research each option in detail. โThe DoiT CREs talked us through the pros and cons of each database for our specific needs, and the associated costs,โ says Peeters. โThis will allow us to make an informed decision, instead of spending time investigating all the options.โ
Protecting patient data with hardened cloud security
With so much sensitive patient data in its system, QMENTA is continually looking for ways to strengthen its cloud security to ensure that data remains protected. An identity and access management workshop with DoiT CREs helped to familiarize the QMENTA team with industry best practice, including how to use user groups effectively to streamline access, rather than assigning roles to users directly.
The company also sought recommendations on how to use Google Cloud Intrusion Detection System (IDS) in a cost effective way, opting to use it in a test project first, rather than in production, to ensure that costs didnโt balloon. The company is now working with DoiT to explore the optimal use of Google Cloud Security Command Center.
The Results
Greater expertise for faster decision making
With around-the-clock, unlimited access to DoiT CREs, QMENTA has learned how to make more informed decisions about its cloud usage more quickly, leaving the team more time to focus on growing the business.
DoiT cloud training sessions have also helped to arm QMENTA engineers with the expertise they need to make the most of the cloud. โI often get feedback from our engineers telling me how good the DoiT CREs are,โ says Peeters. โThey understand our questions and can answer them in a very useful, informative way. Our team is happier and more confident as a result.โ
22% reduction in cloud costs with efficient cloud usage
As a result of the consultancy with DoiT and commitment automation, QMENTA reduced Google Cloud expenditure by 22% in nine months.
What's Next?
Having built and validated its product and completed fundraising, QMENTA is now scaling customers and users, which will significantly increase stored and processed images. The cloud optimization journey with DoiT has future-proofed the company for this expansion.
โWe expect QMENTA to grow to two or three times its current size over the next year, which will see our cloud usage increase considerably,โ says Rodrigues. โThrough our work with DoiT, we have future-proofed our growth, ensuring that as we scale exponentially, our cloud spend remains firmly under control.โ