DIA Receives $14M to Expand AI Ultrasound Analysis | Israel

DiA Imaging Analysis Secures $14 Million in Series B Funding

DiA Imaging Analysis, an Israel-based AI healthtech firm specializing in the automation of ultrasound scan analysis through deep and machine learning, has successfully completed a $14 million Series B funding round.

Investment Details and Existing Funding

This growth round, occurring three years following DiA’s previous funding, attracted new investors including Alchimia Ventures, Downing Ventures, ICON Fund, Philips, and XTX Ventures. Existing investors, such as CE Ventures, Connecticut Innovations, Defta Partners, Mindset Ventures, and Dr. Shmuel Cabilly, also participated. The company’s total funding now reaches $25 million.

Expansion Plans

The newly acquired financing will facilitate DiA’s continued expansion of its product portfolio. It will also support the development of new and enhanced partnerships with ultrasound vendors, PACS/Healthcare IT companies, resellers, and distributors. Furthermore, the company intends to strengthen its presence across its three key regional markets.

AI-Powered Ultrasound Analysis

DiA provides AI-driven support software to clinicians and healthcare professionals, assisting them in capturing and analyzing ultrasound imagery. Manual analysis of this imagery traditionally relies on human expertise and visual interpretation of scan data. DiA’s technology aims to eliminate subjectivity from these manual estimation processes.

Targeted Clinical Applications

The company has developed AIs capable of automatically identifying key details and abnormalities within ultrasound images. These products cater to diverse clinical needs, with several specifically focused on cardiac analysis. For instance, the software can measure and analyze ejection fraction, right ventricle size and function, and assist in the detection of coronary disease.

Automated Bladder Volume Measurement

DiA also offers a product that utilizes ultrasound data to automate the measurement of bladder volume, streamlining a common clinical procedure.

Mimicking Human Visual Perception

DiA asserts that its AI software replicates the human eye’s ability to detect borders and motion. This is presented as an improvement over subjective human analysis, offering increased speed and efficiency.

Clinician Support and Image Acquisition

“Our software tools are designed to support clinicians in both acquiring optimal images and interpreting ultrasound data,” explains Hila Goldman-Aslan, CEO and co-founder of DiA.

Global Market Presence and Partnerships

DiA’s AI-based analysis is currently utilized in approximately 20 markets, including North America and Europe. In China, a partner has received approval to integrate DiA’s software into their own device. The company employs a go-to-market strategy centered around collaborations with channel partners like GE, Philips, and Konica Minolta, who offer the software as an add-on to their ultrasound or PACS systems.

User Base and Vendor Neutrality

Currently, over 3,000 end-users have access to DiA’s software. Goldman-Aslan emphasizes the vendor-neutral and cross-platform nature of the technology, enabling it to function on any ultrasound device or healthcare IT system.

FDA/CE Approvals and Competitive Advantage

“We have established partnerships with over 10 device and healthcare IT/PACS companies,” she states. “To date, we possess 7 FDA/CE approved solutions for cardiac and abdominal areas, with more in development.”

Data Training and Quality Control

The performance of any AI system is contingent upon the quality of its training data. In healthcare, data efficacy is paramount, as biases in training data can lead to inaccurate diagnoses or risk assessments.

Diverse Data Collection

Goldman-Aslan explained to TechCrunch that DiA has access to hundreds of thousands of ultrasound images through collaborations with numerous medical facilities, allowing for rapid adaptation to new areas of automatic analysis. The company collects data from diverse populations, encompassing various pathologies and utilizing different devices.

Data Integrity and Physician Oversight

“The principle is ‘Garbage in, Garbage out.’ The key is to avoid introducing flawed data,” Goldman-Aslan stated. “Our datasets are meticulously tagged and classified by experienced physicians and technicians.”

Image Rejection System

DiA also employs a robust rejection system that discards incorrectly acquired images, mitigating the subjectivity inherent in data acquisition.

Regulatory Approvals: 510(k) vs. PMA

DiA’s FDA clearances are 510(k) Class II approvals. Goldman-Aslan confirmed that the company has no plans to pursue Premarket Approval (PMA) for its products.

Limitations of the 510(k) Pathway

The 510(k) route is commonly used for medical device approval in the U.S., but it has faced criticism for being a less stringent process compared to the more comprehensive PMA pathway.

Regulation of AI in Healthcare

Regulation of rapidly evolving AI technologies often lags behind their implementation, particularly in healthcare, where significant potential exists alongside inherent risks. This creates a gap between the promises of device manufacturers and the level of regulatory oversight their tools receive.

EU Regulatory Landscape and the AI Act

In the European Union, the CE scheme allows manufacturers to self-declare conformity without independent verification in some cases. However, the EU is developing the Artificial Intelligence Act to introduce additional conformity assessments for “high-risk” AI applications, including healthcare use-cases like DiA’s ultrasound analysis. The implementation of this regulatory regime is still several years away.