Abstract

Universally, deep learning researchers and medical professionals have always had a gap that hinders their interaction: their expertise or use of technology. To bridge this gap, we have developed a medical image analysis platform that uses deep learning to make it easier for them to work together. We have built the third version platform where users can upload and process a medical image to check, research, aid in their diagnosis or training. Unfortunately, open-source codes are unavailable, but we allow users to download the free trial version on their local server. This article will discuss the three main contents in detail: how we have built the platform, how it supports users, and how we use models in our system. Primarily, we have integrated some popular models into this platform with a dataset of over 180,000 patients. In addition, the platform can help us detect 20 different diseases related to the lungs, such as Covid-19, Pneumonia, Atelectasis, Consolidation, Edema, Effusion, Lung Lesion, etc. The models' Accuracy is over 90 percent. The platform is freely available on the website: https://aceso.tech/ (**We deployed the platform to the domain during the conference. But now, we have taken down the model demos and will redeploy them once we get funding to operate. Of course, you can access our website's information at the above link.). Note: we have split our platform into 14 classes, 18 classes, and 20 classes.

Keywords: Radiology , Aceso Platform , Deep Learning , Python , Medical Image Analysis , Machine Learning