SelfCue：让3D CT报告生成器说出它已知的内容
SelfCue: Making a 3D CT Report Generator Say What It Already Knows
AI总结：
SelfCue通过对比解码，将3D CT报告生成器隐藏状态中的信息显式化，提升报告临床效能，并蒸馏为SelfCue-KD以零额外推理成本集成。
AI中文摘要：
3D CT报告生成的进展通常依赖于日益复杂的架构和更大规模的训练数据。然而，我们发现3D CT报告生成器其实已经包含了其报告所遗漏的信息，但当隐藏状态转化为词元时，这些信息便丢失了。在18个CT-RATE异常中，这种隐藏到报告的显现差距反映为宏AUROC从隐藏状态的0.848下降到生成报告的0.739。我们提出了基于对比解码的SelfCue方法。它促进隐藏状态已支持的内容，并抑制其不支持的内容。它将临床效能F1提升至0.481，LLM评判的GREEN评分提升至0.510。将该行为蒸馏到权重中，得到SelfCue-KD，一个学生模型，保留了大部分增益，推理时无需额外开销，并可无缝集成到任何已服务于基线的流程中。代码可在该https URL获取。
英文摘要：
Progress in 3D CT report generation is usually sought in increasingly sophisticated architectures and larger pools of training data. We find instead that a 3D CT report generator already holds what its report leaves out, and loses it when the hidden state becomes tokens. Over the 18 CT-RATE abnormalities, this hidden-to-report surfacing gap is reflected by a drop in macro AUROC from 0.848 in the hidden states to 0.739 in the generated report. We propose SelfCue based on contrastive decoding. It promotes what the hidden state already supports and suppresses what it does not. It raises clinical efficacy F1 to 0.481 and the LLM-judged GREEN score to 0.510. Distilling that behaviour into the weights gives SelfCue-KD, a student that keeps most of the gain, needs nothing extra at inference, and drops into any pipeline already serving the baseline. Code is available at https://github.com/renjie-liang/SelfCue-CT.
