# AI基于错误公式计算CT扫描对比剂体积，超过肾功能不全患者的安全碘剂量

- **ID:** `medical/ct-contrast-renal-dose-miscalculation`
- **领域:** medical
- **类别:** data_error
- **错误码:** `CT-CONTRAST-VOL-ERR`
- **验证级别:** ai_generated
- **修复率:** 85%

## 根因

对比剂体积应根据患者体重和估算肾小球滤过率计算（例如，eGFR>60时1.5毫升/千克，最大125毫升），但AI使用固定体积（例如所有成人100毫升）或体表面积公式，导致高风险患者可能发生对比剂肾病。

## 版本兼容性

| 版本 | 状态 | 引入 | 弃用 |
|------|------|------|------|
| Siemens Somatom Definition Flash v7.0 | active | — | — |
| GE Revolution CT v3.2 | active | — | — |
| Canon Aquilion ONE v6.5 | active | — | — |

## 解决方案

1. ```
   Use the ACR contrast manual formula: for eGFR > 60, give 1.5 mL/kg up to 125 mL; for eGFR 30-60, give 1.0 mL/kg up to 100 mL; for eGFR < 30, consult radiologist for alternative imaging or pre-hydration protocol
   ```
2. ```
   Implement a clinical decision support tool (e.g., Nuance mPower) that checks eGFR, weight, and contrast type before approving the scan order
   ```
3. ```
   Use iso-osmolar contrast (iodixanol) for all patients with eGFR < 45, which reduces CIN risk, and calculate volume based on 1.0 mL/kg max 100 mL
   ```

## 无效尝试

- **** — This is safe only for normal renal function; for eGFR < 30, even 1.0 mL/kg can exceed safe iodine load (e.g., 350 mgI/kg) (50% 失败率)
- **** — Age alone is not a reliable predictor; some elderly have normal eGFR and need full dose for diagnostic quality; under-dosing causes nondiagnostic scans (65% 失败率)
- **** — AI may use outdated formulas (e.g., Cockcroft-Gault instead of CKD-EPI) or ignore acute kidney injury, giving falsely normal eGFR (55% 失败率)