Enhancing Chest X-Ray Diagnosis with Deep Learning

Challenge Completed!

DR CADx developed a model for 14 chest findings, which has not yet been validated, and Omdena’s team helped to expand it to 17 findings by adding COVID, TB screening, and rib fractures to the algorithm.

The impact

Dr CADx’s AI solution helps minimize missed radiological abnormalities, which are common due to certain areas on chest X-rays where lesions can be easily overlooked. This reduces diagnostic delays and delays in treatment and symptom relief. By analyzing medical images for disease features, Dr CADx’s deep learning algorithms can help doctors reduce the risk of missed diagnosis by up to 20%. In Africa, where 45 million patients lack access to a radiologist’s report on their medical image, Dr CADx’s solution has the potential to reduce misdiagnosis for at least 1.3 million patients annually, particularly in the case of chest X-rays, of which at least 50% are performed digitally.

The problem

Radiology plays a crucial role in medicine, but radiologists have limited accuracy with up to 30% error rates. The shortage of radiologists worldwide leads to delays in reporting, with up to 30 days for 10% of cases in regions like the EU. Only 60% of medical images are read by radiologists globally, and in Africa, where only 10% of images are read, an estimated 45 million patients lack access to a radiologist report. Non-radiologists have higher miss rates, up to 45%, resulting in severe patient outcomes, including 1.5 million deaths annually. Misdiagnoses also cost the healthcare sector over $115 billion per year.

The project outcomes

DR CADx has developed a model for 14 chest findings, which has not yet been validated in a clinical study. Omdena’s team helped to finetune the algorithm for the 14 chest findings by adding COVID, TB screening, and rib fractures to expand it to 17 findings.