MDF-Net for abnormality detection by fusing X-rays with clinical data - crossword

12345678910
Across
  1. 5. Adjective used to describe deep learning that enhances diagnostic accuracy by using multiple data sources (like images and text) rather than vision alone.
  2. 6. The category of data that includes patient history, vital signs, and lab results, which acts as "critical context" for the diagnosis.
  3. 9. The "trap" encountered when using larger models like ResNet-50, where the model memorized training data instead of learning generalizable patterns.
  4. 10. A specific type of sign (like Heart Rate or Temperature) included in the structured clinical data used for the study.
Down
  1. 1. A term describing the nature of image data (high-dimensional, dense pixel arrays) as opposed to the linear, 1D nature of clinical data.
  2. 2. The medical professional whose workflow serves as the model for the study; they do not diagnose in a vacuum but combine visual evidence with context.
  3. 3. The specific "backbone" architecture that was selected as the "winner" because it offered the optimal balance of computational efficiency and accuracy.
  4. 4. The core strategy of the MDF-Net architecture, occurring at both "Early" and "Late" stages to merge different data dimensions.
  5. 7. A critical lung condition mentioned in the case study that appears as "hazy, diffuse opacities" and can look very similar to Pneumonia on an X-ray.
  6. 8. The approach that combines visual evidence with clinical context, which the presentation states moves AI closer to the reasoning of a human doctor.