一毫米的线,还是一串点?——EIT 想量的那个「宽度」,先得问清楚是谁的宽度

Written by

in

Language

Evidence level: B

用 EIT 量经络有多宽、是连着还是断着,看着是个测量问题;实际上,你量出来的那个毫米数,可能大部分是仪器的宽度,不是经络的宽度。

正文

先说一个不好听的算术

源设计给的待测目标很明确:经典经络宽度按 1–10 mm 算,量出来多宽、连不连续,就能裁决”连续线”还是”离散点串”。

问题出在分母上。EIT 的空间分辨率约为图像直径的 5%–10%(PMID 27203477)。成人前臂周长 20–28 cm,折成直径 6.4–8.9 cm,于是 EIT 的模糊核(点扩散函数,PSF)宽度约 3–9 mm。而你要找的东西是 1–10 mm。

两个区间几乎完全重叠。PSF 比目标还宽,或者勉强一样宽。 用高斯近似估:真实宽度 1 mm、PSF 宽 5 mm,读出约 √(1²+5²) ≈ 5.1 mm——误差四倍,方向固定,永远偏宽。

更要紧的是不可逆。读到 5 mm,真值可以是 5 mm,可以是 1 mm,也可以是 0:一个根本不存在的东西被 PSF 抹一下,也能长出一个 5 mm 的团。同一个读数对应一整族真值,即不可辨识。影像学的经验判据是物体直径须大于 PSF 全宽半高(PMID 22225287)。1 mm 对 3–9 mm,比值 0.1–0.3,远远不够。

所以第一个结论是硬的:在这套几何下,”宽度不匹配”不能当”经络不是 1 mm”的证据。 读数被 PSF 地板托住,无论经络多细,你都读不到 1 mm。

第二个坑:这个 PSF 还会自己变

“多横断面宽度一致性”是好指标——前提是同一根管子各处读出来一样宽。但 EIT 重建是空间变异的:同一个异常放在图里不同位置,重建签名不同、量值也不同(PMID 19202236,提出 PSF 均衡化正为治此)。环形阵列里越靠电极越清、越靠中心越糊。

于是”一致性”检验会尴尬:它测到的变异,可能主要是 PSF 沿周向漂移,不是低阻带自身粗细的变化。 得先在体模里逐位置标定恢复系数,把仪器那部分变异减掉;不做,就在测仪器。

第三个坑:连续性由算法说了算

“连续 vs 点串”听着是形态判决,其实是正则化选择题。以全变差(TV)为例:L1 惩罚保边界,却偏向”分片常数”,倾向于长出边缘。同一批边界电压,换算法形态可以完全不同——系统比较显示 PDIPM 边界最锐但对比度偏低、SB 误差最低、LADMM 最快但分辨率最差(PMID 26008768)。TV 还会在背景生伪影。

翻译过来:用偏好锐边的算法,点串被连成块;用偏平滑的 L2,断点被糊掉,看着连续。 “看它连不连续”这个动作,在选定算法那刻已被预答一半。故必须报”多算法一致性”,三种不一致就判”本分辨率下不可判”。

还有个采样问题。1992 年 68 名受试者的计算机扫描研究结论是低阻点绝大多数在十四经线上或旁开 5 mm 内,但明确写道:低阻分布不是不间断实线,而是一系列点沿经排列(PMID 1291821)。这是”点串”说法的原始出处之一。可它是点测量的产物——以厘米级间距去扫,任何小于间距的连续段都会读成点串。 断点能否看见,取决于采样步长,不取决于组织。

第四个坑:那个”1 mm”,是谁的数字

“经络宽约 1.0 mm”在文献里的出处是定位仪的标称能力——宣称能定位十二经、宽 1.0 mm、与古图相符(PMID 34319572 转述)。这是设备规格,不是独立测量。 这一先验本身是软的。

更要紧的错位:经络低阻的证据大多来自皮肤表面点测量,而环形 EIT 测的是深部断面电导率。皮肤电测量中电流密度最高处在电极附近,局部皮肤电阻率主导读数,深层变化对总阻抗影响很小(PMID 17983337);电极间距决定穿透深度,两电极系统约为间距一半,拉大则更深但能量密度下降(PMID 37177455)。环形阵列的间距被周长除以电极数定死——“能看多深”和”能看多宽”在这套几何里绑在一起,不能既要深又要细。

若经络低阻本质是皮肤层(角质层、汗腺、NO–去甲肾上腺素通路)现象,搬到深部断面上找,可能一开始就找错了层(机制文献 PMID 34319572)。

所以先做三件事,再谈裁决

一、把宽度定义写死:半高全宽、-6 dB 切片,还是阈值等值线?在 PSF 主导时,半高全宽读的其实是 PSF 的半高全宽。

二、先测比值,再测宽度:体模逐位置标定恢复系数,算出该深度的 PSF 宽度;真实宽度大于 PSF 才有意义,否则只报”上限”。

三、连续性多算法报:TV / L2 / SB 各出一套,一致才采信。

「三百六十五」这个数字

《灵枢·九针十二原》:「节之交,三百六十五会……所言节者,神气之所游行出入也,非皮肉筋骨也。」

三件事值得注意。第一,它数的是点;而《灵枢·经脉》通篇「循」「过」「挟」「贯」「属」「络」,画的是线。点与线两套语言在《内经》里本就并存。第二,「三百六十五」不是数出来的,是”以应一岁三百六十五日”按历法定的。第三,它自己声明「非皮肉筋骨」:这个”节”不在解剖层,就不该拿解剖的尺子去核。

古人配数字用的是「应」,不是「量」。一毫米是仪器标称,三百六十五是天数——都属”为让体系自洽而选定的数”。今天拿游标卡尺去核它,不是数字错了,是问句错位了:能被毫米裁决的是组织;不能被毫米裁决的是那个「非皮肉筋骨」的层面。 分不清这两层,测出来的每个毫米都很精确,也都很无关。

证据与局限

已验证: EIT 空间分辨率约为图像直径的 5%–10%,受电极数与独立测量数限制(PMID 27203477);重建问题空间变异,PSF 需均衡化,同一目标在不同位置重建量值不同(PMID 19202236);部分容积效应下,物体直径需大于 PSF 全宽半高才能有效恢复定量值(PMID 22225287);TV 与 L2、SB 等正则化对边界锐度、对比度与伪影的影响有直接比较(PMID 26008768);皮肤电测量的电流密度集中于电极附近、电极间距决定穿透深度(PMID 17983337、PMID 37177455)。

部分验证: 低阻点沿十四经分布、多数在线上或旁开 5 mm 内,但呈”系列点”而非连续实线(PMID 1291821,1992 年点测量研究,采样间距决定断点可见性);经络电学特性系统综述判定证据不足以确证(源文已引 PMID 18240287);”经络宽约 1.0 mm”在文献中的出处为定位仪标称(PMID 34319572 转述),非独立测量。

待验证: ① 四肢环形阵列在目标深度的实测 PSF 宽度,与 1–10 mm 先验的比值——尚无研究给出该比值;② 三种正则化下连续性统计的一致性程度——无直接数据;③ 经络低阻信号究竟位于皮肤层还是深筋膜层——决定环形 EIT 是否是对的探测器,属实证空白。三项均为本路线的前置门槛,未过门槛之前,”宽度匹配与否”不能作为任何一方证据。

相关阅读

证据等级: B(分辨率、部分容积效应、正则化行为、电极穿透深度均有 PubMed 直接支撑,但”EIT 低阻带宽度与经络宽度匹配”本身无直接研究,属方法学前置门槛的论证)
参考文献: PMID 27203477, PMID 19202236, PMID 22225287, PMID 26008768, PMID 1291821, PMID 17983337, PMID 37177455, PMID 34319572, PMID 18240287


免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

One Millimeter of Wire or a String of Dots? — Before Asking About EIT’s \”Width,\” We Must Clarify Whose Width It Is

Evidence level: B

Using EIT to measure the width of meridians, determine continuity or discontinuity appears to be a measurement issue; however, the millimeter value obtained is likely dominated by the instrument’s resolution rather than the actual width of the meridian.

Main Text

An Unpleasant Arithmetic to Start With

The target defined by the source design is clear: classic meridian widths are estimated at 1–10 mm. Whether the measurement yields a continuous line or a discrete string of dots will determine the classification.

The problem lies in the denominator. The spatial resolution of EIT is approximately 5%–10% of the image diameter (PMID 27203477). With an adult forearm circumference of 20–28 cm (equivalent to a diameter of 6.4–8.9 cm), the Point Spread Function (PSF) width of EIT is approximately 3–9 mm. The target feature to be detected, however, is 1–10 mm.

These two intervals almost completely overlap. The PSF is wider than, or comparable to, the target. Using a Gaussian approximation: if the true width is 1 mm and the PSF is 5 mm wide, the reading is approximately $\sqrt{1^2 + 5^2} \approx 5.1$ mm—an error of fourfold in a fixed direction, always biased wide.

More critically, this is irreversible. A reading of 5 mm could correspond to a true value of 5 mm, 1 mm, or even 0: an entity that fundamentally does not exist can be “smudged” by the PSF to appear as a 5 mm cluster. A single reading corresponds to an entire family of true values, rendering them indistinguishable. The empirical criterion in imaging is that the object diameter must exceed the PSF full width at half maximum (FWHM) (PMID 22225287). For a 1 mm object against a 3–9 mm PSF, the ratio is 0.1–0.3, which is far insufficient.

Therefore, the first conclusion is firm: Under this geometric framework, “width mismatch” cannot serve as evidence that “the meridian is not 1 mm.” The reading is supported by the PSF floor; regardless of how fine the meridian is, you cannot read 1 mm.

The Second Pitfall: This PSF Varies Itself

“Consistency across cross-sections” is a good indicator—provided the width reads the same at all locations along a single tube. However, EIT reconstruction is spatially variant: placing the same anomaly at different positions in the image yields different reconstruction signatures and values (PMID 19202236; PSF normalization is proposed to address this). In a ring array, clarity increases near the electrodes and decreases toward the center.

Consequently, an “consistency” check becomes awkward: The variation it detects may primarily reflect PSF drift along the circumference, rather than actual changes in the thickness of the low-resistance band. One must first calibrate recovery coefficients at each position using a phantom to subtract the instrument-induced variation; otherwise, one is merely measuring the instrument.

The Third Pitfall: Continuity Is Decided by the Algorithm

“Continuous vs. discrete string” appears to be a morphological judgment but is actually a choice regarding regularization. Taking Total Variation (TV) as an example: L1 penalization preserves boundaries but favors “piecewise constant” solutions, tending to grow edges. With the same boundary voltages, different algorithmic morphologies can yield completely different results—systematic comparisons show PDIPM has the sharpest boundaries but lower contrast, SB yields the lowest error, and LADMM is fastest but has the poorest resolution (PMID 26008768). TV also generates artifacts in the background.

Translated: Using an algorithm that favors sharp edges connects discrete points into blocks; using a smoother L2 blurs discontinuities, making them appear continuous. The act of “observing whether it is continuous” is already partially predetermined at the moment an algorithm is selected. Therefore, one must report “multi-algorithm consistency”; if three algorithms disagree, the result should be judged as “unjudgable at this resolution.”

There is also a sampling issue. A 1992 computer scan study of 68 subjects concluded that low-resistance points are overwhelmingly located on or within 5 mm of the Fourteen Meridian lines, but explicitly stated that the low-resistance distribution is not an uninterrupted solid line, but a series of points arranged along the meridian (PMID 1291821). This is one of the original sources for the “discrete string” terminology. However, it is a product of point measurement—scanning with centimeter-level spacing causes any continuous segment smaller than the interval to be read as a discrete string. Whether discontinuities can be seen depends on the sampling step size, not the tissue properties.

The Fourth Pitfall: Whose Number Is That “1 mm”?

The claim that “meridian width is approximately 1.0 mm” originates from the nominal capability of a localization device—stating it can locate the twelve meridians with a width of 1.0 mm, consistent with ancient diagrams (PMID 34319572, paraphrased). This is equipment specification, not an independent measurement. This prior assumption itself is soft.

A more critical mismatch exists: evidence of low resistance in meridians mostly comes from surface skin point measurements, whereas ring EIT measures deep cross-sectional conductivity. In skin electrical measurement, current density is highest near the electrodes, where local skin resistivity dominates the reading; deep changes have little impact on total impedance (PMID 17983337). Electrode spacing determines penetration depth; in a two-electrode system, this is approximately half the spacing—increasing it goes deeper but reduces energy density (PMID 37177455). In a ring array, spacing is fixed by dividing the circumference by the number of electrodes—“how deep one can see” and “how wide one can see” are bound together in this geometry; one cannot have both depth and fine resolution.

If the low-resistance nature of meridians is essentially a phenomenon of the skin layer (stratum corneum, sweat glands, NO–norepinephrine pathways), searching for it on a deep cross-section may start with the wrong layer (mechanism literature PMID 34319572).

Therefore, Three Things Must Be Done Before Any Verdict

  1. Fix the definition of width: Full Width at Half Maximum (FWHM), -6 dB slice, or threshold isochrone? When dominated by the PSF, the FWHM read is actually the PSF’s FWHM.
  2. Measure the ratio before measuring width: Calibrate recovery coefficients at each position using a phantom to calculate the PSF width at that depth; meaningful true width only exists if it exceeds the PSF, otherwise report only an “upper limit.”
  3. Report continuity across multiple algorithms: TV / L2 / SB should each produce a set; only if consistent is the result accepted.

The Number “365”

Neijing · Nine Needles and Twelve Origins: “At the junctions of the joints, there are 365 meetings… The ‘joints’ referred to are where the divine qi travels and enters and exits; they are not skin, flesh, bone, or tendon.”

Three points deserve attention. First, it counts points; whereas Neijing · Meridians uses terms like “circles,” “passes,” “clutches,” “pierces,” “belongs,” and “connects,” drawing lines. The languages of points and lines coexist within the Neijing. Second, “365” is not counted but determined by the calendar to correspond to the 365 days of a year. Third, it explicitly states “not skin, flesh, bone, or tendon”: This “joint” does not exist at the anatomical layer; therefore, it should not be measured with an anatomical ruler.

Ancients paired numbers using “correspondence” (应), not measurement. One millimeter is an instrument nominal value, while 365 is a day count—both belong to “numbers selected to ensure system consistency.” Using a vernier caliper today to verify them is not a mistake in the number, but a mismatch in the question: What can be adjudicated by millimeters is tissue; what cannot be adjudicated by millimeters is the level of the “non-anatomical” layer. Without distinguishing these two layers, every millimeter measured is precise yet irrelevant.

Evidence and Limitations

Verified: EIT spatial resolution is approximately 5%–10% of the image diameter, limited by electrode count and independent measurements (PMID 27203477); reconstruction exhibits spatial variability, requiring PSF normalization as values for the same target differ at different positions (PMID 19202236); under partial volume effects, object diameter must exceed PSF FWHM for effective quantitative recovery (PMID 22225287); direct comparisons exist regarding the impact of TV and L2/SB regularization on boundary sharpness, contrast, and artifacts (PMID 26008768); skin electrical measurement shows current density concentrated near electrodes, with electrode spacing determining penetration depth (PMID 17983337, PMID 37177455).

Partially Verified: Low-resistance points distribute along the Fourteen Meridians, mostly on or within 5 mm of the lines, but appear as a “series of points” rather than a continuous solid line (PMID 1291821, 1992 point measurement study; visibility of discontinuities depends on sampling interval); a systematic review of meridian electrical properties concludes evidence is insufficient for confirmation (source cites PMID 18240287); the “meridian width approx. 1.0 mm” claim originates from device specification (PMID 34319572, paraphrased), not independent measurement.

To Be Verified: ① The measured PSF width in four-limb ring arrays at target depth and its ratio to the 1–10 mm prior—no study provides this ratio; ② The degree of statistical consistency in continuity under three regularization methods—no direct data available; ③ Whether low-resistance meridian signals reside in the skin layer or deep fascia layer—determining whether ring EIT is the correct detector, which remains an empirical gap. These three are prerequisite thresholds for this approach; before crossing them, “width matching or mismatching” cannot serve as evidence for either side.

Related Reading

Evidence Level: B (Resolution, partial volume effects, regularization behavior, and electrode penetration depth are directly supported by PubMed; however, “matching between EIT low-conductance band width and meridian width” itself lacks direct research, belonging to methodological prerequisite justification)
References: PMID 27203477, PMID 19202236, PMID 22225287, PMID 26008768, PMID 1291821, PMID 17983337, PMID 37177455, PMID 34319572, PMID 18240287


Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *