接上篇：临床脑电图操作的最低技术要求之设备

3. 记录

3.1 记录出的图像必须符合指南6“临床脑电图学标准图像指南”要求，这样可以使每个实验室出来的图像更加一致以利于彼此间的交流与比较。数字系统能够让我们修改出更合适的反应患者病情的图像，即使可以进行这样的弹性改动，我们还是应该记录患者的原始图像以作参考。系统的参考值，本身是无法修正的，所以数字记录参考电极应该是一个额外的电极（或几个电极的联合），而不是10:10系统或10:20系统中某个电极，这种电极往往使安置在Cz和Pz间的，如果将双耳连接点作为参考电极的效果往往是不理想的。

3.2 记录时应将患者的年龄、姓名、记录的数据、标识的数字、技术员的名字或名字的首字母标注在记录纸上。在记录的同时就应该标注好标识，否知容易导致医疗纠纷。每次记录都应该有一份基本的数据表。这份数据表包括记录的时间，最后一次痫间发作的时间及其他相关资料（如果有的话），患者的行为状态、患者所用药物的明细（这份明细也应该包括诱导睡眠而预先服用的药物）以及其他相关疾病的病史。

3.3 每次EEG测量开始及完成时均应对仪器进行适当的校准，如果可能的话，应该在记录开始处标出全套电极连接的所有导联。然而，对这种全数字系统进行生物学校准是很难完成的。必要时，我们在开始测量时对所有导联进行校准，以使它们能与校准信号更为吻合。每次校正过放大器，我们都应该对仪器再进行一次新的校正。每次EEG测量时这种校正都是不可或缺的，这种校正对于操作者而言就好比一个用以平衡EEG灵敏度的尺度，这个尺度可以衡量高低频的反应，噪音的水平，记录笔尖的校直程度和润湿程度。反之，从这种校正中，我们也可以看出操作者的水平和细心程度。校正电压必须适合灵敏度所需。除了标准的方波之外，生物定标会对检测图形选择过程的错误或记录仪器的错误有所帮助，所以我们需要用前后起源（额枕起源）的概念，因为它包括快的α范围的必行，同时也包括θ范围内眼球运动的图像，数字系统不能完全自动化地完成仪器的校准和生物校准，所以操作者必须从最初的系统参数的图像中观察判断前30s记录过程中所记录图形的情况。

3.4 在常规记录时，EEG仪器的灵敏性应设定在5~10μV/mm笔尖偏移距离。灵敏度被定义为输入电压后与笔尖偏移的距离的比率，单位为微伏每毫米（μV/mm）。常用的灵敏度为7μV/mm，它意味着50μV的校正信号会导致笔尖7.1mm的偏移。如果灵敏度下降（例如从7降至10μV/mm),EEG所做出的图像的振幅也会下降，相反地，若灵敏度上升（例如从7升至5μV/mm)，图像的振幅也会加大。如果灵敏度低于10μV/mm（例如20μV/mm），振幅会低得难以识别。若灵敏度高于5μV/mm（例如3μV/mm)则会导致波峰过度以至于超出记录纸的边界或造成电脑显示屏的信号轨迹重叠。5μV/mm的灵敏度意味着若要使笔尖偏移仅1mm，就需要输入5μV的输入电压（相应地，若要使笔尖偏移仅10mm内的话，就需要50μV的输入电压）。若灵敏度降至10μV/mm，则若想笔尖偏移控制在1mm内就需要更大的输入电压。例如10μV就比5μV要大（也就是说10μV/mm意味着将笔尖偏移控制在10mm的话需要100μV的输入电压而不是50μV）。所以灵敏度增加的话，它的数值就会变小；相反地，灵敏度降低的话，它的数字值就会变大。这里表面看上去似乎是一个矛盾的关系，但实际上灵敏度定义为输入电压每笔尖偏移单位是一个逻辑结论。但在数字系统中，这种直接的物理关系就没有了，因为计算机显示器的空间维度是变化的，所以显示时应标上清晰的尺度标志。

对脑电图仪放大器的操作可描述为增益，它定义为输出电压与输入电压的比率。例如，如果脑电图输入信号10μV被增益至1.0来趋动脑电图描记笔尖的移动，则它的增益就是1.0/0.00001=100 000，对于使用者而言，一个系统（模拟或数字）的增益，不会像灵敏度那样明显。在常规记录的校正期间，不能扭曲记录信号，但应使其扩大到足够大，以使得任何两个导联间的信号差异大于5%。不论记录前选择的哪个灵敏度（带有上述限制的），为了记录的准确性，不论脑电图形出现的是过高的振幅或过低的振幅，我们都应对灵敏度进行恰当的调整。

3.5 对于标准记录而言，低频滤波不应高于1Hz（3dB），对应时间常数至少需要0.16s。高频滤波不应低于70Hz（-3dB）。然后需要注意的是显示频率为70Hz时，计算机显示器在数据显示区域水平分辨率为至少1400，否则操作者会发现有高频信号的丢失和低频信号的失真。常规使用高于1Hz的低频滤波背景会减少记录的慢波伪迹，在δ范围内的病理活动容易丢失一些重要的信息，同样地，低于70Hz的高频背景滤波会扭曲或减少弱波的峰尖及其他的病理放电，使之不易辨认，或产生由于肌肉活动而出现的伪迹波形，而一份信息不全或不准确的记录往往会干扰医生的正确用药。值得强调的是，正确运用低频或高频滤波——需要在记录上标上正确的注释——能起到在记录上强调或清楚阐明图形模式的作用，所以对于滤波的选择我们必须谨慎。

3.6 60Hz滤波会扭曲或减少图形的峰尖，所以出现这种设置我们往往只用于其他方法失效的情况下。

3.7 走纸速度为3cm/s或数据显示速度为10s/一张纸是常用的记录指标。走纸速度为1.5cm/s或显示速度为20s/一张纸，有时也会用于新生儿或其他特殊情况下的EEG测定。

（未完待续）

英文原文

3. Recordings

3.1 Montages should be designed in conformity with Guideline 6: A Proposal for Standard Montages to Be Used in Clinical Electroencephalography. It is desirable that at least some montages in all laboratories be uniform to facilitate communication and comparison. Digital systems allow reformatting of montages to provide optimal display of activity at the time of interpretation. To permit this flexibility, initial recording must be made from a referential montage; but the system reference itself cannot easily be reformatted. For this reason the digital recording reference should be an additional electrode (or combination of electrodes), and not one of those in the 10:10 or 10:20 system. An additional electrode between Cz and Pz is commonly used. The use of linked ears as a digital recording reference is specifically discouraged.

3.2 The record should have written on it as a minimum the name and age of the patient, the date of the recording, an identification number, and the name or initials of the technologist.

Identifications should be made at the time of recording. Failure to do so may result in errors that have adverse medical and legal consequences. A Basic Data Sheet, attached to every record, should include the time of the recording, the time and date of the last seizure (if any), the behavioral state of the patient, a list of all medications that the patient has been taking, including premedication given to induce sleep during EEG, and any relevant additional medical history.

3.3 Appropriate calibrations should be made at the beginning and end of every EEG re-cording. If feasible, a recording with all channels connected to the same pair of electrodes should follow at the beginning. However, this biological calibration may not be possible with all digital systems. At the outset, all channels should be adjusted, if necessary, so that they respond equally and correctly to the calibration signal. When doubt as to correct functioning of any amplifier exists, a repeat calibration run should be made.

The calibration is an integral part of every EEG recording. It gives a scaling factor for the interpreter, and tests the EEG machine for sensitivity, high and low-frequency response, noise level, and pen alignment and damping. It also gives information about the competence and care of the technologist. Calibration voltages must be appropriate for the sensitivities used.

In addition to the standard square-wave calibration, the biologic calibration (“bio-cal”) may at times be of additional help in detecting errors in the montage selection process or in the pen-writing mechanism. For this purpose, an anteroposterior (fron to occipital) derivation should be used, since it can include fast and alpha range patterns as well as eye movement activity in the delta range. In digital systems that lack full provision for instrumental and biological calibration, the first 30 seconds of recording should be observed by the technologist from the primary system-reference montage.

3.4 The sensitivity of the EEG equipment for routine recording should be set in the range of 5—10 μV/ mm of pen deflection.

Sensitivity is defined as the ratio of input voltage to pen deflection. It is expressed in microvolts per millimeter (μV/mm). A commonly used sensitivity is 7 μV/mm, which, for a calibration signal of 50 μV, results in a deflection of 7.1 mm.

If the sensitivity is decreased (for example, from 7 to 10 μV/mm), the amplitude of the writeout of a given EEG on the paper also decreases. Conversely, if the sensitivity is increased (for example, from 7 to 5 μV/ mm), the amplitude of the writeout of a given EEG increases.

When the sensitivity is less than 10 μV/mm (for example, 20 μV/mm), significant low-amplitude activity may become indiscernible. If the sensitivity is greater than 5 μV/mm (for example, 3 μV/mm), normal EEG activity may overload the system, causing a squaring off of the peaks of the writeout onto the paper or overlapping of traces on the computer monitor.

Note that a sensitivity of 5 μV/mm means that, to obtain a pen deflection of 1 mm, a 5-uV input voltage is required (and correspondingly, to obtain a 10-mm deflection, an input of 50 μV is required). If the sensitivity is decreased to 10 μV/mm, the same 1-mm pen deflection now requires a larger input, i.e., 10 μV rather than 5 μV (and correspondingly, a 10-mm pen deflection now requires an input of 100 uV rather than 50 μV). Thus, as the sensitivity is increased, its numerical value becomes smaller. Conversely, as the sensitivity is decreased, its numerical value becomes larger. This perhaps seemingly paradoxical relationship is actually a logical consequence of the definition of sensitivity as input voltage per unit of pen deflection. With digital systems, this straightforward physical relationship is lost. Because the dimensions of computer monitors will vary, clear scale markers must be available as part of the display.

The operation of EEG amplifiers can also be expressed as gain, defined as the ratio of the output voltage to the input voltage. For example, if an EEG input signal of 10 μV is amplified to 1.0 V in order to move the mechanical pens of an electroencephalograph, then the gain is 1.0/.00001 = 100,000. The gain of an analog or digital system is not as obvious to the user as the sensitivity.

During calibration for routine recordings, the recorded signals should not be distorted but should be large enough to permit measurement to better than ±5% between any of the signals on the different channels.

No matter which sensitivity (within the above limits) is chosen prior to the recording, appropriate adjustments should be made whenever EEG activity encountered is of too high or low amplitude to be recorded properly.

3.5 For standard recordings, the low-frequency filter should be no higher than 1 Hz (—3 dB) corresponding to a time constant of at least 0.16 s. The high-frequency filter should be no lower than 70 Hz (-3 dB). Note, however, that to display frequencies as high as 70 Hz, a computer monitor would need a horizontal resolution of at least 1400 pixels in the data display area. Interpreters should be aware that some loss of high-frequency resolution will otherwise occur, along with the possibility of lower-frequency distortion due to spatial aliasing.

A low-frequency filter setting higher than 1 Hz should not be used routinely to attenuate slow-wave artifacts in the record. Vital information may be lost when pathologic activity in the delta range is present. Similarly, a setting lower than 70 Hz for the high-frequency filters can distort or attenuate spikes and other pathologic discharges into unrecognizable forms and can cause muscle artifact to resemble spikes. Production of a record with lost or inaccurate information is poor medical practice.

It must be emphasized, however, that judicious use of the low- or high-frequency filters—with appropriate annotation on the record—can emphasize or clarify certain types of patterns in the record. These filter controls, therefore, should be used selectively and carefully.

3.6 The 60-Hz (notch) filter can distort or attenuate spikes; it therefore should be used only when other measures against 60 Hz interference fail.

3.7 A paper speed of 3 cm/s, or digital display of 10 seconds/page, should be utilized for routine recordings. A paper speed of 1.5 cm/s, or 20 seconds/page, is sometimes used for EEG recordings in newborns or in other special situations.

翻译：陈玉娟

编辑：格格