FAQ
Brainchop is a web-based tool that allows scientists and clinicians to perform volumetric analysis of structural Magnetic Resonance Imaging (MRI) using pre-trained deep learning models, without the need for technical expertise or setup of AI solutions.
To preserving end-user MRI data privacy, unlike traditional web applications, brainchop operates within the browser environment by performing client-side processing to process the end-users data locally on their machines without the need for server-side processing and the associated privacy and portability issues.
There are several reasons why an inference model might not produce an output or produce unexpected output. These reasons include, but are not limited to:
1- The browser: Sometimes, changing the browser can resolve this issue.
2- The sample: If the sample is a clinical scan, it may not show output with small models such as GWM light models. Try using more robust models.
3- The model: If the issue is caused by a fast version of the model (e.g. Aparc+Aseg 104 (High Mem, Fast)), try using the low memory and slow version of the model (e.g. Aparc+Aseg 104 (Low Mem, Slow)).
Finally if you need help with any issue please feel free to report the issue or open a discussion.