Pune-based medical imaging AI startup, DeepTek.ai, has unveiled an FDA-approved AI solution called Augmento X-Ray, designed to alleviate radiologist workload and enhance the quality of chest X-ray reporting. The tool utilizes deep learning algorithms to automatically detect and highlight abnormalities in chest X-rays, assisting clinicians in making accurate interpretations.
With an estimated 1.5 billion chest X-rays performed annually, there is a critical need for timely and precise interpretations. However, with only 20,000 radiologists in India’s population of 1.4 billion people, there is a shortage of expertise in the country. Augmento X-Ray addresses this issue by reducing the radiologist workload by 30-50%, enabling them to focus on critical cases.
DeepTek’s co-founder, Dr. Amit Kharat, brings over 20 years of radiology experience to the team, which also includes a dedicated tech team and a panel of 70-80 radiologists. The company has a strong global presence, serving over 500 customers in APAC countries and Japan. Recently, DeepTek received FDA approval for Augmento and a chest AI model, opening doors for expansion into the US market.
One of DeepTek’s solutions, Genki, is a public health screening tool that has been employed in India for over four years. It is deployed in vans equipped with portable X-ray machines, enabling swift detection of tuberculosis and chest abnormalities. This tool has significantly enhanced patient detection capabilities, especially in remote areas without immediate access to radiologists.
DeepTek is also partnering with state governments and city bodies in India to contribute to the country’s efforts in combating tuberculosis. The company is collaborating with institutions like the Clinton Health Access Initiative and the Bill & Melinda Gates Foundation to support India’s goal of eradicating tuberculosis by 2025.
DeepTek’s CXR Analyzer is a unique AI model focused on chest X-rays in India. Unlike typical pathology-specific models, it takes an organ-based approach, covering a wide range of abnormalities present in chest X-rays. This comprehensive approach streamlines the identification process for radiologists and improves turnaround time for critical cases.
The company’s proprietary models are developed using Python programming and trained on platforms such as PyTorch and TensorFlow. DeepTek collaborates with NVIDIA, Microsoft Azure, Google Cloud, and AWS for GPU-powered and cloud-agnostic models that meet diverse customer needs.
To validate their models, DeepTek leverages a training dataset comprising millions of radiology scans sourced from their teleradiology services. This dataset undergoes an extensive third-party validation process involving over 3,000 scans reviewed by US Board Certified radiologists. The accuracy of DeepTek’s AI model in detecting and categorizing abnormalities exceeds 95%.
DeepTek’s visionary strategy revolves around advancing two key product lines: Augmento as an AI deployment platform and AI models developed in collaboration with the US FDA. They are simultaneously focusing on expanding their impact in the APAC region, with ongoing pilot initiatives in Singapore and plans to onboard more hospitals in the future.
Through collaborations with renowned entities like Shimadzu, BJC Healthcare, NTT Data, and Shimadzu, DeepTek aims to extend its services to leading hospitals in Thailand, Malaysia, the Philippines, and beyond. The company’s innovative AI solutions have the potential to revolutionize the field of radiology by improving efficiency, accuracy, and accessibility in healthcare diagnostics.
