A machine learning company is collecting images for a rare chest disease called cystic fibrosis where it affects only 0.01% of the USA population. The company wants to automatically detect if the person has this disease from his/her X-ray image. The main obstacle remains in the lack of images containing the disease, patient’s privacy where some patients refuse to cooperate. The company cannot obtain a massive dataset due to these restrictions. Given these restrictions, what should the company do to somehow enhance their model’s training?