Artificial intelligence (AI) in detection of lung malignoma

<bold>Purpose:</bold> The incorporation of artificial intelligence (AI) in radiological practice in detection of lung malignoma <bold>Method:</bold> Machine intelligence (AI) is potentially the single most disruptive impact on radiology in a few decades, and possibly since the beginnings of our specialty. We used a database of 250000 views to create a model for automatic tumor detection. After creating the neural network, we created the image smoothing software. The next step was to load the neat findings, without pathology. In the next step, we inserted single lesions, and later switched to multiple lesions. Two radiologists with decades of experience controlled the findings. <bold>Results:</bold> In the first phase, the main goal was the detection of normal findings. Accuracy was 48%. After that, we changed the contrast of the output image, which led to an increase in accuracy to 76%. In the case of single lesions in the first stage, the results were compromised by the quality of the output image and the accuracy was poor. By changing the image resolution, the accuracy was 53%. Currently, the accuracy is around 69%, and we expect it to be 81% in a year. With multiple lesions, detection is high, but specificity is low. In further work we expect to achieve a specificity of 90% <bold>Conclusions:</bold> The introduction of AI into healthcare could alter the dynamic between physicians and patients, with potential implications for patient trust. Artificial intelligence provides the chance of alter that goes far over former disruptions.