AI-Based Screening Solutions for COVID-19 Using Radiology Images

Rohit Lokwani

The world is facing an unprecedented crisis due to the COVID-19 pandemic. Healthcare professionals are trying their best to contain it. At DeepTek, we have a team of expert radiologists who guide our data scientists in creating AI-based solutions based on diagnostic procedures in real-world scenarios.

We have developed a COVID-19 X-ray screening solution that can be used by public and private hospitals, healthcare clinics, and even mobile diagnostic vans. The solution can be used with or without Internet access.

We began by developing an AI-based solution for pneumonia detection in chest X-rays. Once our models started detecting pneumonia accurately, we extended our solution to differentiate between COVID-19 and other types of pneumonia.

In our evaluations, the model demonstrated a positive-patient sensitivity of 90% and an accuracy of 96%.

In addition to classifying the input X-ray as positive or negative, the model also highlights the region of interest within the X-ray image that exhibits COVID-19 imaging findings. We are also developing models for CT scans, which can detect imaging characteristics at an earlier stage.

Our solution streamlines screening workflows, reduces the burden on imaging experts, and enables quicker diagnosis. Notifications and alerts via Whatsapp, SMS and email ensures no patients are lost for followup.

We look to continue extending our support to governments and private hospitals, lowering costs, simplifying caregivers workflows, providing accurate and timely diagnosis with the aim of democratising healthcare by making it accessible to all including people living in remote areas where there are no experts.