{"id":1088394,"date":"2024-09-30T12:00:00","date_gmt":"2024-09-30T19:00:00","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?p=1088394"},"modified":"2024-11-05T06:42:27","modified_gmt":"2024-11-05T14:42:27","slug":"stress-testing-biomedical-vision-models-with-radedit-a-synthetic-data-approach-for-robust-model-deployment","status":"publish","type":"post","link":"https:\/\/www.microsoft.com\/en-us\/research\/blog\/stress-testing-biomedical-vision-models-with-radedit-a-synthetic-data-approach-for-robust-model-deployment\/","title":{"rendered":"Stress-testing biomedical vision models with RadEdit: A synthetic data approach for robust model deployment"},"content":{"rendered":"\n

This paper has been accepted at the 18th European Conference on Computer Vision (ECCV 2024) (opens in new tab)<\/span><\/a>, the premier gathering on computer vision and machine learning.<\/em><\/strong><\/p>\n\n\n\n

\"<\/figure>\n\n\n\n

Biomedical vision models are computational tools that analyze medical images, like X-rays, MRIs, and CT scans, and are used to predict medical conditions and outcomes. These models assist medical practitioners in disease diagnosis, treatment planning, disease monitoring, and risk assessment. However, datasets used to train these models can be small and not representative of real-world conditions, which often leads to these models performing worse in actual medical settings. To avoid misdiagnoses and other errors, these models must be rigorously tested and adjusted to perform reliably across different conditions.<\/p>\n\n\n\n

To mitigate the dataset challenge of not having enough diverse data and to improve the testing of biomedical vision models, we developed \u201cRadEdit: Stress-testing biomedical vision models via diffusion image editing<\/a>,\u201d presented at ECCV 2024. Aligned with the Microsoft Responsible AI principles<\/a> of reliability and safety, RadEdit helps researchers identify when and how models might fail before they are deployed in a medical setting. RadEdit uses generative image editing to simulate different dataset shifts (e.g., a shift in the patients’ demographics), helping researchers to identify weaknesses in the model. By employing text-to-image diffusion models trained on a wide array of chest X-ray datasets, RadEdit can generate synthetic yet realistic X-rays.<\/p>\n\n\n\n

RadEdit\u2019s approach involves using multiple image masks (binary images representing designated regions of a reference image), as illustrated in Figure 1, to limit changes to specific areas of the image, therefore preserving their integrity. It generates synthetic datasets free from spurious correlations and artifacts, addressing shortcomings in existing editing techniques. Traditional editing techniques often overlook biases within the generative model, leading to synthetic data that perpetuate these biases. Alternatively, these other editing techniques restrict edits to the point of unrealistic outputs.<\/p>\n\n\n\n\t

\n\t\t\n\n\t\t

\n\t\tvideo series<\/span>\n\t<\/p>\n\t\n\t

\n\t\t\t\t\t\t
\n\t\t\t\t\n\t\t\t\t\t\"On\n\t\t\t\t<\/a>\n\t\t\t<\/div>\n\t\t\t\n\t\t\t
\n\n\t\t\t\t\t\t\t\t\t

On Second Thought<\/h2>\n\t\t\t\t\n\t\t\t\t\t\t\t\t

A video series with Sinead Bovell built around the questions everyone\u2019s asking about AI. With expert voices from across Microsoft, we break down the tension and promise of this rapidly changing technology, exploring what\u2019s evolving and what\u2019s possible.<\/p>\n\t\t\t\t\n\t\t\t\t\t\t\t\t

\n\t\t\t\t\t
\n\t\t\t\t\t\t\n\t\t\t\t\t\t\tExplore the series\t\t\t\t\t\t<\/a>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t<\/div>\n\t<\/div>\n\t\n\n\n

How RadEdit works<\/h2>\n\n\n\n

RadEdit improves biomedical image editing using three key inputs, as illustrated in Figure 1:<\/p>\n\n\n\n