Dose error checking device and method of checking a dose error

A dose error checking device comprising:

• a processing circuit configured to generate key data based on a final feature map and a respective weight for each class of dose error causes,
• wherein the final feature map is based on a convolution and pooling process for a convolutional neural network (CNN) input image,
• wherein the key data include class activation mapping (CAM) image and a row corresponding to an error risk sensor from the CAM image,
• wherein the processing circuit is configured to
• display a sensor activation mapping (SAM) image that corresponds to the error risk sensor, the SAM image being based on (i) selection data that select at least one row of the key data and (ii) the CAM image;
• analyze a cause of a dose error based on the SAM image;
• and
• provide an action to resolve the cause of the dose error.

A method of checking a dose error comprising:

• generating, using a processing circuit, a convolutional neural network (CNN) input image by arranging, based on a sensing group, raw data generated by multiple sensors for measuring dose error shots;
• generating, using the processing circuit, a class activation mapping (CAM) image based on a final feature map and a respective weight for each error cause class, the final feature map corresponding to the CNN input image;
• generating, using the processing circuit, a key data including at least one row of the CAM image, the at least one row corresponding to at least one error risk sensor;
• and
• displaying, using the processing circuit, a sensor activation mapping (SAM) image that corresponds to an error risk sensor of the at least one error risk sensor, the SAM image being based on the key data, selection data that select at least one row from the key data, and the CAM image.